Autonomous driving using Xsens MTi-G-710 in Formula Student Racing
KA-RaceIng is a team of 80 students from the Karlsruhe Institute of Technologyparticipating in the international Formula Student competition. KA-RaceIng is the only team able to start in all three disciplines: combustion, electric and driverless. The competition consists of dynamic elements like acceleration and long distance races as well as static events like engineering design and even business plan or cost report. In addition to the classic challenges lightweight design, drivetrain, suspension, electronics and aerodynamics, the main focus in driverless racing is on sensor processing and planning algorithms. All three cars are completely self-designed, manufactured and tested.
The track is not limited by lines but with cones in 3 to 5 meters distance from each other. Going at speeds up to 60 km/h on very narrow and curvy tracks, a fusion of LIDAR and camera sensors is necessary to estimate the course of the track. Besides the perception of the environment, the inner sensors for tracking the status and location of the vehicle are very central. Especially disciplines with unknown tracks require a high accuracy of gyroscopes and accelerometers to create a map and localize the car in it while going at high speeds.
Xsens in a race car
The Xsens MTi-G-710 is one of the most important measurement devices in our race car, delivering among other measurements acceleration and angular velocity. Both are used to estimate the position of the vehicle. The gyro bias correction makes it possible to end up in a distance of less than 2 meters after one lap of several hundred meters. Combining the excellent raw data with a probabilistic approach for the mapping task (SLAM) we are able to close the track and perform the following rounds at much higher speed.
The acceleration measurements are also used to calculate the behavior of the car and the resulting wheel loads that are essential in racing. Knowing the load on each wheel the maximal torque that can be transferred is estimated; otherwise the wheel would just spin without actually accelerating the race car.
Download MTi Whitepaper
Xsens has a whitepaper available that explains the MTi System Architecture: “The Next Generation Xsens Motion Trackers for Industrial Applications”. It reveals in great detail what happens inside the orange box.
The whitepaper explains in short:
- MTi System Architecture
- Algorithm Engine
- MTi Portfolio