5G based self-driving car demonstrations in collaboration of the Academic and Industry sectors in Hungary

APZ – BME– Ericsson Hungary – iMAR Navigation – Hungarian Telekom – T-Systems Hungary

ABP_TSM_800x534

The official opening ceremony of the Zalaegerszeg ZalaZone Autonomous Vehicle Proving Ground took place on May 20, 2019. While the investment continues, the test track will be 100% complete by the end of 2020, but from that day on, the finished parts and constructions of the track are available to tenants – vehicle and component developers, testers. The speciality of the track is that it is not only suitable for performing conventional vehicle tests, but also for receiving special vehicles equipped with autonomous driving functions. To this end, in addition to the previously completed LTE advanced (4G +) coverage, also the standard 5G test mobile station was completed for the ceremony, providing communications background for the 5G-based self-guided demos accomplished by the parties involved:

  • Automotive Proving Ground Ltd. (APZ)
  • Budapest University of Technology and Economics (BME)
  • Ericsson Hungary Ltd.
  • iMAR Navigation GmbH.
  • Magyar Telekom Plc.
  • T-Systems Hungary Ltd.

During the demonstration, the APZ operated test track in Zalaegerszeg, Hungary, was the first in the world to conduct real traffic situation tests with self-driving cars using an actual standard 5G radio data connection provided by standard 5G devices.

AB_TSM_313_600x400

No doubt that some of the equipment used are yet so called pre-commercial devices, and as the 5G frequency tender was still under preparation in Hungary at that time, the 3.5 GHz frequency used was still available only in test mode.
The test vehicles themselves are capable of autonomous driving up to SAE level 4, the traffic simulation vehicles operated also in fully automated mode (SAE level 5) to perform a predefined traffic scenario.

The demo consisted of two parts.

The first part involved two cars provided by iMAR, powered by their own developed positioning and self-guiding technology.
IMAR's ultra-precise positioning technology has its roots in the land navigation, marine, aerospace, military and other special areas. Their usage in the industrial-level automotive testing applications gives the users access to easy-to-use systems with highest reliability and performance at economical effort.
The main benefit of this solution is that the vehicle control constantly accesses the position data of the vehicles in real time, in cm level accuracy and can make decisions based on these information. In practice, this means that the car's position, obtained by iMAR’s iTraceRT-MVT INS/GNSS solution, is being updated internally 500 times per second, placed on a high-resolution (HD) map in real-time and stored internally for post-view or post-processing, so it’s always known in cm accuracy with less than 1.5 ms latency, at which point exactly the vehicle is.
This makes it possible to respond to traffic situations at high speed, even when traveling on the highway, faster than your own reaction time, even with vehicles in close proximity.Of course, to achieve this objective it is also necessary to distribute these highly updated extremely precise position data with sufficient speed (i.e. with extreme low latency) between all the vehicles participating in the traffic scenario, and if necessary, between a central control unit and the vehicles, too.
This low latency – and at the same time appropriate bandwidth and highly reliable, stable data connectivity – arrived with 5G in wireless technology.
The latency of 5G may be reduced to 1 ms by standard. During the demo, the actual transmission delay reached was 7 ms.
This means that position data with update rates of up to 100 times per second are shared between the cars involved in the traffic situation as they occur. All this is not achieved with direct V2V (Vehicle to Vehicle) technology, WiFi-like solutions, but with cellular technology, at an officially protected, safe frequency.
So where – 5G – mobile coverage is provided, these, or even better conditions will be available in the future, as demonstrated by the real-world demos.

During the demonstration, iMAR vehicles - as there was only 400 m of straight track available for the demo - traveled at a maximum speed of 60 km/h after the start, followed each other at short distances, then have overtaken each other; and finally stopped at the end of the lane – on a programmed route, in fully autonomous mode; the following car, the VUT (Vehicle Under Test) was continuously monitoring and reacting to the movement of the leading car (TSV, Traffic Simulation Vehicle).

Main technological components used in the frame of the demonstration:

AB_TSM_175_600x400 AB_TSM_249_600x400

The test cars presented real-time control situations: demonstrating the functionalities of iMAR real-time trajectory and scenario control, using as an input the scenario description based on OpenScenario, while meeting the requirements of ISO 22133-1 object communication, which is expected to be released in the near future.

The vehicles themselves are commercially available mid-range passenger cars (KIA Niro as TSV and VW Golf VII as VUT), the TSV retrofitted with self-driving technologies and both equipped with INS/GNSS reference systems of type iTraceRT-MVT-510.

However, the modified TSV can be used in common road traffic with conventional driving and in self-driving mode in suitable areas. This is made possible by simply disabling the self-driving function of the car at any time: the devices that control the vehicle in autonomous mode are connected to the CAN-bus to control the steering, brakes and throttle – which are electronically controlled by default anyway. By disconnecting the retrofit control unit from the CAN-bus, the vehicle can be driven in a conventional manner, as there are no "hard" modifications in the vehicle's own factory control equipment, and can be used as an ordinary car anywhere in the traffic.
This allows test vehicles to arrive at the test sites "on their own feet" where the self-driving support systems can be enabled.

The test vehicles are equipped with the already mentioned positioning devices. They use simultaneously several satellite positioning systems on several frequencies (GPS, Galileo, GLONASS, BeiDou), improved by correction data from a GNSS ground reference point network (iREF), fused inside iMAR’s complex solution with the data of an inertial measurement system, consisting of three precise gyroscopes and accelerometers. So up to 20 obtained satellite positions or raw data sets per second are complemented by 500 times per second with inertial data, ensuring not only extreme accuracy but also error tolerance (in the case of possible satellite positioning outage).

The exact position data is being placed on a HD resolution base map, using a public format like OpenStreetMap or a specifically surveyed map.

Position data (and any other vehicle data if applicable) are shared among vehicles on the mobile radio network. The demo included Ericsson’s 5G modem device, which provides GB bandwidth for the connected IP network, and is a small, compact device that could be easily mounted into the cars.

On the test track, Magyar Telekom's telecommunication tower was equipped with Ericsson's 5G antenna, and Ericsson's mobile-core network solution was also installed onsite, which was connected to their headquarters in Aachen. This setup ensured the experimental 5G network.
The management platform for control functions during the demo was ensured by means of edge computing technology: devices that are part of the mobile network, installed as part of or near cells, provide the computing capacity, running environment required for network-connected systems; also enable one of the key features of 5G, i.e. extremely low latency.
T-Systems provided uplink, broadband optical connectivity and Internet exit, as well as physical infrastructure. Telekom's full 4G+ coverage – as part of the normal commercial network – was continuously available as a back-up.
The implementation of the 5G network was the result of a cooperation of Ericsson, Telekom and T-Systems engineers.

NMHH (National Media and Info-communications Authority in Hungary) provided the test frequency in the 3.5 GHz range.

In the second autonomous driving test combination, the solutions of another domestic automotive workshop were presented, partly utilizing the same technologies and tools used in the previous demonstration, but combining and further developing them: accurate position data were supplemented by sensors monitoring the car's surroundings, and even information fed from a virtual space also influenced the vehicle’s behavior.

In the course of this, an autonomous car developed by the BME presented a unique test case supplemented with a so-called “Scenario in the loop” solution, which is a special method for testing self-driving vehicles.

In this case, the demonstration took place with one test car, but pedestrians and other vehicles also appeared as traffic barriers.While the vehicle is autonomously operating (driving), the car is moving the same way in a simulation, in virtual space, but its environment may differ from the real test track. In the virtual space, objects (pedestrians, vehicles) have the same impact on the vehicle's operation as those perceived by their own sensors, injecting them into the vehicle control system bypassing the sensor signals. In addition to virtual objects, real dummies like test dummy-pedestrians and vehicles can also be controlled by the SW system, which the car can detect and react to with the help of its own sensors. Communication between the vehicle, the simulation environment and the controlled test dummy is unique in the world through 5G, providing real-time responses.

Two demos were performed with the test vehicle. The first was a so-called “Valet Parking” feature: the car automatically parked out for a phone call and drove to the caller. Meanwhile, during this scenario, a virtual and then a controlled dummy-pedestrian crossed its path, which were detected and responded to: the car stopped in front of them and continued its journey only after the obstacle had disappeared.
The second scenario was the “Traffic Jam Pilot”. Here, the car follows the lanes sensed with its cameras, steering in self-driving mode and adjusting its speed to the vehicle ahead. During the presentation, it first followed a virtual vehicle and then a real car.
The test car was fully autonomous driven in these two cases as well.

Detailed scenario of the demonstration:
The vehicle traveled through a section of the smart city zone of the proving ground, on a programmed route. The "Valet Parking" function was introduced first, followed by the "Traffic Jam Pilot" function.

  1. Parking,
  2. Parking out,
  3. Stopping in front of the virtual pedestrian,
  4. Then stopping in front of the physical (dummy) pedestrian,
  5. After turning off, stopping behind the virtual vehicle and then overtaking it,
  6. Turning back, following the real car, then stopping.

Main technical components of the BME demo:

  • Smart Fortwo passenger car: the car has undergone unique modifications – BME development – steering, gas, brakes, etc., controlled by a special computer in the car,
  • „Scenario in the loop” virtual simulation environment, BME development with the cooperation of APZ,
  • On-board support equipment in autonomous mode: iMAR positioning system (basic version), 1 pc camera, 2 pc LIDAR and 1 pc radar equipment.
  • 1 electronically controlled dummy pedestrian, APZ development with BME
  • 1 ordinary car to represent traffic situation,
  • Ericsson 5G modems (same as in the previous demo),
  • Also the 5G test network was provided by Magyar Telekom Group and Ericsson.

Importance of the demonstration
The demo takes place in real and virtual space at the same time, and the vehicle can respond to both virtual and physical, real-world obstacles and traffic situations. With this solution, the development and testing of self-guiding systems can be faster and more cost-effective, since something is tested in the real environment only after the individual functions and systems are already working properly in the simulation.

The role of the 5G network in the demo
The vehicle travels autonomously, responds to situations based on its own sensors and information from the simulation space. In this case, 5G ensures a stable, real-time connection between the car and the simulation environment, which means that it happens exactly the same as it happens in the real-world and the virtual space, the car reacts to obstacles appearing in virtual space and also in the physical space real-time. The dummy is also controlled via the 5G network.

All of the demos were live at the opening ceremony at the T-Systems stand, and were broadcasted with external cameras (mounted on drones and vehicles) and inside the vehicles onto a giant display, along with the virtual space in the simulation. As a result, it was noticeable that the backup drivers sitting in the cars did not intervene during the maneuvers.

In the complex demonstration of extensive collaboration between academic and industry stakeholders it was the first time in the world to see how the benefits of a low-latency 5G mobile network can be utilized to monitor and control self-driving cars.
Magyar Telekom was the first in the country to create a real 5G test environment in the Zalaegerszeg area. Testing the 4G+ (partially 5G-ready) station built in partnership with Ericsson on the proving ground has been in progress since spring 2018, and another milestone of this process was the launch of the standard 5G test facility in May 2019, as well as the joint demonstrations with the partners.

Share