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.


Fast Stochastic Model Predictive Control for Autonomous Racing


P. Aeschbach

Master Thesis, HS15 (10392)

In this thesis, we develop robust MPC controllers for miniature race cars. Designing such controllers is often a challenging task, since (i) the car models are not entirely accurate, which may lead to accidents, and (ii) the time to compute the control inputs is limited to tens of millisecond. The current practice to introduce robustness is to resort to standard (deterministic) MPC with manually tightened constraints. In this project, we consider alternative methods using ideas from robust and sampling-based MPC, which introduce robustness by design. We show that these controllers work well in simulations and consistently outperform the deterministic MPC, both in terms of lap time and number of accidents. Motivated by these results, we have developed a heuristic sampling-based MPC controller, which we validate in experiments. Although our heuristic controller uses a much smaller sample size than theoretically required, it has the advantage of being realtime implementable with a computation time of tens of milliseconds. The performance of our heuristic controller is validated through experiments, where we show that it can vastly outperform the currently implemented controllers in terms of robustness while being competitive performance-wise.

Supervisors: Alexander Liniger, Xiaojing Zhang, Angelos Georghiou, John Lygeros


Type of Publication:

(12)Diploma/Master Thesis

File Download:

Request a copy of this publication.
(Uses JavaScript)
% Autogenerated BibTeX entry
@PhdThesis { Xxx:2016:IFA_5410
Permanent link