RoboRace



RoboRace
RoboRace
RoboRace
RoboRace
RoboRace

RoboRace - A Self-Driving Racecar



This project is about developing an autonomous system for a racecar to be able to self-drive on a given path within the track boundaries. The autonomous system consists aspects such as environment perception, vehicle state estimation, mapping, SLAM, trajectory planning and control strategy.

In the environment perception, a system will be trained to detect the cups as the track boundaries and obtain the position of the cups in relation of the camera using YOLOv3. The IMU sensor and Ground Speed sensor that are attached on the racecar will measure and integrate data over time in order to get the vehicle's position, speed and heading. Furthermore, a mapping will be implemented in order to translate the cups from the local camera frame to a persistent global map as well as to frame the transformation according to time using basic ROS tools. A SLAM algorithm will be included as well to correct the integration error in the vehicle state estimation. Besides, a path will be planned within the track boundaries for the racecar in the trajectory planning. In the control strategy, we have the track boundaries and vehicle state estimation as the inputs in order to compute a suitable control action to operate the racecar safely.


  • Jahr:
  • Studierende: Phillip Strobl, Dejan Pekez, Alexander Fahn, Sarah Giet Yee Choong, Guido Ivan Garcia Duva
  • Semester: Semester 5
  • Studiengang: Informatik
  • Supervision: Prof. Dr.-Ing. Thorsten Schöler, Hannah Lisa Walkiw, Sebastian Pröll
  • Tags: automation, RaceCar, ROS, Python, SLAM, Mapping, CNN, Vehicle State Estimation, Trajectory Planning, Control Strategy, Simulation, YOLOv3, Darknet