The chase for precious seconds is the key to race car driving. For maximum speed, race cars must be as light as possible. Until now, Colin Chapman, founder of the Lotus race car brand, was considered the undisputed lightweight champion. However, the students at OTH Regensburg (East Bavarian Technical University) are stepping up the game. For their latest race car design, they decided to omit the driver and thus save a significant amount of weight. Strictly speaking, the OTH Regensburg students did not actually come up with the idea, but they were part of the group that transformed the idea into reality.
In the Formula Student competition, teams from all over the world are facing off with their own monoposto race car designs. The teams are competing against each other in three different categories. Race cars with combustion engines are competing in the “Formula Student Combustion (FSC)” category, while electric vehicles are facing each other in the “Formula Student Electric (FSE)” category. In 2017, the race series introduced a third category: “Formula Student Driverless (FSD)”. This category was established for autonomous race cars using combustion engines or electric motors.
The OTH Regensburg Dynamics e.V. team has been competing in the race series since 2008. Next season, Dynamics e.V. will compete for the first time with a race car that can drive with a driver or autonomously. Accordingly, this allows the team to compete in both the Electric and Driverless categories. At the heart of the intelligent control system for autonomous operation is an AI-enabled (artificial intelligence) vehicle computer designed by Embedded Specialist Syslogic. As the designated hardware partner, Syslogic is providing the AI computer for the racing team.
Unlike most motorsport competitions, Formula Student is not solely about the best lap time. To secure a place on the winner podium, the teams also have to prove themselves off-track. In addition to the design of the race car, they are also judged by their business plan or their financial budgeting. But ultimately, Formula Student is also about fast lap times. In a total of four dynamic disciplines, the students demonstrate what their race cars are capable of. The highlight in the “Driverless” category is the “Track Drive” discipline, in which the race cars complete ten laps on a previously unknown course.
Currently, the Dynamics e.V. students are in the midst of testing. After initial tests in the laboratory and with a conventional car, the new technology is currently being tested in a race car. But how exactly does the autonomous race car work? Ferdinand Wohlstein, Team Manager Driverless, explains how. He says: “Simply put, we’re dealing with three sub-areas: Sensors, actuators, and software.”
The team uses LIDAR (light detection and ranging) technology to recognize the environment. Laser pulses are emitted and the reflected, scattered light is measured. This allows the environment to be captured with high precision in a 3D model. Two additional cameras are used for color perception of the differently colored pylons that outline the circuit. Their images are then processed by a self-trained algorithm called YOLO (You only look once). It recognizes the pylons and their color.
The camera and LIDAR data are processed almost instantaneously. It can only be accomplished with a high-performance, GPU-based embedded system. Ferdinand Wohlstein states: “With the Jetson AGX Xavier platform and the robust design, the Syslogic AI vehicle computer is particularly well suited for our race car.”
In addition to processing the sensor data, the vehicle computer also handles all communication of the cameras and the LIDAR via Ethernet, as well as the GPS via a USB connection. Both data sets, the distances measured with LIDAR and the colors measured with the cameras, are then merged into a single, common data set. This in turn facilitates route planning and makes it possible to detect the track ahead. A controller is used to formulate the commands for the actuators. It consists of an additional steering unit and the existing drivetrain. The two units communicate with the Syslogic AI vehicle computer via a CAN bus. The high-precision GPS and GNSS units installed in the vehicle computer also allows the race car’s position to be determined with pinpoint accuracy. Ferdinand Wohlstein summarizes: “Long story short, the Syslogic AI vehicle computer is the brain of our autonomous vehicle.” According to Wohlstein, it receives environmental data, processes it, and sends commands to the actuators.
The hardware for the driverless system is a Syslogic AI Vehicle Computer RML A3, which is based on the Jetson AGX Xavier™ module by NVIDIA®. With up to 32 TOPS AI computing power, the embedded computer evaluates sensor data from various sensors in parallel and practically in real time. Accordingly, it is suitable for object recognition, environmental perception, and autonomous control of the race car. Because the vehicle computer has been specifically developed and manufactured for vehicle use, it can easily withstand the high shock and vibration loads in race cars.
Raphael Binder, CEO of Syslogic, says: “We’re keeping our fingers crossed for Dynamic e.V. for the remaining test phase and especially for the upcoming season.” Binder adds that he was excited about the fact that Syslogic was able to support an innovative team of young students.
We will provide periodic updates on the project status on our blog and social media channels.