For motion planning and control of autonomous vehicles to be proactive and safe, pedestrians' and other road users' motions must be considered. In this paper, we present a vehicle motion planning and control framework, based on Model Predictive Control, accounting for moving obstacles. Measured pedestrian states are fed into a prediction layer which translates each pedestrians' predicted motion into constraints for the MPC problem. Simulations and experimental validation were performed with simulated crossing pedestrians to show the performance of the framework. Experimental results show that the controller is stable even under significant input delays, while still maintaining very low computational times. In addition, real pedestrian data was used to further validate the developed framework in simulations.
Real-Time Constrained Trajectory Planning and Vehicle Control for Proactive Autonomous Driving With Road Users
2019
7 pages, 9 figures, accepted to European Control Conference (ECC) 2019
Preprint
Electronic Resource
English
FOCUSED TRAJECTORY PLANNING FOR AUTONOMOUS ON-ROAD DRIVING
British Library Conference Proceedings | 2013
|SAGE Publications | 2021
|