Optimization‐based autonomous racing of 1:43 scale RC cars

• DOI: 10.1002/oca.2123
• Publication Date: 2014-07-17T12:45:42Z
• Authors: Alexander Liniger; Alexander Domahidi; Manfred Morari
• Keywords: FOS: Computer and information sciences; Computer Science - Robotics; 0209 industrial biotechnology; Optimization and Control (math.OC); FOS: Mathematics; FOS: Electrical engineering, electronic engineering, information engineering; Systems and Control (eess.SY); 02 engineering and technology; Mathematics - Optimization and Control; Electrical Engineering and Systems Science - Systems and Control; Robotics (cs.RO)

This paper describes autonomous racing of RC race cars based on mathematical optimization. Using a dynamical model the vehicle, control inputs are computed by receding horizon controllers, where objective is to maximize progress track subject requirement staying and avoiding opponents. Two different formulations presented. The first controller employs two-level structure, consisting path planner nonlinear predictive (NMPC) for tracking. second combines both tasks in one optimization problem (NLP) following ideas contouring control. Linear time varying models obtained linearization used build local approximations NLPs form convex quadratic programs (QPs) at each sampling time. resulting QPs have typical MPC structure can be solved range milliseconds recent exploiting solvers, which key real-time feasibility overall scheme. Obstacle avoidance incorporated means high-level corridor dynamic programming, generates constraints controllers according current position opponents layout. performance investigated experimentally using 1:43 scale cars, driven speeds more than 3 m/s operating regions with saturated rear tire forces (drifting). algorithms run 50 Hz rate embedded computing platforms, demonstrating high optimization-based approaches racing.