Risk Assessment and Planning with Bidirectional Reachability for Autonomous Driving

Ming-Yuan Yu · Ram Vasudevan · Matthew Johnson-Roberson

Fig. 1: The ego vehicle (blue box) intends to perform an unprotected left turn to the goal (yellow star) at an intersection. The green shaded region is the observable polygon and arises due to the 1) limited sensor range and 2) occlusions f — FIGURE FROM THE PAPER

TL;DR

Combining forward and backward reachability lets a self-driving car assess occlusion risk efficiently, driving faster and cutting median collisions 7.5× at occluded intersections.

BRIEF

Driving through crowded intersections is hard when buildings and other vehicles block the view. Common risk methods either assume the worst about everything hidden, which makes the car overly cautious, or try to predict many unseen agents, which is expensive to compute. This work uses both forward and backward reachability to assess risk under occlusion. It can locate the sources of risk and feed into motion planning. In tests at occluded four-way intersections with up to five other vehicles, the ego vehicle reached its goal faster and the median collision rate fell 7.5 times versus state of the art.

Abstract

Risk assessment to quantify the danger associated with taking a certain action is critical to navigating safely through crowded urban environments during autonomous driving. Risk assessment and subsequent planning is usually done by first tracking and predicting trajectories of other agents, such as vehicles and pedestrians, and then choosing an action to avoid future collisions. However, few existing risk assessment algorithms handle occlusion and other sensory limitations effectively. One either assesses the risk in the worst-case scenario and thus makes the ego vehicle overly conservative, or predicts as many hidden agents as possible and thus makes the computation intensive. This paper explores the possibility of efficient risk assessment under occlusion via both forward and backward reachability. The proposed algorithm can not only identify the location of risk-inducing factors, but can also be used during motion planning. The proposed method is evaluated on various four-way highly occluded intersections with up to five other vehicles in the scene. Compared with other risk assessment algorithms, the proposed method shows better efficiency, meaning that the ego vehicle reaches the goal at a higher speed. In addition, it also lowers the median collision rate by 7.5× when compared to state of the art techniques.