Hermes: Hemera Returning Messenger

A common issue for long-duration balloon flights in the polar area is high bit rate data transferring. Just a few hours after launch balloons are nor reachable with direct radio link, and often satellite links are not fast enough to allow the necessary transfer rate or, simply, too expensive. For this reason, stratospheric balloon borne experiments carry out on-board data recording. Data recorded need to be recovered after termination, which is, sometimes, a slow, difficult and expensive task. Not always it is easy or possible to reach the landing site, especially during the polar winter. The aim of the project is to provide an autonomous glider capable of physically carrying the data from the stratospheric platform to a recovery point on the ground. This can also transport physical objects (like air samples) collected at float or along the flight. We estimate that an electrical motorglider released in the stratosphere can fly for several hundreds of kilometers. The glider is installed on the balloon payload through a remotely controlled release system, and connected with the main computer to receive data and the geographic coordinates of the recovery point. The glider trajectory can be monitored with Iridium SBD (Short Burst Data), and simple commands can be issued as well as using Iridium.


Introduction
We describe an affordable and easy to use vector able to safely transport scientific data and samples from the stratosphere to the ground, a desirable tool for several scientific measurements. Such a system was originally thought for physically recovering data media from the Olimpo [1] experiment, but it can be used as well for i.e. sampling the volcanic ash from ground to the stratosphere or to make better and more complete sounding balloons because you are not forced to use expandable payloads with limited capacity, since the payload can be recovered and reused.
The HERMES payload is a project funded and coordinated by Agenzia Spaziale Italiana (ASI), and whose scientific coordinator is Istituto Nazionale di Geofisica e Vulcanologia (INGV).
The launch of HERMES is part of the European project HEMERA [2, 3], born with the aim of improving the scientific activities in the stratosphere by offering to researchers an easy access to the space in individual or shared stratospheric payloads.
As part of this project, two calls for proposals have been organized to select eligible scientific experiments. Such projects, including HERMES, were offered a free balloon flight.
In the case of HERMES, the launch campaign will take place at the Swedish space base of Esrange (Kiruna), and will be carried out by the Swedish Space Corporation (SSC).
With the HERMES project, a stratospheric platform equipped with a glider was designed and developed. The glider, once released, is capable of transporting a copy of the data of the scientific experiment hosted by the platform to a pre-established recovery point on the ground. This is achieved thanks to the presence on board the glider of a Solid State Drive (SSD) memory, connected to the payload flight computer.
This would allow the scientific data of the experiment on board the balloon to be available within a few hours and in an accessible place. This is of prime importance for scientific experiments on Long Duration Balloons (LDB) that take place on polar flights [4][5][6][7][8][9]. In fact, recovering the 1 3 payload in these cases is rather difficult and expensive (there are special recovery teams dedicated to such activities), but it is the only way to recover the scientific experiment data set. Furthermore, during the landing of the payload, damage to the same may occur, compromising the outcome of the scientific experiment.

Project History and Evolution
The project started from the ABACHOS (Automatic BACk HOme System) [https:// strat hosph ereff ect. com/ ABACH OS] platform modified to meet the HERMES needs. ABA-CHOS is launched using inexpensive latex balloons and is recovered at launch point. The use of affordable latex balloons helped experiment with this design with more than 15 flights. The analysis of an ABACHOS flight appears further in the article. The original platform was modified by adding some electronics (mass storage, communication, umbilical connector, battery heater) and a tail motor to increase the flight length. The glider is no longer released by the latex balloon explosion but from a command issued by a ground station and is tied to a releasing platform attached mechanically and electrically to the main gondola. The sketch of the original releasing platform appears in Fig 3. Here, the glider is held in place using a couple of brackets on the wings and the releasing is assisted by a pushing arm who also helps detaching the umbilical magnetic connector.
Some modifications had to be done to meet the SSC requirements, linked to three considerations: 1) The release of a load can change the mass center of the gondola and be a source of disturbance. While the attitude variation can be tolerated on a large LDB (Long Duration Balloon), where the weight of the glider is negligible compared to that of the gondola, it is no longer so in a small experimental flight. This problem cannot be overcome by releasing the glider at the end of the experiment, because the Swedish rules for testing UAVs (Unmanned Aerial Vehicle) limit the distance from the launch point, which would be exceeded after a few hours at altitude. 2) The balloon system telemetry (with transponder included) is service telemetry of SSC which the payload cannot use (the only communication facility provided to the payload was a relay contact to be used as emergency release command). Therefore, HERMES had to provide its own auxiliary telemetry in line of view. The generous use of equipment capable of generating RF (Radio Frequency) disturbances (two Iridium SBD modems and two 1 GHz line-of-sight telemetry) to the balloon's natural telemetry would require specific integration and a compatibility test for the experiments hosted on the same gondola.
3) The mechanics of the releasing system had to be tested with an acceleration of 10 g, to allow (in case of failure in releasing) the glider attachment to resist the acceleration induced by the parachute opening.
Because of the first two points the system cannot share the gondola with other experiments and needs to be completely autonomous, providing its own power and communication. A sketch of this configuration appears in Fig. 1, where the absence of a common gondola has led to revisit the project, designing a self-consistent mechanics and modular electronics, which allows to easily compose the characteristics necessary for use on any gondola. All the modules that make up the system are autonomous and can be easily integrated into any stratospheric experiments.

System Description
The system is composed of: payload (hosting HERMES electronics and the releasing system), glider and GROUND STATION. In its nominal configuration HERMES expects the use of solar panels, as well as an external power source from the main gondola.
In the case of short-term experimental flights (a few hours), the system can operate exclusively by the on-board battery. Figure 1 shows the overall view of the payload without the solar panels (short-term experimental configuration).
The main modules that make up the payload handle battery power and communications. Also present is the Flight Computer Simulator (FCS) and the glider release system (RELEASER). All modules communicate over CAN (Controller Area Network) bus.
The GROUND STATION can decide the flight termination through a command via a satellite channel, after uploading the coordinates of the landing point to the autopilot. Figure 2 shows the block diagram of the payload. It is divided in two sections, the RELEASER who hosts the battery charger and the parts to carry out the release, and the gondola frame which contains all the facilities we expect to have in a host gondola. The computer simulator, logically part of the gondola frame, has been installed in the RELEASER to facilitate connections.

The Launch Platform
The BATTERY PACK module manages the battery that powers the system. The battery, consisting of seven pure lead elements, was chosen for the extended temperature range. A microprocessor manages the battery correctly by charging it from the panels (using the MPPT algorithm) or from some other dc source, taking into account the temperature. The module powers the system through the CAN connector and supplies a specialized power supply to the glider's battery charger which is operated, before release, to switch the glider's batteries from storage mode to full charge. The BATTERY PACK communicates all the values concerning the power supplies through the CAN bus. Using two switches, the battery can be separated for transport and the system can be set up for battery operation only, ignoring external power sources.
The COM and CONTROL module is responsible for managing satellite and line-of-sight communications. A reduced version of this module is designed to be housed on board the glider to allow commands to be sent over the satellite network.
The IMU (Inertial Measurement Unit) block contains the inertial sensors (three-axis accelerometer, three-axis gyroscope), a three-axis magnetic compass, a barometric sensor and a GPS (Global Positioning System) receiver. The module, self-consistent for general use, allows to know the payload's attitude.
The MONITOR and SD RECORDER module makes a local acquisition of the log file of the CAN traffic and provides an interface RS232-CAN helpful for debugging.
The GLIDER CHARGER and JUNCTION BOX module manages the connection with the glider's umbilical connector (connector J21 in Fig. 2), powers the glider's SSD, recharges the glider's batteries and maintains its temperature by means of a heater.
The FCS produces synthetic data with which to fill the SSD, simulating the presence of a flight computer. In fact, HERMES was designed to transport the data produced by a balloon borne experiment to the ground, and the FCS, since there is no real data, produces test data, which loads through the umbilical connector on the SSD present on the glider. These data are the video taken by a camera, active from the moment of launch, and the house keeping data acquired by all the modules and transferred through the CAN bus.

The RELEASER
The RELEASER module, housed under the plate that holds the glider, controls the servo motors that release the glider. The servomotors are controlled through a position signal (PWM, Pulse Width Modulation) and through the power supply that can be supplied individually. The temperature of each servo and the overall current supplied are checked.
As mentioned, three servomotors are used, of which two free the wings and one activates a thrust mustache which helps release. Currently, the umbilical connection of the glider to the platform is obtained with a magnetic connector. Figure 3 shows the sequence of the glider release. The presence of the aircraft on the platform is revealed through a proximity sensor installed near the rear engine. A manual switch allows you to operate the servomotors for installing the aircraft on the release platform.
The releasing command can take place: (a) through a command via the CAN bus (b) through a contact provided by logistics telemetry, to be activated in the event of a defect in communications. (c) through a manual switch when installing the glider on the platform.
Compared to the original design just described, the mechanics of the separation system has been modified to comply with the request of the Swedish authorities which requires that the system resist an acceleration of 10 g.
The original system was supported by two motorized brackets that operated on the wings. Now, a safety screw (Fig. 4) has been added to this system which secures the fuselage to the RELEASER and which is unscrewed and extracted when the aircraft is released. The brackets that support the wings have been maintained to keep the glider in position but they no longer have a support function. The servo used to move the pushing mustache is now used only to help detach the umbilical connector. Figure 5 shows the glider block diagram. J21 is the magnetic connector that connects the glider to the payload through the umbilical cable.

The Glider Electronics
Through this connector the payload supplies the heater for the battery of the glider, keeps the battery charged, transfers the backup data from the FCS to the SSD, and allows communications with the glider through the CAN bus.
The possibility is foreseen that the glider can host a COM and CONTROL module (with IRIDIUM SBD), connected via CAN with the autopilot, of which it reads the positioning system to allow you to track the glider when you are not in line of sight.
It can load a new landing point on the autopilot and turn the engine on and off.

Final Considerations
The actual version of HERMES has been designed for summer flights. Under these conditions, the main issue to consider is the poor heat dissipation provided by the atmosphere (5 mb at the float altitude) and the internal pressure of  A much more critical scenario is a flight during the winter in the polar regions, where the temperature reached may be of the order of -80 °C [10]. These extreme conditions will put a strain especially on the mechanical parts and electronics.
Note that HERMES payload modules are not pressurized, special attention should be paid to heating the batteries and electronics.
Concluding, HERMES is built in the name of modularity. This permits to: -Make modules usable separately in other balloon projects; -Make debugging easy; -Make the project updatable: if a module is technically obsolete it can be replaced with an updated module that respects the functionality, with a minimum redesign of both HW and SW; -Easily add new features. In fact, all modules communicate on the bus and use the same connectors so you can add functionality simply by adding a module.

The Glider
The glider has been designed and made on purpose. It is built of foam reinforced with carbon fiber tubes. All parts are obtained by machining a foam block with a numerically controlled hot wire cutting machine (Fig. 6). In Fig. 7, there is the mechanical drawing, while in Fig. 8, there are some pictures of the glider during the machining process.
The glider is equipped with a commercial autopilot for navigation (Pixhawk). The system combines advanced sensors, including an accelerometer, gyroscope, magnetometer, barometer, and GPS, a powerful microcontroller, and a range of connectivity options. The open-source nature of this device allows the users to have access to the source code, and this enables us to modify and customize the code to meet specific needs.
In Fig. xx, it is visible how the glider, during an experimental flight, correctly gets into flight attitude at an altitude of 22,000 meters.
Normally, the autopilot controls the aircraft through the left and right ailerons, and receives commands from the radio telemetry (connection with the glider control station) and from the remote radio (remote control for landing in line of sight).
Usually in planning a mission, the Home Position is set as the location where the aircraft is armed. This is the mode used in the experimental flights carried out so far.
Alternatively, there is the possibility to use waypoints to set up an alternative return point. The control station of the glider (telemetry in line of sight) permits entering new waypoints, setting geographic coordinates (latitude, longitude)  and altitude. This is the mode we are developing for the next payload configuration, where the glider will also have an Iridium SBD modem on board.
A balloon flight is unpredictable in the long term. The flight has to be continuously monitored from a ground team who decides if to release the glider. The glider will be released if (1) there is the necessity to recover the data (2) there is the possibility to land to a safely reachable place. In this case, the landing point is uploaded to the glider, and the releasing command is issued.
Once released, the glider is able to autonomously reach the landing point. In any case, in the final phase of landing, when the glider is visible to the operator, there is the possibility of manually taking control of the aircraft (RC commands).
From the point of view of logistics, when the command release has been executed, the flight support team travels to the prearranged location, now just expect to see the glider in line of sight and then manually take control of it with radio command to land it. To facilitate its contact with the ground, a carpet of plastic material (about ten meters long) is spread where it will land.

Experimental Flights
As previously mentioned, the tests carried out so far have been performed with latex balloons. Figure 9 shows the launch phase and the re-entry phase of an experimental flight conducted with a latex balloon. Experimental results show the glider, after releasing, gets in attitude between 10 and 20 km height. The length of the flight depends on the attitude height, the efficiency and, if the tail propeller is used, on the efficiency of the thruster and on the battery capacity. A 100 Wh battery can increase the flight length of roughly 50 km. If in attitude at 20 km the glider can fly for more than 100 km in calm air. This distance can be  The result of the flight test with the tail propeller conducted in 2019 in Benevento, suggests that the presence of the engine also helps the aircraft to get into flight attitude (22 km) 2 km higher than the best flight without engine. Figure 10 shows the plot of the vertical speed in the experimental flight of the glider equipped with the engine. Figure 12 shows a 3D plot of the flight path and Fig. 11 shows the plot of the distance from the base and of the altitude of the flight. Figure 13 shows the glider efficiency (horizontal speed/vertical speed) vs. height, calculated from the flight test data. The graph shows an apparent (the engine was on) efficiency greater than 20 at 13 km: unfortunately a malfunction occurred during this promising flight and it was not possible to examine the prototype to understand the cause of the malfunction.
The presence of the RC pilot to facilitate the landing is not mandatory although safer. The glider can land nicely on a flat surface (like the snow or ice we expect to have in polar areas). However, the memory stick containing data can survive even in case of a bad landing.
In order to be able to compare the flight described above with a flight performed without an engine, an experimental flight performed on is now analyzed. First of all it is shown in Fig. 14, the altitude-time diagram of the flight, which shows how the glider gets in attitude at 20,000 meters.
The same graph is shown on a three-axis Cartesian system (Fig. 15), where on the (x,y) plane, we have the geographical coordinates, while on the z axis, we have the height. With this representation, all phases of flight from launch to landing are clearly visible. Once close to the ground, the glider spirals until manual control is taken.
On the figures below, there is the ground projection of the descending spiral (Fig. 16) and the perspective view of the spiral with the starting point of the manual piloting highlighted (Fig. 17).
To determine the acceleration to which the glider is subjected, the log data relating to the inertial platform (IMU) are now analyzed.
The diagram of Fig. 18 shows a plausible baseline at 9.8 m/ s 2 with occasional fluctuations below 1/2 g on takeoff, a bump at 10 g on release and some spikes to 2 g on landing.

Future Developments
The use of autonomous stratospheric gliders is a promising method for collecting and transporting scientific samples and data. The results of experimental flights suggest several improvements that will make this technique affordable and usable. The tail propeller showed to be a good way to increase the flight length, and a higher efficiency means a bigger choice of landing points and a longer flight time to collect samples during the descent. Retractable wings (to be deployed after entering the troposphere) will help in this task as well as a better energy management and a better propulsion efficiency. High efficiency space-grade solar panels may be hosted on the wing surface without significantly affecting the weight and may increase the time of flight and (more important) keep communications alive allowing tracking the glider even in case of a delayed recover.
The experimental flights showed the glider sets itself in flight attitude at various altitudes, making the flight length unpredictable. A better autopilot algorithm or a glider design should make this height known in advance and therefore the flight length more predictable. The autopilot actually relies    Total average power consumption: 10W. The CAN I/F is the one that allows all the modules to communicate with each other and with the glider.
The USB I/F manages the data transfer between the FCS and the SSD on board the glider before its release.
Speaking of power interfaces, the Battery Pack and the Glider Charger and Junction Box modules have on-board step down converters that generate stabilized 9V and 5V. These supply voltages are distributed to all modules and to the glider.

Connections Between Modules
All system modules communicate with each other via CAN bus. Only 2 types of connectors (except glider umbilical J21 and RF connectors J6, J7, J8) are used for communications: a 4-pin connector and a 10-pin connector. The connectors can have different purposes but the connections are arranged so that inserting a connector incorrectly does not cause an electrical fault.
The 10-pin connectors are used for: See Tables 1, 2, 3. The 4-pin connectors connect all the modules by carrying the USB or CAN signal and 9 V voltage. To avoid conflicts in the event of incorrect insertion, the USB connectors (on the modules) are not connected to the power supply.
J21 is the magnetic connector that connects the glider to the payload through the umbilical cable. Through it the payload supplies power (5 V) and data (USB interface) to the SSD memory on board the glider.
In addition, this cable supplies the glider battery maintenance voltage, the power to the heater (to keep in temperature the glider battery) and CAN signals. The latter allows landing coordinates to be loaded onto the autopilot before releasing the glider.
Connector J13 refers to the cable carrying the release command coming from the host telemetry. This is a Fig. 18 The Z-axis acceleration plot Therm -CAN Therm 8 Gnd th Gnd Gnd th 9 Gnd Gnd Gnd 10 Gnd Gnd Gnd Funding Open access funding provided by Istituto Nazionale di Geofisica e Vulcanologia within the CRUI-CARE Agreement.

Conflict of Interest
On behalf of all authors, the corresponding author states that there is no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.