Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Online model learning in autonomous racing using Gaussian Processes



Alexander Liniger, Marius Schmitt

When implementing a controller in the experimental set-up one of the biggest challenges is to deal with the ever changing model. Where possible model changes may come from wear of the tires, ageing of the motor, dust on the track, different batteries with different state of charge, and other reasons. The changing model makes the controller tuning task a lengthy (sometimes frustrating) process. However, at the same time the model is constantly exited due to the aggressive driving in autonomous racing, which should allow to learn the model online.

The goal of this project is to investigate the use of Gaussian Processes (GP) to learn varying parts in the model. To achieve this task it is necessary to investigate tractable methods to incorporate new data, as well as finding the correct time scale at which the model needs to be updated. Additionally to make use of the online learned GP model, it is necessary to modify the existing optimization-based controllers to work with GP models.

Weitere Informationen


John Lygeros

Art der Arbeit:
Anzahl StudentInnen:
Status: done
Semester: FS17 or starting now