Title: Racing line optimisation for an advanced driver assistance system

Authors: Falk Salzmann; Sofiane Gadi; Ingmar Gundlach

Addresses: Department of Vehicle Mechatronics, University of Technology Dresden, Georg-Bähr-Street 1b, 01069 Dresden, Germany ' Department of Information Technology and Electrical Engineering, Automatic Control Laboratory, ETH Zürich, Tannenstrasse 1, 8092 Zurich, Switzerland ' Department of Control Systems and Mechatronics, Technical University of Darmstadt, Landgraf-Georg-Straße 4, 64283 Darmstadt, Germany

Abstract: This paper deals with an accurate, robust and efficient optimisation method for time optimal path planning on circular tracks. Starting with a general description of the problem, suitable method domains for time-optimal path planning are qualified. In terms of reproducibility and accuracy, we propose an algorithm combining a model- and a policy-based method which takes car, track and driving data gathered from connected cars into account. Hence, it can provide a consistently learning as well as a sufficient constant time-optimal racing line on worldwide race tracks for different driver assistance purposes. We evaluated the calculated racing lines with respect to heuristic criteria like curve cutting behaviour and by comparing them to ones driven by professional race drivers.

Keywords: advanced driver assistance; connected car; reinforcement learning; trajectory optimisation; vehicle dynamics.

DOI: 10.1504/IJVP.2022.119437

International Journal of Vehicle Performance, 2022 Vol.8 No.1, pp.46 - 73

Received: 26 Aug 2020
Accepted: 25 Jan 2021
Published online: 05 Dec 2021 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article