AVENUE Site Demonstrators: Geneva, Lyon, Luxembourg, and Copenhagen

Abstract

The AVENUE project, which was carried out in Geneva, Lyon, Luxembourg, and Copenhagen, demonstrates the innovative integration of automated vehicles (AVs) into urban public transport systems. This initiative investigated the use of AVs in different urban environments and achieved remarkable progress in the areas of mobility, environmental sustainability, and accessibility. Through real-world testing, the project identified key success factors such as technological integration, social acceptance, and infrastructural adaptation. Achievements include on-demand services, seamless integration into existing transport networks and the promotion of multimodal transport solutions. The experiences from the AVENUE project provide crucial insights into operational and societal aspects of AV implementation in public transport and emphasise the importance of ongoing innovation and research in the field of automated mobility for future urban transport ecosystems.

1 Introduction

As urban populations continue to surge and our cities grapple with ever-increasing traffic congestion and environmental concerns, the quest for innovative and sustainable public transportation solutions has intensified. Among the emerging technologies at the forefront of this transformation is fully automated vehicles. These self-driving vehicles have the potential to revolutionise public transportation, offering eco-friendly, efficient, and convenient mobility options for commuters in urban, suburban, and even rural areas.

Today, automated vehicles are proposed and employed across various use cases, leading to some confusion and unrealistic expectations. To provide clarity on the objectives of the AVENUE project, we categorised them into the following applications of automated¹ vehicles in today’s context:

• 1. Public Transport Fixed Line or Predefined Geofenced Zone Local Transport:

  • Operates within predetermined geofenced areas with established stop points.

  • May or may not require physical infrastructure.

  • On-demand transport is possible, with ride pooling options typically mandatory.

• 2. Shared Autonomy for Long and Short Distances (VDV, 2015):

  • Involves shared or private transportation using non-privately owned vehicles.

  • These vehicles are capable of traveling and stopping at various locations.

• 3. Owned Autonomy (ibidem):

  • Refers to private transportation using privately owned vehicles.

  • Offers 100% control over mobility requirements for the vehicle owner.

This categorisation helps define the different levels of automated mobility services, each catering to specific needs and preferences in urban transportation. Public transport, which is the target of the AVENUE project (Konstantas, 2021), targets the first of the above categories. Public transport serves a distinct role that distinguishes it from a taxi service as it is intended to be the most cost-effective mode of non-human-powered travel, remaining cost-effective and accessible, while other options provide varying degrees of flexibility and control to users but at higher costs.

The deployment of automated shuttles for public transportation is a promising development, but it is crucial to emphasise that their success hinges on testing and evaluation under real-world conditions, as the road to a seamless automated future is paved with uncertainties, technical challenges, and regulatory hurdles (Antonialli et al., 2022). Therefore, it is imperative to conduct real tests in actual deployment scenarios to identify potential issues, correct mistakes, and obtain invaluable feedback from users.

In this chapter, we present the deployments in the different sites of the AVENUE project where fully automated shuttles were (and in some still are) being deployed for shared public transportation services. These case studies serve as compelling examples of how automated shuttle technology is evolving and adapting to the unique challenges and opportunities presented in different urban landscapes. Through these deployments, we were able to understand and concretise the hidden issues, for which we had just some vague idea, and refine this models, technology, and services developed in order to ensure integration into our daily lives and related urban environment.

The significance of this research lies not only in its contribution to the development of automated shuttle technology but also in its potential to reshape the way we envision public transportation in the twenty-first century. By highlighting the challenges faced, lessons learned, and the ultimate goal of enhancing the daily commute for countless individuals, this chapter aims to provide a comprehensive perspective on the deployment of automated shuttles for public transportation.

In the pages that follow, we will delve into the specific case studies, examining the unique contexts and challenges of each deployment site (an overview is provided in the following table). Through these analyses, we hope to shed light on the multifaceted nature of automated shuttle integration and inspire continued research, innovation, and collaboration in the field of automated public transportation (Table 2.1).

Table 2.1 Summary of AVENUE operating site (+ODD components)

2 The Changing Landscape of Mobility

There is a noticeable trend of people leaving rural areas and migrating towards more densely populated cities and suburbs. This mass movement results in higher population concentrations per square mile (United Nations, 2018). The consequence of this increased population density is an upsurge in commuter numbers and a growing demand for urban and suburban transportation services, leading to an increased use of private vehicles for commuting.

In response to the ever-increasing number of private vehicles within cities, a current prevailing trend in Europe is the restriction of traditional motorised individual traffic in city centres. Instead, there is a focus on providing shared public transportation options with the aim of reducing environmental impacts and enhancing the overall quality of life for both citizens and commuters (Nieuwenhuijsen & Khreis, 2016).

The future of transportation hinges on collaborative efforts to create sustainable mobility options, with on-demand automated public transport at the forefront. This represents a transformative shift in public transportation and has the potential to be a catalyst for social change.

In the following sections, we will explore the AVENUE deployments in Europe. Before delving into these specific deployments, we will elucidate the driving factors behind this trend and explore potential solutions that have paved the way for the adoption of automated vehicles in public transportation.

2.1 Fighting Congestion

The primary goal of contemporary mobility regulation is to mitigate the environmental impact, specifically stemming from CO₂ emissions resulting from various modes of transportation (European Commission, 2020). A fundamental step towards addressing climate protection is the transition from internal combustion engine vehicles to electric ones. However, this shift alone may not be adequate, as it fails to alleviate the overall strain on road infrastructure. Mere replacement of traditional individual vehicles with electric counterparts still results in the same road occupancy and congestion issues.

An alternative approach to combat traffic congestion could involve restricting the entry of single-occupant or dual-occupant vehicles into inner cities. Instead, priority could be given to purpose-designed vehicles capable of efficiently transporting a significant number of passengers or goods while optimising road space utilisation. This approach might encompass larger vehicles for mass transportation along prominent routes, alongside smaller vehicles ready to promptly respond to on-demand requests.

By adopting this strategy, non-essential traffic could be reduced, and curbside parking areas could be repurposed for intelligent mobility solutions, bicycle lanes, and other innovative transportation methods. This transformation would inevitably reshape urban infrastructure and create fresh opportunities for public transportation agencies, provided they can offer more customer-centric transport services (OECD/ITF, 2015, 2016, 2017, 2018).

2.2 The Transformation of Public Transportation

In response to the growing need to transport larger numbers of people efficiently, especially in urban centres and mobility hubs, the conventional approach has often involved separate ground-level railway and bus lines. These systems serve the majority of commuters effectively but fall short of providing the door-to-door convenience associated with individual car travel.

Public transportation journeys often consist of a collective core, with personalised segments at the start and end to accommodate the individual needs of passengers. To address the growing demand for smooth door-to-door transportation solutions within the public transport system, it is imperative to adopt a multifaceted approach. This approach entails the integration of various transportation modes into a unified and interconnected system, commonly referred to as mobility as a service (MaaS). The central challenge for public transport operators is to create a multimodal transportation solution that not only matches the convenience of private car usage but actually surpasses it in terms of quality.

The transformation of city centres, coupled with the phasing out of private cars, necessitates the introduction of green and health-conscious mobility alternatives or robust public transport systems. While walking and cycling are gaining traction, there will always be a segment of the population reliant on vehicle transit. On-demand public transport with smaller vehicles could address this need within car-free city areas. To integrate the related external costs (competition with public transport in particular mass transit, additional congestion costs, and/or other costs) for the city and the citizen, a different price should be claimed to the users.

On-demand public transport represents enhanced efficiency and sustainability. Passengers can schedule their trips at their convenience within designated service hours, arranging for minibus pickups and drop-offs at agreed-upon locations, all facilitated through online booking or dedicated applications. This flexibility streamlines the public transport experience, making it more convenient for daily commuters (Alonso-González et al., 2018).

Another crucial aspect of this evolution is the development of high-capacity, demand-responsive vehicles designed to maximise passenger capacity and minimise overall travel distance. The primary challenge lies in coordinating passenger collection and drop-off for on-demand travel, which may require either a few larger vehicles or numerous smaller ones, with the latter incurring higher costs due to the need for dedicated drivers for each.

Currently, public transport operators tend to utilise vehicles that are specialised for either bulk transportation or individual travel. However, it can be highly inefficient and impractical to transfer a large group of, for example, 200 people using minibuses designed for just 12 passengers each, as this approach essentially forces the implementation of a monomodal transport solution.

Nevertheless, there are innovative options available. For example, large trolleybuses, specifically designed for fixed main routes equipped with overhead electricity lines, can offer efficient transportation solutions. These vehicles can also serve the first and last mile on an on-demand basis by using an additional battery pack when traveling off the main electrified routes. However, it is important to note that large trolleybuses may come with a higher cost and could present challenges when navigating through smaller neighbourhoods due to their size and overhead line constraints.

A potential solution to overcome these challenges is the utilisation of shared driverless vehicles for public transportation. Automated and connected vehicles have already undergone successful testing, with the primary hurdle being their interaction with older vehicles, bicycles, pedestrians, and existing infrastructure.

2.3 Readiness to Adopt New Transportation Means

End users seek a seamless, user-friendly travel experience. They desire an all-encompassing travel assistant that plans their journey, anticipates their travel habits, and provides real-time information on vehicle arrivals, appearance, pickup locations, drop-off points, delays, and essential itinerary changes, all while ensuring an efficient flow of pertinent information.

In the modern transportation landscape, passengers are increasingly open to adopting innovative transportation modes that make use of cutting-edge technologies and alternative means of mobility. A prime example of this is the surging popularity of electric bicycles. These battery-powered models have become a captivating category within the evolving transportation ecosystem, attracting a growing number of enthusiasts. Electric bicycle riders often sidestep the complexities of multimodal transportation chains, as they can journey directly from their starting point, “A”, to their destination, “B”, without relying on other modes of transport. The introduction of electric bicycles has not only extended the range of cyclists but has also enhanced their capabilities, allowing them to cover even greater distances by combining their pedal power with electric assistance. This trend underscores passengers’ readiness to embrace innovative, eco-friendly transportation solutions. Holiday travel stands as a well-established example of the widespread shift towards multimodal transportation. It vividly illustrates the transformation of our travel habits over time. Historically, families used to embark on monomodal journeys, primarily traveling by car directly from their residences to their holiday destinations. However, the landscape of holiday travel has evolved significantly, and contemporary vacations now often involve a more intricate and multimodal approach.

Today, families embark on individualised journeys, departing from their homes to reach airports. Once at the airport, they collectively board planes, and upon arriving at their destination, they are individually dispatched to reach their final holiday spots. Surprisingly, many travellers have come to embrace this somewhat complex multimodal approach, despite its inconveniences. This shift is primarily driven by the numerous advantages it offers, including cost-effectiveness, speed, and the ability to access more remote and distant locations. It underscores how passengers are adapting to and appreciating the benefits of multimodal travel experiences.

2.4 Challenges for Public Transport Operators (PTOs)

We find ourselves at a critical juncture in public transportation, where a paradigm shift is on the horizon. The necessary technology is readily available, and passengers are eager to embrace a new and improved transportation experience. However, public transport operators (PTOs) face a complex array of challenges that must be addressed to make this transformation a reality. These challenges include V2X communication, fleet orchestration intelligence, GDPR compliance, cybersecurity, data collection, legal considerations, societal changes, economic implications, environmental costs, and more (Davidsson et al., 2016; Konstantas & Fournier, 2023).

In addition to addressing these challenges, achieving passenger and citizen acceptance is a pivotal aspect of the successful transformation of public transportation. This involves not only overcoming technological and regulatory challenges but also re-evaluating the cost structure and providing new services that align with evolving passenger expectations. Here we can mention some expectations of passenger like transparent and affordable pricing, recognising that passengers have diverse needs and preferences and offering personalised services, introducing innovative and convenient services that enhance the passenger experience, creating feedback channels for passengers to express their opinions, and finally effectively communicating the benefits of the new transportation options and services.

By focusing on these aspects, public transport operators can foster passenger acceptance and encourage the adoption of multimodal and innovative transportation solutions. This, in turn, can contribute to the success of the paradigm shift in public transportation towards more sustainable mobility.

In the subsequent sections of this document, we delve into the endeavours of public transport operators within the AVENUE project. They aim to address the questions raised earlier by drawing from their experiences gained through various deployments in diverse European cities.

3 The Geneva Sites

3.1 Objectives

Starting 2017, Transport Publics Genevois (TPG), the public transport operator in the Swiss Canton of Geneva, has been testing an automated vehicle on a fixed 2.1-km route with four bus stops. This first experience allowed TPG to be a partner in the European co-funded H2020 AVENUE project and to set up a fully automated transport system on the Belle-Idée estate in Thônex, Geneva, Switzerland, combining all future key elements of on-demand automated public transport (Fig. 2.1).

Fig. 2.1

The Belle-Idée with 75 stop points

As part of the AVENUE project, the aim was to demonstrate a comprehensive, shared, on-demand, door-to-door, and dynamically routed public transport service. This service utilised a fleet of automated minibuses operating within a geofenced zone, eliminating the need for fixed routes, set timetables, or traditional bus stops. The system was seamlessly integrated into the existing transport network.

Passengers had the exclusive option to book rides through a dedicated application. The project also explored the feasibility of allowing passengers to board and disembark at system-defined stop points without the need for physical infrastructure like ground markings or information display poles.

The objective was to develop a 100% automated public transport service in which the passenger is the only human link.

The following steps showcase how this objective was brought into practice:

• 1. A user on-site wants to be transported as soon as possible from his current position to the entrance of building 8, named reception.

• 2. The user takes his smartphone and opens up the tpgFlex on-demand application from TPG.

• 3. The user only has to submit his final destination. His smartphone already knows the user’s current position and, if not changed by the user, assumes that he is alone and wants to travel right away.

• 4. The moment the user validates one of the proposed offers according to his requirements regarding travel time, service level, and pricing, a door of the on-site depot opens, and one of the induction-charged electric minibuses drives out automatically and heads down to the user’s location via the fastest route.

• 5. Once the vehicle arrives at the pickup location, the user opens the doors of the minibus, gets in, and finds a seat. The vehicle automatically closes its doors and directly hovers to the user’s destination by choosing the fastest or most convenient route.

• 6. If another person makes a reservation at the same time on a very similar itinerary as the first user, the minibus will divert its trajectory to combine both reservations and serve the other user as well.

• 7. After it arrives at the user’s destination, the on-demand system can send the vehicle to another location in order to achieve a new mission, back to the depot to charge batteries, or place the vehicle at a key location since the on-demand system knows from experience that most of the bookings are being made at that bus stop at that time of the day.

The 100% automated system, comprising three to four vehicles, is continuously supervised by an on-site supervisor who is supported by two automated monitoring systems:

• 1. The vehicle monitoring system supervises the vehicle’s road-handling capabilities, such as monitoring harsh braking and object detection.

• 2. The passenger monitoring system is responsible for ensuring the passengers’ well-being while on board. It monitors for situations like passengers falling, experiencing health issues, engaging in altercations, or potential incidents like bag snatching. Additionally, this system checks the available interior space for the on-demand system.

Both systems provide the supervisor with visual and acoustic alerts, enabling him/her to take appropriate action. These actions may include observing monitoring tools to assess the situation, inspecting the interior of the vehicle, stopping the vehicle remotely, opening doors from a distance, or contacting emergency services such as calling an ambulance (Fig. 2.2).

Fig. 2.2

Belle-Idee on-site supervision office

It is worth noting that, as of the current date, there is no functionality for the supervisor to remotely control or manoeuvre a vehicle at low speeds (<6 km/h) to a safe position. This capability is not included in the project’s objectives due to technical limitations.

The on-demand system is fully accessible from the supervision office, allowing the supervisor to dispatch a vehicle to individuals who may not have access to a smartphone and have made reservations by telephone.

As of the date of this document and not included in the project’s objectives due to legal restrictions, the possibility of removing the operator (driver) from the vehicle is not yet allowed. However, the primary objective within the AVENUE project was to demonstrate the ability to operate without the need for an onboard operator.

3.2 Deployment

Due to the complexity of the project and the fact that most partners did not have experience in combining all these technologies, the deployment of the entire project was divided into three phases.

Phase 1: Between July and September 2020

• Mapping of the entire Belle-Idée site (summer map)

• Deployment of one vehicle on 45% of all available roads

• Creation of 27 stop points, 4 existing TPG bus stops, and 23 virtual stop points

• First tests with a safety operator and without passengers

• Training of safety operators

Phase 2: Between October and December 2020

• Deployment of/increase to two vehicles on 70% of all available roads

• Creation of/increase to 53 stop points, 4 existing TPG bus stops, and 49 virtual stop points

• Vehicle testing with a safety operator and without passengers

• First “fully automated on-demand system” testing.

• First software release 6.0 test

• Training of safety operators

Phase 3: Between January and March 2021

• Mapping of the entire Belle-Idée site (winter map)

• Deployment of/increase to three vehicles on 99% of all possible roads

• Tracing of a route to the TPG Seymaz bus stop outside the domain

• Creation of/increase to 75 stop points, 5 existing TPG bus stops and 70 virtual stop points

• Installation of software update to version 6.1 which enabled better road holding and easier on-demand

• Fully automatic depot entrance and exit of vehicles

• Vehicle testing with a safety operator and without passengers

• On-demand system testing with application booking, 100% automated

• Training of safety operators

At the end of 2021, software update 6.2 was installed which enabled for a dynamic update of an ongoing on-demand mission, hence vehicles already on their way to the next bus stop could be rerouted to being able to pick up or drop off passengers elsewhere (Fig. 2.3).

Fig. 2.3

Belle-Idee service travel information

At the end of 2022, the operating hardware and software of the vehicles was changed from Windows to Linux for better system stability.

During the deployment phases, every safety operator drove numerous kilometres on-site to test and identify issues as well as to get acquainted with the overall traffic after being trained and accompanied by the automated vehicle assistant trainer.

3.3 Achievements and Key Success Factors

Due to the willingness to make the project happen and the combined forces from the Swiss Federal and Geneva Cantonal Authorities, the AVENUE project partners, as well as the TPG Managing Board, the project team was able to create and set up the project and start giving demos for select groups of people and testers after 2.5 years.

Unfortunately, the partner in charge of the on-demand vehicle dispatching system went bankrupt soon after and the project was brought to a halt. It then took another 8 months to start for the second time. Since the service started its operation again, various public transport delegations from all over the world who systematically validated the importance of the solution have been welcomed.

At the end of the project, the only issue that kept the vehicles from driving fully automated was obstacles on the streets, such as wrongly parked cars, which the vehicle cannot bypass (yet). When the vehicle detects an object that is not moving, it will stop driving until the object has been removed. When it arrives at an object moving in the same direction as the vehicle, the vehicle will adapt its speed and follow the object from a distance. When an object is moving in another direction as the vehicle, the vehicle comes to a halt and waits for the object to go out of the way, and when the road is free, it starts driving again.

3.4 Recommendations

SAE Level 4+ for Public Transport

The development of fully automated vehicles has progressed at a slower pace than initially anticipated and previously announced by international car manufacturers. The most notable achievements currently available on the market can be categorised into two main use cases:

• 1. SAE Level 3 Automated Highway Lane and Traffic Jam Driving: This represents an advanced driver assistance system that enables automation for highway lane keeping and managing traffic jams.

• 2. SAE Level 4 Fully Automated Driving: This pertains to fully automated driving on pre-programmed virtual routes, typically at very low speeds.

These advancements illustrate the current status of automated driving technology, where different levels of autonomy and capabilities are achieved. The key distinguishing factors revolve around speed and the complexity of driving conditions. As of the present, achieving SAE Level 5 fully automated driving, as well as SAE Level 4 pre-programmed automated driving at speeds exceeding 30 km/h, remains unattainable. Commercial availability for such advanced capabilities is not anticipated before 2027. This, however, does not mean that the development and testing of fully automated vehicles need to stop. With regard to public transport operators, there is an important use case being created on a political level: thermic-powered vehicles are starting to be banned from city centres, parking spots removed, and speed limits reduced to 30 and even 20 km/h. This means that, as public transport is deployed in a well-defined geofenced area in the city, with all the roads and bus stops known in advance, a highly efficient public transportation service can be provided with SAE Level 4 vehicles, integrated in a MaaS system with the existing main lines of bus, metro, and tram.

As it is not economically feasible for public transport companies to equip every fully automated vehicle with a legally required safety operator, removing the safety operator as in SAE Level 5 is essential. The interim solution could be to create an extra SAE Level, between Levels 4 and 5, for public transport only which aligns with the following requirements:

• 30 km/h speed limit

• Pre-programmed virtual line driving

• Without a safety operator

• Supervision at near distance

The creation of such a “SAE Level 4+ for public transport” or SAE Level 4.5, hence between SAE Level 4 and 5, has already been proposed by the Association of German Transport Companies—VDV—in 2020.²

3.5 Future Developments

Due to the maturity of the project and Europe-wide recognition as being an urban public transport solution, TPG will continue gaining experience with new transport modes and technologies within the Horizon Europe Ultimo project. Firstly, TPG will scale up the current Belle-Idée living lab test site with the following steps:

• 1. Connect the Belle-Idée estate to the nearest train/metro station via a 1.5-km route.

• 2. Test at least one vehicle in 100% fully automated mode, without an operator on board, with the authorisation to supervise several vehicles simultaneously at a distance.

• 3. Reduce or eliminate the motorised traffic on the Belle-Idée estate completely.

• 4. Offer a made-to-measure mobility service to be able to transport everyone on-site.

• 5. Test the transport of goods in off-peak hours.

The second objective is to transform the existing classic driver-based rural on-demand service tpgFlex of the Champagne/Mandement region in the Swiss Canton of Geneva into an on-demand, door-to-door, fully automated, people and goods transport service integrated in the TPG transport system and later perhaps in a MaaS.

4 Denmark and Norway

In the AVENUE project, AM was running three test sites:

• Nordhavn, Copenhagen, and Denmark

• Ormøya, Oslo, and Norway

• Slagelse and Denmark

The Ormøya route was originally initiated without being a part of AVENUE, but an agreement has been made to include the site for 5 months, to begin with. The Norwegian site ended in December 2020. A new Danish site in Slagelse Hospital began in September 2021, with a focus on on-demand transportation service.

4.1 Nordhavn

The Copenhagen pilot site was situated in an area of the city called Nordhavn. Nordhavn is an active industrial port that is undergoing a transformation—turning into Copenhagen’s new international waterfront district offering residential and commercial buildings. When the development of Nordhavn is done, the area will house more than 40,000 residents and 40,000 employees.

Nordhavn aims at being an eco-friendly neighbourhood and contributes to boosting Copenhagen’s image as an environmental metropolis. The city should vibrate with life as a versatile urban area with a multitude of activities and a wide range of shops, cultural facilities, and sports facilities. The area is becoming more and more populated, and the need for local transportation is expected to keep growing (Fig. 2.4).

Fig. 2.4

The Nordhavn route area seen from above

Currently, the Nordhavn area is serviced by a nearby S-train station (approx. 1.1 km away) and bus stops located near the train station. There are, however, no buses or trains running directly in the area—creating a great opportunity for automated vehicles to function as a new public transport solution, connecting the area much better than it is today. In 2020 two new metro stations have been built—opening in the periphery of the neighbourhoods.

4.1.1 Objectives

The main users of the shuttle service have been the residents of Nordhavn (including families, children, and elderly), commuters working in Nordhavn, and visitors to the area. Several usage scenarios can thereby be anticipated:

• Ease mobility within the area for the residents and commuters working in the area.

• Used for the first/last mile from the main road/entry point to the different stops within the area for residents and commuters working there.

• Provide easier access from the main road to, e.g. the harbour pool, restaurants, and cultural facilities for visitors and families.

Planned services provided for the end users:

• The shuttles are free of charge during the pilot project in Denmark, so there is no ticketing yet.

• There are static bus stops providing the position of the bus, relative to the given stop.

• The real-time location of buses can be seen in the mobile application.

• Besides the bus stop signs, users can find information about the pilot project on the AM website and in the AVENUE mobile application.

During the project period, it was the aim to test the services developed through the AVENUE project, e.g. real-time position of the bus, on-demand booking, and accessibility for disabled persons.

4.1.2 Deployment

The first route was placed in the area called Århusgadekvarteret. This area was the first one finished, and residents started moving there in 2015. Since then, different squares, the harbour promenade, and a rooftop gym have been evolved and taken into use. Furthermore, special attention has been paid to developing local retail, so today there are supermarkets, cafes, restaurants, and different specialised retailers. There are several shared space areas on the route including a bathing zone.

The first route is a circle line around the area (blue line on the map below), making it easier to get around and to enter the area from outside Nordhavn. Our garage is located on the next peninsula close to Århusgadekvarteret (the red line on the below map) (Fig. 2.5).

Fig. 2.5

Map showing Nordhavn route

The pilot route is going to be in mixed traffic with cars, pedestrians, bicycles, etc. The area is, in general, a low-speed area with 20–50 km/h speed limits on the route, and in the 50 km/h limit areas, the recommended speed for cars is 30 km/h.

Operation facts:

• 2 AVs running (initially)

• Mon–Fri 10.00–18.00

• Loop route with six stops

4.1.3 Achievements and Key Success Factors

4.1.3.1 Passengers and Distance Driven

Passengers we counted every day of operation. In total, we have transported 1579 passengers. It is seen that most passengers were transported at the beginning of the pilot, possibly due to the route location close to the harbour popular in the summer time. Another possible explanation could be that locals got used to the shuttle as an attraction and as time went by a sense of novelty disappeared (Fig. 2.6).

Fig. 2.6

Distance and passengers per month in Nordhavn

A total of 2.417 km has been driven in Nordhavn. The count can be seen increasing in almost linear progress since August 2020 (Fig. 2.7).

Fig. 2.7

Accumulated distance driven in Nordhavn

4.1.3.2 Driving Speed and Automated vs. Manual Mode

The driving speed is seen to be increasing slightly over the operational time period from an initial 7 km/h to 7.89 km/h. This is despite an increase in manually driven kilometres primarily related to an increase in parked cars on the route.

In total, close to 992 h have been driven during the pilot. In regard to the navigation mode, 82.6% of overall driving on the route was done in automated mode. The rest of 17.4% driven in manual mode was mostly due to a large amount of illegally parked cars on the route and due to roadworks. Driving manual to and from the garage is filtered out.

4.1.3.3 Issues Reported on Route

The main issue reported through the safety operator app was parked cars on the road. The parked cars disturb the driving, as the AV is not able to diverge from its planned trajectory, and hence the operator is forced to do a manual takeover in order to pass the parked vehicle. In total, this has been registered 1995 times.

The issue of parking on the road has been reported in several locations on the route. Mostly around Århusgade, Sandkaj, and Göteborg Plaza. The issues with parked cars in the Nordhavn area were significantly contributing to impeding the operation, finally resulting in its closure.

4.1.3.4 Downtime and Cancelled Operation

The following figure illustrates the distribution of the five most common reasons for downtime/cancelled operation per downtime hour it caused. The downtime was primarily due to roadwork on the route, which made driving in the designated trajectory impossible and manual takeover necessary (Fig. 2.8).

Fig. 2.8

Causes of downtime in Nordhavn by downtime hours

4.1.4 Recommendations

During the deployment of the shuttles in Nordhavn, multiple learnings were achieved. They are described in the following.

4.1.4.1 Object-Detection Challenges

The Nordhavn area is packed with restaurants, cafes, and shops with outdoor seating and display areas. Larger parts of the Nordhavn route are shared space areas, where pedestrians, bicycles, and vehicles interact in a shared road, rather than separated lanes. This type of city planning causes many situations where the automated shuttles detect obstacles and stop as a consequence. Furniture and other objects cross into the driving lanes, causing the shuttle to stop and requiring the safety operator to move the objects away from the driving lanes or manoeuvre manually around obstacles. For a completely automated operation without a safety operator, the area would probably not be ideal because of these shared space areas, and at least it would require significant technological breakthroughs before an automated vehicle would be able to operate in the Nordhavn area without a safety operator. This is also due to narrow roads, where two cars cannot pass each other without human interaction.

4.1.4.2 Increased Mixed Traffic in High Seasons

During the summer period, the swimming areas at Nordhavn attract many visitors who arrive by bike. Due to the large number of bikes, bikes are sporadically parked outside of the appointed areas for bike parking. This caused difficult driving conditions, and many operational days were shortened as the vehicles were not able to drive due to the many parked bikes, and it was deemed impossible to have the safety operator remove the bikes during every round of driving.

4.1.4.3 Consequences of Construction Work

As the Nordhavn area is under heavy construction, the shuttles have had to operate in an area with many trucks and work-related vehicles being parked illegally and shutting off parts of the route. This has caused delays and cancellations of the operation for shorter periods and sometimes days.

4.1.4.4 Lack of Parking Spots Compared to the Number of Cars

The Nordhavn area is a very busy area with local residents, offices, shops and restaurants, and many daily visitors. Some of these visitors travel by car and can find it very difficult to find a place to park the car during peak hours. This resulted in many illegally parked cars on the route, and the safety operators have been taking over manually multiple times every day to overtake parked cars. Driving in an area that requires manual overtaking does affect the total percentage of kilometres driven in automated mode.

4.1.4.5 Low-Speed Limit

Most streets in Nordhavn have a speed limit of 30 km/h which suits the automated vehicles’ speed capabilities. In general, a low gap between the vehicle’s top speed and the speed limit provides a more safe operation, with less risky overtaking from other road users. This contributes to Holo’s general assessment of the Nordhavn area as a safe environment to drive and test low-speed automated vehicles.

4.1.5 Future Developments

Unfortunately, the route in Nordhavn had to be shut down, and AM has investigated several scenarios to still meet the learning objectives for the AVENUE project. In the following the process and reasoning for shutting down the Nordhavn site will be further elaborated.

4.1.5.1 Complications in Nordhavn

In the fall of 2020 AM started to investigate the potential for expanding the Nordhavn site. AM investigated several routes which could be added to the already active route in Nordhavn. The selection and evaluation focused on last-mile transportation from the two metro stations and to both residential and various business areas of Nordhavn.

In November 2020 all the potential new routes were discarded by By & Havn, as construction plans for the area would interfere with the automated minibuses. The area of Nordhavn is undergoing heavy construction, and due to delays in the construction work, several roads have been unavailable to use for the remaining period of the AVENUE project.

The above challenges in Nordhavn caused AM to seek alternatives for automated operations in the Copenhagen area. AM has evaluated several scenarios for how to still meet the objectives of the AVENUE project and decided to use the AVENUE shuttles on an existing Amobility site. In order to minimise the risk of delaying operations substantially, caused by long approval processes, AM investigated the use of an existing site in Denmark for the remaining months of operations in AVENUE.

AM was granted approval to start operations at Slagelse Hospital (Denmark) in March 2020. However, the project had been paused due to COVID-19. In the fourth quarter of 2020 AM and Movia (PTA) decided to restart the project in August 2021. AM and Movia discussed the possibility of integrating AVENUE as part of the Slagelse Hospital site and agreed to proceed with this possibility.

Even though this alternative only can support operations with Navya vehicles, AM has identified several crucial learnings in relation to on-demand and integration with PTA for ordering.

4.2 Ormøya

Due to the delays in launching the Danish demonstration site, it took some time before all four buses were in operation. Therefore, in May 2019 AM agreed with the consortium to include our subsidiary in Norway as a third party, so that two of AM’s AVENUE buses could be deployed on a route there. This way AVENUE would still gain useful insights into the operation while awaiting the launch of the Copenhagen site.

Originally the plan was to include the two buses on the route Akershusstranda in central Oslo for 5 months beginning in June 2019. This would be a route with four buses running and the service fully integrated with existing public transport in Oslo. However, due to heavy construction, this route had to be cancelled. It was therefore decided to integrate AM’s AVENUE buses on the second route “Ormøya” just outside Oslo centre. Ormøya is an island south of Oslo city centre connected by a bridge to the mainland and a bridge to a second island called Malmøya.

4.2.1 Objectives

AM is collaborating with Oslo Municipality, the Norwegian Public Roads Administration, and Ruter³ on a 3-year self-driving trial project. The project is an important milestone in the process of getting self-driving buses to the Oslo area. Oslo and Akershus wish to have 0% emissions across their public transportation, and this project will test if self-driving buses can support these ambitions for a sustainable public transport system. The end goal is for automated buses to be part of Ruter’s regular offer in a few years.

The main purpose of the project was to investigate what self-driving vehicles can mean for everyday logistics in a neighbourhood. By increasing the frequency of public transport by means of small self-driving vehicles, the goal was to reduce the need for private cars in the area. One road leads in and out of the two islands that have a total of around 500 households. The local residents have a 12-m bus service which departs around once an hour for most of the day. On the mainland just off the inland is one of the major thoroughfares going into Oslo from the south, Mosseveien/E18. This main road has frequent express buses going in and out of Oslo. The automated bus service provided a high-frequency last-mile solution for the residents of Ormøya and Malmøya which connected them to the express service on Mosseveien/E18.

4.2.2 Deployment

The route is 1.6 km one way (3.2 km round trip) and has six bus stops. It runs from Nedre Bekkelaget bus stop which is located near Mosseveien/E18 where users can access high-frequency express buses to and from Oslo. Also near this end point is the local area public school that kids from Malmøya and Ormøya attend.

The other end point, the Malmøya bus stop, is right on the landing on the island of Malmøya where there is a turning place for the vehicles. This bus stop is also located close to a marina, where lots of Oslo residents keep their recreational boats. The four other bus stops are evenly distributed along the two end points. The bus stop Mailand is also located close to a public beach/swimming area and a marina which attracts lots of visitors in the summer. The route can be seen below with the stops marked (Fig. 2.9).

Fig. 2.9

Map of Ormøya route

The speed limit on the entire route is 30 km/h, and it contains several speed bumps which generally keep the speed in the area low. The condition and build of the road vary quite a bit along the route. Several places are very narrow, just barely wide enough for two vehicles to pass each other, and several stretches have poor asphalt quality. There is also a lot of vegetation close to the route.

In order to be able to offer the inhabitants a valuable self-driving travel service, we had to ensure high operational stability along the stretch. This has been challenging due to several elements, and we have therefore made ongoing adjustments in the offer to explore what it takes to ensure stable and reliable operation. Operational stability will be a success factor in initiating new, more complex self-driving bus lines in the years to come.

4.2.3 Achievements and Key Success Factors

4.2.3.1 Passengers and Distance

In total, 6637 passengers were transported and 22,984 km driven on the Ormøya route. During the 1-year pilot, there were over 5233 h of operation (total number when all three vehicles are added together). That is equal to approximately 395 operational days or 131.67 full operational days with 3 vehicles (1 operational day = 13.25 h) (Figs. 2.10 and 2.11).

Fig. 2.10

Accumulated distance and passenger count at Ormøya

Fig. 2.11

Distance and passenger count per month at Ormøya

The route in Ormoya was severely impacted by operational challenges and COVID-19 lockdowns in some months during the project period. This resulted in some periods with few passengers and less distance driven. In terms of passenger distribution, passengers were using the shuttle service for transport every day of the week. However, there was a somewhat higher number of passengers during the weekdays than the weekend.

4.2.3.2 Automated vs. Manual Driving

In regard to the navigation mode, 93.8% of overall driving on the route was driven in automated mode. The rest of 6.8% driven in manual mode was mostly due to technical issues. Driving to and from the garage is filtered out.

Driving speed on the Ormøya route was between 9.44 and 10.53 km/h. The overall average speed was approximately 10 km/h. By the summer of 2020, a recommissioning was done to improve driving in areas of the route.

4.2.3.3 Issues Encountered on the Route

When looking at the distribution of different issues reported by the safety operator, again for this route “parking on road” is the most frequent issue occurring with a total occurrence of 1725 times. Operators fall as many as 62 times, signalling a safety issue in the working position of the operators (Fig. 2.12).

Fig. 2.12

Issue distribution at Ormøya

4.2.4 Recommendations

4.2.4.1 Public Transport in Oslo

The service at Ormøya functioned as an integrated part of Ruter’s public transport offerings in the general Oslo area. This meant that the bus required a standard ticket which gives access to the entire network and that the service was included in Ruter’s overall route planning tools for users. Neither Holo nor Ruter performed checks of valid tickets.

By all indications, users did not complain about the ticket requirement but viewed it as natural for service in Ruter’s network.

4.2.4.2 User Experience

During the final quarter of operations, Ruter conducted a user survey around the shuttle. The survey was conducted by interviewing people walking or moving around the general area of Ormøya. In total Ruter collected 107 interviews each lasting 5–8 min during weeks 38, 39, 41, and 42 in 2020. The main purpose of the survey was to evaluate what users and local residents think of the self-driving bus service.

In general, respondents were positive towards the tests and felt that it was safe. However, the survey also showed clearly that the service was not providing a valuable mobility solution. Approximately 82% of passengers took the bus merely out of curiosity, while only 12% used the service for their daily commute. When local residents at Ormøya and Malmøya were asked why they did not use the automated bus for their daily mobility needs, two of the top answers were the low speed and that the bus did not go where they needed to go. Furthermore, low reliability was also noted as a reason why the service was not used for their daily mobility needs.

4.2.4.3 Vegetation and Snow

The narrow road on Ormøya makes this site especially challenging when it comes to vegetation. There is little to no margin for the growth of vegetation. Both vegetation on the sidewalk and overhanging bushes and branches are stretching into the safety zones interfering with smooth operation.

Another issue at Ormøya is the high and low overhanging branches interfering are then detected by the vehicle causing it to brake or slow down (Fig. 2.13).

Fig. 2.13

Photos showing the vegetation close to the vehicle trajectory on the route in Ormøya

The rapid growth of vegetation in the summer months has led to constant interference with operations. In periods operations have been halted until the vegetation has been properly cut. In the summer months, April to September, we have seen the need to cut vegetation up to every second week.

Vegetation-related issues have heavily impacted the stability of the operation on this route. Comparing this site to other routes with less vegetation clearly shows a negative impact on the performance of the vehicles. Additionally, sudden braking caused by vegetation is a major safety concern. The sudden braking can cause cyclists, pedestrians, and following cars to potentially collide with the vehicle when it makes unexpected and seemingly irrational decisions like braking hard for a small branch near the trajectory.

In the winter months snow and ice proved a similarly big challenge. The operational design domain (ODD) for the Navya Arma vehicles states that they cannot operate when snow is falling or when the route is covered in snow. Falling snow proved quickly to be unsuitable for operation. With snow in the air, the vehicles make continuous false obstacle detections, and every snowfall resulted in a halt of operations.

The year 2020 did not see very much snow at the Ormøya site, and only on a couple of occasions did snow on the ground cancel operations, mainly because snow clearing was not done well enough to clear the vehicle’s trajectory as highlighted in the image below (Fig. 2.14).

Fig. 2.14

Photos showing snow and snow banks in and close to the vehicle’s trajectories on the route at Ormøya

4.2.4.4 Major Safety Issues

Already in the risk analysis of the route, before the operation commenced, the major safety concerns were identified as other road users behaving dangerously around the shuttle due to low speed and slow reaction to clear road or similar. Also, passengers and operators getting hurt inside the shuttle due to sudden and hard braking were highlighted along with intentional disturbances. These safety concerns proved to have been correctly assessed from the beginning, and they were the major safety concerns throughout the operation.

Through the Holo operator app, Holo was able to collect data on the potentially most dangerous situations that occurred on the route. The four most important categories that data was collected on are dangerous overtaking of the vehicle, intentional disturbance, operator falls, and passenger falls. Two major safety concerns that remain with an operation like the one on Ormøya are:

• The operators of Navya Arma vehicles do not have ideal working conditions. Sudden and hard braking has resulted in several mild injuries because of the lack of proper seating and suitable working positions.

• The low speed of the shuttle does cause other road users to act irresponsibly and attempt overtaking in areas that are not suited for overtaking like areas with bad overview or near pedestrian crosswalks.

4.3 Slagelse

The Nordhavn area should originally have been finished in 2020, but due to major delays in construction plans, parts of the Nordhavn route (streets) were closed permanently for longer periods of time during the remainder of the AVENUE project in that area. Hence, AM and Copenhagen area PTA Movia agreed on introducing an AVENUE shuttle on the two shuttle project at Slagelse Hospital. Here the main learnings were aimed at on-demand driving and integrations with public transport PTA Movia, their client systems, etc.

The Slagelse site was an interesting case, as the distances between the departments in the hospital are too long for patients to walk between them and a shuttle moving between different departments and connecting parking lots would improve the mobility in the area.

4.3.1 Objectives

The goal for the on-demand service is (1) hospital staff can book trips for patients and visitors, and (2) patients and visitors can book their trip. Transportation between parking lots and other entrances makes good sense because of the large distances in the hospital area. The status of the trips booked is updated through the tool used by hospital staff, hereby passengers can get info on when to expect pickup time. Safety stewards will greet passengers when the vehicle arrives at the pickup point. Without any direct input from the safety steward, the vehicle will begin its trip once the doors are closed.

The on-demand service initiated with one vehicle. The second vehicle was included in the on-demand service as soon as possible. The goal was to service both vehicles on demand as much as possible.

At this stage, there were some technical activities that need to be performed in order to prepare the service:

• Update Navya vehicle software to 6.1

• Integrate Holo system with Navya API

• Integrate Holo system with a dispatcher at Movia

• Movia to develop User Interface for booking

There will be a full focus on the technical integration and stability/performance of the service created, in order to gain the full learning experience in trying to service a robust on-demand service.

There will be less focus on the development of apps and screen content, hereby limiting the investigations into the whole automated minibus customer journey. This is in order to favour the technical development and achievement of back-end software—customer interfaces are owned by Movia.

This priority is possible because the user who is booking the trips will be hospital staff and passengers via the Movia interfaces. The passengers will receive the necessary info from the hospital staff and the interfaces. Furthermore, the safety operator still has to be present in the vehicles and will be utilised to give the needed information to the user.

The practical test process will look like this:

• Outline vehicle behaviour in all possible on-demand situations on Holo’s test track in Copenhagen. SW version 6.1. Holo internally dispatches missions to vehicles.

• Outline vehicle and integration behaviour in all possible on-demand situations with missions received from Movia. Still on Holo’s test track.

• Move to Slagelse and perform similar tests on the real route. Reach a satisfactory level of performance before servicing passengers.

4.3.2 Deployment

The shuttle was driving on a 770-m-long stretch on Fælledvej at Slagelse Hospital. Besides driving on Fællesvej, the shuttle was driving in five parking areas, with multiple stops, to turn the shuttle and re-enter the stretch of Fælledvej. The route and the stops can be seen below. Besides the stops placed in the parking areas, the shuttle also stopped at the western part of Fælledvej, in both directions (Fig. 2.15).

Fig. 2.15

Map showing the route at Slagelse

The shuttle was driving on sections of the road with different infrastructure settings. These different sections are shown in the following picture and further described below (Fig. 2.16).

Fig. 2.16

Map showing the route sections at Slagelse

4.3.2.1 Red Section

The red section is a 240-m stretch with parking spots alongside the road on the south side, parking booths between the two driving lanes, and a double-edged bicycle lane on the north side of the stretch. There is a sidewalk for pedestrians on both sides of the road and no facilities for scooters in the eastbound direction. Meaning, scooters and, to some extent, bicycles must use the road in the eastbound direction. The speed limit for the stretch is today 50 km/h, and the red section has a width of 6 m. On the red section, there are three road connections with an unconditional right of way for the road users on Fælledvej. The municipality of Slagelse has approved the speed limit to be decreased to 30 km/h during the hospital pilot project.

4.3.2.2 Green Section

The green section is a 300-m stretch with parking booths between the two driving lanes and a walking path on the west and south sides of the road, as shown in the picture below. There is no sidewalk on the north side of the stretch (meaning that it is assumed that pedestrians may walk on the roadway on the north side). On the green section, there are no implemented facilities for scooters or bicycles, meaning that they will drive on the roadway on the green section. The speed limit for the stretch is today 50 km/h, but it is assumed that the driving speed is lower because of two sharp curves on the stretch, with a small turning radius, which cannot be driven with 50 km/h. The green section has a width of 6 m. On the green section, there are five entry roads for parking facilities. From the parking facilities to Fælledvej drivers have an unconditional obligation to give way. There is a road connection in one of the sharp curves, where drivers have an unconditional obligation to give way to drivers on Fælledvej. The municipality of Slagelse has approved the speed limit to be decreased to 30 km/h during the Hospital Trials.

4.3.2.3 Blue Section

The blue section is a 230-m stretch with barriers separating the road and newly implemented speed signs, recommending 20 km/h on the stretch. There are sidewalks for pedestrians on both sides of the stretch. There are approximately 35 m of bicycle lanes on both sides by the exit from entrance 11. On the remaining part of the stretch, the road is shared with bicycles and scooters. The driving lane has a width of 3.25 m in both directions, and there are five cross sections with an obligation to give way for the traffic on Fælledvej. Slagelse municipality has approved speed limits of 30 km/h during the Hospital Trials.

A regular Movia bus (line 902) operates on Fælledvej with 30-min intervals during weekdays. The line has two stops (for each direction) on the part of Fælledvej that the shuttles were driving on. At the main entrance of Slagelse Hospital, a patient bus of the same size as Movia’s regular bus departed a couple of times during the day. The patient bus had a marked parking area in front of Slagelse Hospital that has been placed outside the self-driving shuttle’s route. During entry and exit to and from Slagelse Hospital, the self-driving shuttle and the patient bus could lock. This required the safety driver to manually take over and give way for the patient bus.

4.3.2.4 Parking Conditions

The area in which the automated minibus had been operating has a high density of parking areas. There are parking facilities in Fælledvej, where drivers dismount their cars directly onto the road. Further, there are marked parking booths on these parking facilities, which the automated minibus had to drive past. Besides parking areas defined by regulations or signs, there was a high degree of parking outside of these designated areas, as seen below. Based on this, it is uncertain how much effect the parked cars outside of designated areas will have on the operation of the shuttle. Unwanted stops and brakes may occur if parked cars are blocking the route of the shuttle (Fig. 2.17).

Fig. 2.17

Photos showing parking conditions at Slagelse

4.3.3 Achievements and Key Success Factors

4.3.3.1 Distance and Passengers

In total, 1659 passengers were transported and 4595 km driven on the Slagelse route. During the pilot, there were over 2117 h of operation (Figs. 2.18 and 2.19).

Fig. 2.18

Total number of passengers and distance driven in the Slagelse project

Fig. 2.19

Distribution of passengers and distance driven per month in the Slagelse project

4.3.3.2 Automated Vs. Manual Driving

In regard to the navigation mode, 93.9% of overall driving on the route was driven in automated mode. The rest of 6.1% driven in manual mode was mostly due to technical issues. Driving to and from the garage is filtered out.

4.3.4 Recommendations

4.3.4.1 User Experience Learnings

4.3.4.1.1 Patients

Moving patients from A to B on a hospital site has proven to be a very good use case for the Navya vehicles, as patients often do not need high speeds but rely on the comfort and the ability to be moved. This means that even if the shuttle’s top speed is 18 km/h, the patients still experience high value, as the alternative would be to walk or wait for local flex taxis, etc.

4.3.4.1.2 Relatives/Visitors

Being a relative at a hospital site is often associated with difficult parking conditions and lots of walking on the hospital sites. With the service provided in the AVENUE project, relatives and visitors have had the opportunity to park in large parking areas away from the hospital entrances but then being carried by the shuttle to the entrances. This means less congestion and car hassle at the entrances of the hospital, resulting in emergency vehicles having more space and less reckless parking from visitors.

4.3.4.1.3 Employees

As for the relatives and the visitors, the employees have used the shuttle service to get from larger parking areas to the entrances of the different departments—but also to accompany patients from one department to another—cutting off time from walking and waiting on flex taxis.

4.3.4.2 Performance Learnings

4.3.4.2.1 Low-Speed Environment

The Slagelse Hospital site has shown to be a very good fit for the deployment of the Navya vehicles. Being the site in Amobility’s history with the highest automated uptime percentage, 93.9% of the planned operation has been delivered in automated mode. The environment at Slagelse Hospital is a low-speed (20–30 km/h) zone. Passengers, employees, and relatives visiting the hospital with cars drive slowly and are in general not in a hurry—meaning less dangerous overtaking and reckless driving around the shuttle.

4.3.4.3 Low Complexity Environment

The roads on the Slagelse Hospital are public with low-speed zones. The roads are wide and have designated lanes for pedestrians, bicycles, and cars. This has provided a good environment and level of complexity for the deployment of the Navya vehicles. Historically Amobility has experienced many issues with bicycles sharing the roads with cars, resulting in many overtaking close to the vehicle causing severe and hard braking of the vehicle—a huge risk for the passengers inside the vehicle. Having this environment, more suited for the Navya vehicle, has had a huge impact on the high uptime in a positive direction.

4.4 Conclusions

The main conclusions that can be drawn from the Amobility efforts in the AVENUE project and the implementation of self-driving vehicles in Nordhavn, Copenhagen, Ørmoya, Oslo, and Slagelse, Copenhagen, are summarised in the following points.

• The approval process in Denmark has been a slow start where multiple stakeholders in the approval process had to learn the basics of self-driving vehicles, resulting in an approval process with inspirations from the railway approval systems—hence a documentation level out of the ordinary, seen from the perspective of the self-driving vehicles industry, where the technology is still at a low maturity level. A new approval is on average considered to take between 9 and 14 months in Denmark, and the adjustments and changes to an already given approval are rigid and require a new approval in most cases.

• The entire approval process in Denmark with multiple approvers is very expensive and requires a huge amount of work from Amobility, in terms of documentation, separate approvals, tests, risk assessments, and so forth.

• The approval process in Norway is structured in a very different way, allowing for a more dynamic and communicative approach. It is Amobility’s experience that the Norwegian system is more agile and ready to adjust to the rapid development of self-driving technologies. A new approval is on average considered to take between 3 and 5 months in Norway, and the adjustments and changes to an already given approval are seen as dynamic and innovative.

• Based on the experiences and knowledge that Amobility has gained so far, a recommendation for the Danish legal framework has been provided, with both direct and indirect changes that can be made to ease up the approval process.

• Amobility has suggested that when testing/driving in SAE Level 3 or lower, only approval from the DRSA, local police, and road owners should be necessary. Furthermore in projects with SAE Level 4–5 operation, AM recommends having either (1) approval from the appointed assessor, DRSA, police, and road owner(s) or (2) an approval by the Danish Road Directorate (taskforce), DRSA, police, and road owner(s). AM also suggests authorities to allow approvals for areas rather than single route approvals. Also, AM recommends that the Danish authorities target public funds more towards AV implementation. Ultimately AM recommends removing the political approval process entirely. For each project to have its own executive order completely diminishes the flexibility and innovation for existing and future pilot projects.

• Inspiration and know-how from the Norwegian approval system have been recommended to the Danish authorities as a comparison, aiming at showcasing the potential loss of innovative projects in Denmark if the approval system does not adapt to the innovative and rapid developments of self-driving technologies.

  • From a technical perspective, Amobility has experienced that the technology and industry in general have developed at a slower pace than anticipated and the objectives of AVENUE have been difficult to meet.

  • The Navya vehicles are not able to drive in SAE Level 4 as there are still many aspects of the safety-related features that need to mature before Amobility can take out the safety operator from the vehicles. As a part of the risk assessment in Denmark, there are certain jobs related to risks in traffic that the safety operator has to perform to verify and support the vehicle in public transport.

  • During the project of AVENUE, Amobility deployed Navya vehicles at three different sites in both Norway and Copenhagen, Denmark. Experience has shown that the route in Slagelse offers the best circumstances for operating Navya vehicles with the right amount of complexity still ensuring the opportunity for high uptime. The maintenance cost is historically low in Slagelse, and major breakdowns and issues have not been experienced. At the same time, passengers at Slagelse Hospital have been very happy with the service of moving patients from department to department. This highlights why the use case in Slagelse is better than the other sites, as the patients do not demand speed but comfort and the ability to avoid walking. They are not in a hurry, as Amobility experienced passengers in Nordhavn to be due to the city centre position having both residents, employees, and tourists.

  • To be able to reach higher speeds with the shuttles, the sensory systems have to be improved allowing for a larger safety net of lidars, etc. Given the Amobility experience in Nordhavn and Ormøya, the average speed is still as low as 7–8 km/h. The top speed of the Navya vehicle is currently 18 km/h. It is Amobility’s opinion that the average speed is more important to increase than the speed of vehicles as the most important improvement for the operation is to move the passenger faster from A to B and not at the highest top speed. The objective is to try to raise the average speed to 14–16 km/h.

  • Driving in snow and heavy rain in Oslo has turned out to be quite the challenge for the Navya shuttles. The big snowflakes are often confusing the lidars and tricking the system into thinking that there is an object that the vehicle has to avoid, hence usually a severe braking causing the passengers and safety operator to fall.

  • The map of the vehicle (from the commissioning process) does not take into account that vegetation changes from season to season, even from the day the commissioning was executed to the day the operation begins. This has caused many problems for the vehicles that have performed many hard and severe braking. Often the branches are seen as obstacles in front of the vehicle in the safety zones. Amobility and the clients have spent both time and resources on keeping the vegetation to a minimum (preferably exactly as it was during the commissioning).

5 Lyon, France

5.1 Objectives

Launching a transport service with automated vehicles open to the public on an open road in the Confluence district in September 2016, Keolis Lyon is an early adapter of this research field. They are part of a local dynamic that is pushing the development of automated vehicles, with a political vision supported by the public transport operators (PTOs), as the means of the future to complete public transport offers. This first experiment allowed the creation of a public transport line serving a 1.6-km-long neighbourhood, thus connecting residents to a tramway line. This helped Keolis Lyon to lay the foundations for the use of automated vehicles as part of a strategy to bring passengers back to other modes of the urban transportation network (tramway, metro, high service level bus) making the public transport network more attractive.

The objective is to test a technology capable of revolutionising public transport, by offering PTOs a new way of developing their transport offer. Keolis Lyon is hoping that automated vehicles could modify the distribution of funding in public transport, which decides whether new public transport lines can be developed.

To understand the context of this project, it is important to acquire knowledge about the economic background of public transport. In Europe, mobility and the development of public transport in urban areas are two important issues. Therefore, the involvement of public authorities in urban transport is essential to ensure the availability of a quality service at affordable prices. The management of a high-quality urban transport network represents an important public service mission (PSM) for the community, as it justifies the granting of subsidies to transport companies.

Indeed, most urban transport networks are not commercially viable. The public authorities must therefore cover all or at least part of the costs of investment in infrastructure and vehicles. Operators also receive subsidies for the operation of public transport, which vary in size from community to community. For example, in 2002, subsidies covered 45% of operational costs in Helsinki, 50% in Stockholm, 51% in Lyon, and 60% in Brussels. To meet the mobility needs of the urban population and the expansion of the areas served, the supply of urban transport has increased considerably over the last two decades. The French networks have become increasingly busy, and operating costs have increased very rapidly. However, the networks have not compensated for the increase in their operating costs with fare increases. Thus, from 2000 to 2015, the average rate of coverage of operating expenses by fare revenues has been steadily decreasing. It fell from 31% to 18% in networks with 50–100,000 inhabitants, from 33% to 18% in those without metro or tramway systems, and from 37% to 32% in those including them (Coppe & Gautier, 2004).

Reviewing these figures, it is important to point out that a significant part of the operating costs is represented by the drivers’ wage bill. This is how the idea that the creation of a new public transport line depends mainly on the expected volume of demand can be confirmed. Following this logic, it can be considered that if the operating costs of a new line are fixed according to its frequency (driving costs, investment, and maintenance of vehicles and infrastructure), the share of the territorial authority will decrease with the increase in frequency. Busy lines are therefore more profitable because they benefit from additional revenue. Thus, in an area with insufficient passenger density, the local authority will not be able to support the costs of creating and operating a new line.

In this paradigm, the automated vehicle could change the equation by considerably lowering operating costs, particularly by reducing the cost of labour, which is not negotiable for vehicles with drivers.

The experiment carried out in 2016 was therefore intended to enable a better assessment of the capacities of automated vehicles to accompany public transport networks in this transition.

5.2 Deployment

In 2018, after 2 years of operation of the experiment in the Confluence district, the opportunity to join the AVENUE consortium and to participate in the H2020 project led Keolis Lyon to look for a new and more ambitious field of experimentation. The Parc OL experimentation made it possible to work on several use cases.

Economic use case: The Parc OL is the stadium of the professional football team which attracts about 40,000 spectators on event nights. The ambition is to create an economic activity around this stadium, in a project called “OL Vallée”. Therefore, there was a certain number of buildings to be built throughout the experiment which would completely change the way people travel in the area (leisure centre, restaurants, office buildings, medical centre and analysis laboratory, hotel). With a progressive increase in the number of passengers per day, the automated shuttles were seen as a way to offer a complementary public transport solution to accompany the economic development of this district. While it was clear that in the long term a mode of transport with more capacity would have to serve the OL Valley area, the automated shuttles were to provide a link during the increase in transport demand.

Technological use case: The first phase of experimentation was to connect the nearest tram stop (T3 Décines-Grand Large) with the Parc OL, by carrying out a route of approximately 2.6 km round trip. The route that the automated shuttles were to take allowed them to be integrated into a large flow of cars, passing crossroads and roundabouts for which the automated shuttle technology alone was insufficient. It was, therefore, necessary to equip four intersections with communicating traffic lights and to develop communications between the automated shuttles and the traffic light intersections. Thus, several insertion levels with or without priority for the automated shuttles were developed.

Social acceptance use case: The route of this experiment required the shuttles to cross a sensitive neighbourhood, in which road incivilities are regularly observed by the police. The question of the appropriation of these new technologies by the local population was therefore a central issue. This project was an opportunity to meet the inhabitants of the district to present the project, the technology used, its operation, and its constraints. This promotion of the project was done through several communication channels and by going to meet the schools of the district. The objective of this phase was to allow residents to take ownership of the project and to integrate the automated shuttles into the urban landscape as a new everyday object (Fig. 2.20).

Fig. 2.20

Parc OL route

5.3 Achievements and Key Success Factors

Unfortunately, after an inauguration of the experiment on 18 November 2019, the crisis of COVID imposed the stop of this experimentation on 16 March 2020. New health regulations have forced French public transport operators to respect social distancing in their vehicles. However, given the size of the automated shuttles, respecting these social distances was not possible. Although a resumption was allowed in September 2020, the context had evolved and no longer allowed for qualitative observations of this experiment. The number of passengers had dropped significantly on the whole network, with traffic of less than 70% of the normal attendance. The figures presented below are therefore only for the period before the COVID-19 crisis (Fig. 2.21 and Table 2.2).

Fig. 2.21

Kilometres travelled/theoretical kilometres

Table 2.2 Shuttle data

Monitoring of the experiment:

• 4000 passengers

• 6400 km

• 1.13 passengers/journey made

Awareness among CT users (650 people from the panel to link with the social impact chapter):

• 46% are aware of the experiments in Lyon.

• 55% have a favourable opinion of automated vehicles.

• 33% would be ready to give up their car if a TAD service with automated vehicles was proposed.

Rate of completion of the service:

• 85% (theoretical kilometre to kilometre to be achieved by a shuttle when it is out of service and kilometre to be achieved during operational stops caused by external reasons).

• 65% if all theoretical kilometres are kept.

• Equipment reliability problem and known cause:

  • Prototype vehicle

  • (1800 km of untested operation + Transpolis tests where the limits of the shuttles were tested)

The average speed was provided by Navya. It represents the average of all operating speeds. The average speed is usually low because it includes slowdowns when approaching stops where passengers get on and off the vehicle, as well as all the slowdowns caused by events on the route (red lights, crossing times at junctions, awkward parking, etc.).

5.4 Future Development

While the question of the technical capabilities of automated vehicles seems central to the anticipated developments, it is necessary to put them in perspective with the needs of mobility services. For automated vehicles to be commercially successful in public transport, it is imperative that the service offered is relevant.

By integrating this reflection, it is easily understandable that the principle of a fixed line, with regular and reliable schedules, can be a relevant service. The regularity of the passages could facilitate the habitual travels of the users, as well as their confidence in this service. Nevertheless, this is only true when the frequency of passage of automated vehicles at the stops is very high. This allows the customer to remove the constraint of planning these trips. In the same way as a metro, we do not look at the time of arrival because we know that they run frequently and that the waiting time between two metros is always acceptable. The logic for a fixed line of automated shuttles is therefore the same. Even if this is feasible, it implies a constraint on the size of the vehicle fleet. To be able to guarantee a high-frequency operation, the transport operator will have to exponentially increase the size of their fleet with the increase in the size of the area to be served. This principle may therefore create a strain on investment requirements and operating costs. The increase in fleet size implies more vehicles but also more supervisors and intervention teams, which will still be needed in case of problems and disrupted situations.

In this paradigm, we could therefore deduce that the development of fixed public transport lines operated with automated shuttles is only relevant considering certain characteristics of the site, including size or certain constraints related to the types of routes available for automated shuttles (e.g. industrial site with a single road).

Based on these observations, we can therefore anticipate that the aggregation of automated shuttle technology and vehicle fleet management technologies to offer an on-demand transport service would offer an answer to several constraints.

5.4.1 The Constraints of Availability for Users

With an on-demand transport system, the constraints on the availability and regularity of automated shuttles are perceived differently from a fixed route. The automated shuttles will be available through a spontaneous reservation, and waiting times can be anticipated by the users. For example, users of conventional on-demand transport platforms (e.g. Uber) have learnt through use that certain times are to be preferred and others to be avoided to use the services offered.

The second option for the user is to book their journey in advance, giving them more freedom to choose the time they want to be picked up. With this proposal, the transport operator encourages the user to anticipate his trips, which subsequently facilitates the organisation of services and the dispatching of vehicles according to the time slot of the day. The more customers anticipate their journeys, the more efficient the transport operators can be in managing production.

5.4.2 Energy Constraints and Battery Capacity

In addition, on-demand transport helps to increase the ratio of passengers to kilometres travelled, thus improving energy efficiency as the energy requirement on the batteries is reduced. While creating a fixed route, the battery capacity is an important factor for the dimensioning of the vehicle fleet. This is less important for a fleet of automated shuttles in demand-responsive transport because the services are composed of breaks during which the automated shuttles can recharge the batteries.

5.4.3 Facilitate the Relationship with the User

As on-demand transport services are mostly based on a digital solution that integrates a smartphone interface, the relationship with the user becomes easier. Important top-down information such as the waiting time before the next trip, the proposal to use another more efficient mode in a mobility-as-a-service (MaaS) environment, the monitoring of use and the valorisation of loyalty, and the listening of the customer can be managed via this interface. Thanks to this possibility, the transport operator minimises their investment in infrastructure by creating passenger information.

On the other hand, it is important to note that this vision of the service excludes users without a smartphone, which is notably the case for the oldest and most isolated population. It is therefore necessary to think about an accompanying service if automated shuttles are to fulfil a public interest mission and thus be a fully fledged mode of the public transport network. It is important to not exclude any population from a new service.

5.4.4 Pricing Issue

In a regular and fixed service, the question of pricing arises. Even if a validator were to be installed inside a shuttle, it would be difficult to automate the validation process at this stage of technological development. This leaves the possibility of carrying out control actions, in the same way as on the other vehicles of a public transport network. However, when considering the development of automated vehicles, we must consider the use cases that will minimise human intervention. In this context, the use of the on-demand transport service supported by a smartphone interface facilitates the act of payment of the service by the customer. It is easily imaginable that the customer can register their credit or debit card or their transport card in the application (depending on the fare choices of the PTO). In this case, only people who have a good knowledge of the current fare system will be able to make a reservation and use the service.

However, this is only part of the answer to the problem of fraud. It will also be necessary to investigate the control of access to the automated shuttles by people who take advantage of another person in a legal situation to get on board the automated shuttle.

The reasons given above are not intended to be exhaustive, but they do help to understand the interest in on-demand transport for automated vehicle technologies. It also allows us to observe that many questions arise with this type of service. This is why Keolis Lyon has committed to an evolution of the transport service initially proposed. The initial fixed line will have to increase its perimeter and offer a transport-on-demand service within a district that has been energised by the recent arrival of numerous centres generating daily passenger flows, with different motivations, and different travel times, not subject to the two peak full stops of the morning and evening.

6 Luxembourg

6.1 Pfaffenthal

Pfaffenthal is a small, urban living area located in Luxembourg City, the capital of Luxembourg. This urban area with around 1300 inhabitants is based in a valley between the historical centre of Luxembourg City and Kirchberg, the business district that is home to the European Investment Bank and the Court of Justice of the European Union.

Pfaffenthal is connected to the city centre via a public elevator and to Kirchberg via a funicular railway. Several bus connections are available in the surroundings of the elevator entrance at the city centre level. The funicular is part of a multimodal station that has been newly implemented in Pfaffenthal. Besides the funicular station, this multimodal station consists of a train station, a stop for several bus lines, as well as a bike sharing station. A tram connection and additional bus stops are situated on the Kirchberg side (of the topography). Over the course of 1 day, a variety of individuals traverse the Pfaffenthal Valley via a range of transportation methods. During the morning and evening rush hours, the majority of individuals commuting through Pfaffenthal do so by tram (Fig. 2.22).

Fig. 2.22

Route and automated minibuses stops

The specific characteristics of the Pfaffenthal Valley make it the ideal use case scenario for the demonstration of a first and last-mile mobility service. Prior to the beginning of the project, no interconnectivity between the various modes of transportation arriving in the distinct regions of Pfaffenthal existed. Furthermore, Pfaffenthal offers a very diverse traffic situation with all kinds of different road users. It is a showcase to see how an automated vehicle can be integrated into such a diverse environment.

6.1.1 Objectives

Until the beginning of the project, no transportation solution existed to overcome the distance between the residential area, the multimodal station, and the public elevator. The core objective is to fill this lack of transportation to connect the different means of transportation as well as the different areas of Luxembourg City with each other.

During the day, residents, employees, and a vast number of tourists are using the multimodal station in combination with the elevator to get to the different parts of Luxembourg City. The distance between the public elevator and the multimodal station is 500 m, and the distance between the residential area and the public elevator is 800 m. This corresponds to a 5–10-min walking distance. To connect the public elevator, the multimodal station, and the residential area with each other, the following route for the automated minibuses has been selected.

6.1.2 Deployment

The automated minibus shares the street with all kinds of different traffic users from cyclists and pedestrians to trucks, buses, and individual cars. Even with the speed limit of 30 km/h, the automated minibus encounters numerous overtaking manoeuvres, which cause harsh breakings. Complex traffic situations around the automated minibuses cause rough driving behaviour. The shuttles need to identify other traffic and not only detect it. The automated minibuses speed of max. 18 km/h is slowing down traffic in Pfaffenthal, especially in the morning peak hours when the traffic is very dense, and the drivers seem to be very nervous and hectic. This is also the reason for the change of operational hours in 2019. It was decided to keep the automated minibuses out of the morning peak hours to prevent them from slowing down traffic. There is no need for an on-demand system because the shuttle only serves three stations. A safety operator has to be on board all the time by law (at all sites).

6.1.3 Achievements and Key Success Factors

With more than 25,000 passengers and over 9000 km driven in around 18 months, Pfaffenthal is the most successful site regarding the deployment of automated minibuses. The presence of the tourist destination Pfaffenthal, along with the prime location between the two elevators, was the key contributor to the success. The shuttle service did not require any promotional campaigns as it enabled users to travel distances that would normally be traversed on foot, thus providing them with the opportunity to reduce their transit times and to experience new technology.

6.1.4 Future Development

In the near future the city of Luxembourg is willing to restart the service. The restart on the existing route and with enough time to get all authorisations is a good way to get back to the service. Afterwards, the extension or implementation of other options and routes can be conducted, if required.

The vision for the trial in Pfaffenthal is to deploy more routes within Luxembourg City to establish a network and an on-demand service of automated minibuses that are linked to each other, to the different parts of Luxembourg City, and to the public transport of the city. SLA can implement the on-demand option because Esch has shown that this service can be offered in Luxemburg. During the time of the AVENUE project, there were plans to expand the route. The objective of this route extension is to provide a transport service to:

• Hospice de Pfaffenthal (limit in parking space).

• Youth Hostel (limit in parking space).

• Servior, a retirement home (limit in parking space).

These three future partners do not have enough parking facilities and no nearby bus stops for passengers taking the Panorama Lift from the upper town. An emerging factor is providing accessibility to different users where transportation is scarce. The city of Luxembourg wanted to get a few months of experience with the original route before giving the authorisations for on-demand and door-to-door service. Due to COVID and the full stop of the service until the end of the project, the extension has never been realised but is still a possible option in the future.

6.2 Contern

Contern is a city located around 10 km southeast of Luxembourg City. An industrial zone with different companies has been implemented on its territory. A railway station and a stop for public buses are located on the border of the industrial zone. There are several other stops within the zone, but no direct connection exists from the train station to Campus Contern.

In this first phase, the automated minibus was operating between a real estate development company called “Campus Contern” with more than 300 employees and the train station.

6.2.1 Objectives

The aim was to dispatch individuals arriving via public transit to various firms in the industrial section and to offer a transport solution within this area. The trial in the industrial zone of Contern was chosen for its different environment compared to Pfaffenthal. Whereas Pfaffenthal has a busy inner-city traffic situation with various types of road users, the traffic in the industrial area of Contern is primarily composed of industrial vehicles, trucks, and cars, with fewer cyclists and pedestrians. The morning and afternoon hours in Pfaffenthal are marked by a considerable rise in individual car traffic because of people going to and coming from work. This phenomenon is far less accentuated in Contern (Fig. 2.23).

Fig. 2.23

Route Contern

The vast majority of the company’s employees was using their private car for their work commute as well as for transfers inside the zone. The shuttle is a good alternative because it closes the last-mile gap from the train station during the rush hour. Frequent traffic with heavy vehicles and increased commuter activity during peak hours is observed in the industrial area.

6.2.2 Deployment

The percentage of the automated mode with 50% is very low compared to other sites because it takes 30 min for the shuttle to get from Campus Contern to the train station and back. The train arrives every 30 min. To arrive in time, the safety operator switches into the manual mode and drives a little faster than in the automated mode. In Contern very few issues occurred during the operation that were caused by external factors. There is less traffic in the industrial zone of Contern compared to Pfaffenthal; thus the automated minibuses encountered less overtaking. However, there were more heavy vehicles, which made the overtaking more dangerous for the shuttle. The main issue encountered was vehicles parked illegally on the automated minibus’s path. From the view of the other road users, the automated shuttle is driving comparatively slow. This can result in less acceptance of the automated minibuses as a new kind of road user and on the other hand to use the shuttle itself, as long as it is driving slower than their own cars. There is no need for an on-demand system because the shuttle only serves two stations.

6.2.3 Achievements and Key Success Factors

There were over 850 passengers transported and 4000 km driven over 26 months. This is not as much as in Pfaffenthal, but there were a lot of breaks because of construction sites and COVID (including strict home office rules for a long time), and the people accepted the service and used it every time it was running. The length of the route and the slow speed of the shuttle prevented more people from using it.

As the automated minibus was operating exclusively for “Campus Contern”, the total costs were paid by this company. On this extended route, the automated minibus is passing several other companies and thus could connect more of them to the railway station.

The continuation of the contract depends on other companies to join the service and to lower the costs for the participants. It is very important to notice that the operation of the automated minibuses is linked to very high costs—as long as a safety operator is on board and not supervising several shuttles from a central office. A lot of companies, cities, authorities, etc. are very interested in such an innovative mobility solution, but they are not able to bear the respective costs.

Other than in Pfaffenthal, one does not simply walk into the shuttle. The people have to know that there is a service because it is only two times an hour at the stops. Campus Contern engaged in extensive internal communication, and on average there was at least one person per ride in the shuttle. With more companies joining, the amount of passengers can rise and provide a good alternative for taking the car.

6.2.4 Future Development

The industrial zone in Contern was seen as an appropriate use case for the establishment of an on-demand mobility system. The demand varies with the time of the day. During the peak hours in the morning and in the afternoon, the request for the automated minibuses is mainly done between the train station and the different companies. During non-peak hours, there is a shift in demand towards the interior of the industrial area between various businesses and locations of interest, such as restaurants.

It was hypothesised that the routes could be extended to the other side of the industrial area, as well as additional stops being added in front of companies that expressed a desire to use the service. Due to the stop of the service during COVID and the slow return of the employees from the home office, there were too few passengers to justify the (expensive) extension of the existing service. Nevertheless, other companies are still interested.

The vision for Contern is to establish a system of automated vehicles connected to the bus network and other shared transport modes like carsharing, establish the on-demand system, and implement the shuttles into the public transport information system. Like this, it is possible to synchronise the shuttle with the train in time, even in case of delays. Numerous shuttles could be available for the different parts of the industrial zone, which can be ordered on demand. This will facilitate the last mile of travel between public transportation stops and the workers’ final destinations.

6.3 Esch-Sur-Alzette

Esch-sur-Alzette with 35,000 inhabitants from more than 120 nationalities is the second most populated town in Luxembourg and is located on the border to France in the south of the country around 17 km away from the capital Luxembourg City. The introduction of a shuttle service to the longest shopping street in the country could facilitate the revitalisation of the area through improved mobility. The shuttle ends on one side near a bus stop, while at the other end, there is no public transport. In this way, the shuttle closes the gap between the public transport stop and the inner city.

6.3.1 Objectives

In September 2021 the service started in the main pedestrian/shopping street. In this part of the city, no public transport can be provided, but the mobility of the people can be optimised, especially the elderly, wheelchairs, strollers, or people with heavy shopping bags by providing the shuttle service. On rainy days the shuttle service can get people dry from one spot of the shopping street to another.

In September 2022, the planned on-demand service was launched, extending service hours from 11:00 am to 9:00 pm. The shops are closed after 6:00 pm, but the cafés and restaurants are still open. The presence of a shuttle during the evening hours in the more quiet street can provide a sense of security to pedestrians due to its illumination and motion.

The long-term vision is an automated-based, door-to-door service in Esch-sur-Alzette. The objective is to start first trials of an automated minibus with dynamic routing in a geographically defined area, without fixed bus lines or predefined timetables.

6.3.2 Deployment

The shuttle is driving on a 1-km section of the street “Rue de l’Alzette” with several stops at the crossings. It is driving at around 5 km/h as it should not be a disturbing element in the street but merge with the pedestrians. Only pedestrians and bikes are allowed in the street. Before 11:00 am and after 6:00 pm, delivery traffic is allowed, too. The on-demand service started on 12 September 2022, extended the service hours to 9:00 pm, and added four virtual stops on the same route (Fig. 2.24).

Fig. 2.24

Route Esch

In terms of the shuttles’ operating environment, Esch is very different to Pfaffenthal and Contern. The very dense and narrow shopping street is only used by pedestrians and sometimes by bikes and scooters. Some people want to test the “reflexes” of the shuttles’ systems, and others are not paying attention even if the automated minibus rings the bell. Occasionally, there were objects (e.g. chairs from restaurants or a fence of a long-term construction site) on the driving path of the automated minibus. Due to these circumstances, SLA was required to communicate the challenges for the service to restaurant owners or to the city of Esch. The testing and further development of the pedestrian mode resulted in less harsh braking. Furthermore, the implementation of the on-demand service resulted in longer service hours and a demand-orientated service for the users.

6.3.3 Achievements and Key Success Factors

With around 12,000 passengers in 12 months, the site in Esch is nearly as successful as Pfaffenthal.

In the summer of 2022, the mobility service had to be paused for 2 weeks due to technical difficulties. After the break, the passenger numbers were as high as before. This underlines that the people got used to the shuttle service, as they use it despite a longer operational break.

The street is wide enough for one vehicle, but with a lot of pedestrians, bicycles, maintenance cars, events, and construction sites, it can be difficult for the shuttle to drive all the time in automated mode. Intensive communication with the city of Esch is necessary to solve these problems, and it is working out very well.

The ODS started on 12 September 2022, extended the service hours to 9:00 pm, and added four virtual stops to the six existing ones. An app is required to book the shuttle for the desired time. The user cannot see the stops in the app but the service area. The user can choose the pickup and drop-off location freely within the area but is guided to one of the ten stops. These stops are very close to each other, so the user does not need to walk more than 1 min. Although it is a slight restriction, it allows the user to flexibly book the automated minibus.

Similar to the operation in Pfaffenthal, the people see the shuttle running and can use it to shorten their walking route. Since the service hours are the same as the shops’ opening hours, no additional advertisement is required.

6.3.4 Future Development

There was a long-term vision for an automated-based, door-to-door service in Esch-sur-Alzette, which should have covered the whole inner city. Due to COVID, there was a significant delay in the planned schedule. The normal shuttle service started one and a half years later than initially planned. Additionally, it was early obvious that for a useable service covering a whole city centre, much more shuttles are required. This requirement is going to be fulfilled within the ULTIMO project from 2023 on. Although this plan is desirable, the process of service implementation in Esch may have to be conducted with more steps.

The service should continue in 2023. It is also possible that the service hours will be extended. If people use the shuttle starting at 6:00 pm often and get used to it, it will show that the shuttle is part of their mobility and can maybe even drive until 10:00 pm—with or without the on-demand service. Another option is to add another street at the end of the existing route where a lot of restaurants are, enabling that the customers are closer to the shuttle service. In conclusion, the shuttle itself is not only an additional mobility option but is also a contributing factor in revitalising the city centre, making it safer, cleaner, and more liveable.

7 Lessons Learned

Automated vehicles for public transportation are seen as a promising solution to the growing urban mobility challenges. Despite significant investments in improving technology, the transition from a “technology offer” to a “service offer” remains unclear, and until today there has not been a real commercial deployment of automated vehicles in urban mobility. This is primarily due to the lack of experience on how to develop viable and sustainable business models, which, in turn, results from a lack of real large scale.

The several real-life deployments of the automated vehicles in different public transportation set-ups of the AVENUE project allowed us to identify the vast majority of issues and to imagine and implement solutions. In this chapter, we have described the AVENUE site deployments, and we have outlined the key elements that require further study and development. The key lessons learned are that, on one side, long-term planning needs to be established by public transportation operators (PTOs) and public transportation authorities (PTAs) over a decade, with the creation and implementation of new policies, infrastructures, and passenger services. On the other side, new open standards that support the interoperability of automated technologies must be developed and adopted. This will foster competition among technology players, including vehicle manufacturers, fleet management and orchestration providers, V2X technology developers, and more, thereby creating a sustainable ecosystem.

While the issues discussed in this chapter are significant, they are not the only challenges. Additional hurdles include retraining maintenance personnel from mechanical to IT and electronics expertise and addressing the transition of existing drivers to new roles, such as intervention teams or back-office operators. Furthermore, deploying automated vehicles with on-demand, door-to-door public transportation services may face opposition from other transportation actors, such as taxi services, who might view this innovation as a direct threat to their business. Additionally, the certification of automated drivers remains an open question since public transportation vehicles are currently certified for their mechanical capabilities, and human drivers are “certified” for their driving skills. Approaching these issues, along with any unforeseen challenges that arise during actual very large scale service deployment, will require a concentrated effort from PTOs, PTAs, legislators, and service users to develop workable solutions.

Ultimately, this is the objective of the European project ULTIMO, initiated in October 2022 based on the learnings of the AVENUE and other European projects. ULTIMO aims to lay the foundation for a large-scale deployment and develop the necessary standards, policies, and road maps to create a viable, user-oriented, and economically sustainable public transportation service using automated vehicles. We expect that large-scale deployments in Europe and around the world will start appearing from 2026 and on, with the parallel introduction to the market of new types of highly performant SAE L4 vehicles.