Test bench for self-driving software

Arianne : The electric race car built by the students of the EPFL Racing Team, over the 2022-2023 Season.
Group picture of the team at the 2023 Formula Student Germany competition in Hockenheim
Group photo of all the teams competing in the 2023 Formula Student Germany competition at the Hockenheim ring.
Myself on the left, working on the steering column assembly, in the pits of the Formula Student Switzerland competition.
Modified RC car to serve as test platform for the autonomous driving algorithms.
An early version of the RC car on the test track for testing vision algorithms.

Project information

  • Location Swiss Federal Institute of Technology (EPFL) - EPFL Racing Team
  • SkillsCAD, mechanical machining, 3D printing, electronics prototyping, team work.
  • Project date Jul 2022 - Feb 2024

Project details

Context

The EPFL Racing Team is an association of around 100 students of EPFL, with the goal of building an electric race car over the course of one academic year, and then compete against other universities during international competitions across Europe in the summer. These competitions include disciplines both with and without a pilot and as a relatively new team, the goal for the 2022-23 season was to implement, for the first time, autonomous driving capability.

As part of the "driverless hardware" division, my role has been two fold:

  • To come up with a test bench for the vision and control algorithms for the autonomous driving.
  • To structurally integrate the required sensor package on the full scale race car, while maintaining compliance with Formula Student competition rules.

Results

The test bench was implemented by fitting a RC car with the same peripherals that would later be used on the full scale race car. That is a Nvidia Jetson onboard computer along with a camera, LiDAR and INS/RTK unit (see picture above). This system was then able to control the RC car autonomously, through an arduino mega interface, along with a CAN bus converter.

As per the fitting of sensors on the full scale car, the computer and INS were fitted below the driver seat, while the camera and lidar sit next to eachother, attached to the main hoop structure above the driver seat.

Some key results are :

  • A working RC car enabling the software team to directly translate algorithms for risk-free testing.
  • Successful integration of the sensor package to the full-scale race car, which passed competition scrutineering on the first shot.
  • A first experience working in a big team of over 100 students, which is unlike anything else in academia but rather more industry like.
  • The opportunity to travel and take part in international competitions, to meet other teams from around the world, to discover their designs, and of course race them!