Traffic flow harmonization in expressway merging

Abstract

Steering a vehicle is a task increasingly challenging the driver in terms of mental resources. Reasons for this include the increasing volume of road traffic and a rising quantity of road signs, traffic lights, and other distractions at the roadside (such as billboards), to name a few. The application of Advanced Driver Assistance Systems, in particular if taking advantage of Ambient Intelligence (AmI) technology, can help to increase the perceptivity of a driver, leading as a direct consequence to more relaxed mental stress of the same. One situation where we see potential in the application of such a system are merging areas on the expressway where two or more varying traffic streams converge into a single one. In order to reduce cognitive liabilities (in this work expressed as panic or anger), drivers are exposed to while merging, we have developed two behavioral rules. The first (“increased range of perception”) enables drivers to change early upstream into a spare lane, allowing the merging traffic to join into mainline traffic at reduced conflicts, the second (“inter-car distance management” in the broader area of merging) provide drivers with recommendations of when and how to change lanes at the best. From a technical point of view, the “VibraSeat” a in-house developed car seat with integrated tactile actuators, is used for delivering information about perception range and inter-car distances to the driver in a way that does not stress his/her mental capabilities. To figure out possible improvements in its application in real traffic and at a meaningful scale, cellular automaton–based simulation of a specific section of Madrid expressway M30 was performed. Results from the data-driven simulation experiments on the true to scale model indicate that AmI technology has the potential to increase road throughput or average driving speed and furthermore to decrease the panic of drivers while merging into an upper (the main) lane.

Appendix: Related work in traffic merging optimization

Ran et al. [17] focuses on merging in automatic highway systems by application of different options of assisted merging, and by assuming that vehicles on the single-lane main road are operated fully automatic. To improve merging, they used the approach of matching vehicles in the stream on the ramp to gaps in the traffic stream on the main road so that vehicles can merge at the speed of vehicles on the main road. While not directly related to our simulation approach, some aspects they described have affected considerations in the development phase of our model, e.g., the concept that merging vehicles approaching the merging area can change into the main road with almost unaltered speed.

For the results reported in the next four projects, highway sections with two main lanes and a single merging lane were used for simulation.

Laval and Daganzo [22] showed a more real multi-lane hybrid model for traffic flow simulation. They provided evidence that the reduction in throughput is almost caused by the limited ability of lane changing cars to accelerate due to small gaps, leading to the typical “stop-and-go wave.” But they also found out that, if knowing when and how to queue best, the observed speed reduction and “stop-and-go wave” would have been avoided. In applying their model to a bottleneck situation they reached consensus that lane changes are the main cause of the drop in the discharge rate obtained. This is of particular interest as this situation is similar to the merge bottleneck investigated in our case [23]. Nevertheless, results cannot be taken too serious as the movement behavior of vehicles in their simulation is unrealistic (they used only a constant emergence of few vehicles (1,242 veh./h on the main lanes, 416 on the shoulder lane)). Wang [24] presented a combined car following and lane changing model also reproducing the typical “stop-and-go” phenomenon observable in reality. The drawback is, however, that the used a simplified highway model with only one lane on the main road for simulation/interpretation, making it almost impossible to derive qualitative conclusions from the developed model for the reality.

Hidas [14] implemented a rather complex lane changing algorithm for vehicles (“lane-change plan”) and tested its feasibility in a hypothetical road network. A macroscopic comparison of the stop rate per vehicle (as a measure for stop-and-go behavior) indicates similar results for simulation runs with activated lane-change plan compared to disabled lane-change model in uncongested traffic flow, while the stop rate increased steeply with deactivated lane-change plan under fully saturated conditions. The authors conclude that the explicit modeling of forced and cooperative lane change can eliminate the weaving and merging problems under congested conditions; unfortunately, they also used a artificial population of vehicles on a hypothetic road network, which does not allow to draw immediate conclusions for real traffic systems.

Akram et al. [25] studied traffic flow on individual lanes from videotaped entrance ramp junctions on different expressway sites. They found out that merging vehicles entering the traffic stream on the leftmost lane create some turbulence in the vicinity of the entrance ramp and that expressway vehicles approaching the merging area on the left lane move toward the right ² as long there is capacity in order to avoid this turbulence. Furthermore, they found only a weak relationship between accumulated vehicle flow rates on the expressway (v ₁₂) and the stream on the entrance ramp (v _r ).

Fujii et al. [26] reports on traffic flow simulation from real observations (traffic volume, average speed; 5 minute traffic counter update rate) on a narrow highway section, parametrized using the concept of coordination graph [27]. The comparison of simulation output with observations from reality shows high conformity for a coordinative behavior model, while simulation with disabled coordinative behavior shows significant differences in macro-indicators and micro-behavior in traffic flow. What we can learn from this work for our model is that coordinative behavior of car/driver pairs is a must to obtain simulation results close to reality.

Jin [28] studied continuous kinematic wave models of merging traffic flow and demonstrated its validity with numerical examples. Furthermore, the proposed analytical model is consistent with existing analytical models and provides, compared to discrete kinematic wave models, analytical insights of emerging vehicle dynamics at merging areas. However, the merging model used in that work cannot capture the impacts of lane changes, thus formation and dissipation of traffic queues at merging areas are analyzed without lane changing behavior only.

Papamichail and Papageorgiou [29] went one step ahead in the complexity of the merging setting having used dual-branch entrance ramps with both branches physically separated. The developed algorithms were optimized with respect to balancing the queue length on or waiting times in both branches; the total time spent (TTS) by all vehicles in the ramps is similar for each applied optimization strategy. From this point of view, the algorithmic developed is of less interest for our studies; nevertheless, as they have used a stochastic simulator as we did, the findings in terms of required number of replications requested due to the fact that different runs with different seeds produce different results are quite relevant for our simulation planning. They achieved good results with only ten repetitions per setting—which can be considered a sufficient quantity of repetitions to generate also reproducible results for our model(s).

Drivers’ individual preferences in operating a vehicle were considered a important factor in microscopic traffic simulation reported in [30]. Drivers have shown individual behavior, as for merging they may consider different gap length as safe or approaching to the merging point at different velocities. (The approach applied to the model is twofold, first a variation of gap size and second a shift in the driving speed (i.e., speed difference between own car and cars on the main lane), both considered acceptable for the certain driver in order to make the simulation results more consistent to real observations.) The two parameters were varied based on ten “driver types,” and simulation runs were then conducted using standard models leading to results unfortunately not reproducible in reality.