Proactive Risk Mitigation and Reactive Control for Safe and Smooth Automated Driving

Qizhan Tam, Therese Cypher-Plissart, Chris J Ostafew

Abstract: In addition to predictable static and dynamic objects on public roads, automated driving systems must also handle unpredictable hazards such as parked car doors opening, cyclists falling, or pedestrians stepping off curbs. One extreme approach would be to plan such that all hazards would come to pass, though this results in an uncomfortably conservative system. Another extreme approach would be to perform aggressive reactive maneuvers when hazards materialize, though this results in an uncomfortably reactive system. This paper proposes a robust approach that handles hazards through proactive planning and reactive control to achieve a smooth and safe automated driving (AD) while enabling the use of lower cost sensors.

TL;DR: Principled and human-like fusion of proactive planning and reactive control for safe and smooth autonomous driving.

Keywords: trajectory planning, model predictive control, risk mitigation, safety