Collision risk evaluation and collision-free path planning of autonomous surface vehicles considering the uncertainty of trajectory prediction

Abstract

This study presents a collision avoidance strategy for safe autonomous navigation of autonomous surface vehicles (ASVs). To recognize in advance a potential collision situation between the vehicle and an approaching object, a quantitative collision risk degree and a collision risk zone must be evaluated and predicted. However, they are difficult to accurately evaluate and predict, since the motion information of the vehicle and object includes various uncertainties caused by navigation sensors and environmental disturbances. Hence, the collision risk, which is expressed as a collision probability, is evaluated based on the probabilistic method considering the time-varying trajectory uncertainties of the vehicle and object. In addition, the collision risk zone on the predicted trajectory of the vehicle is determined considering the uncertainties and safe separation zones with respect to the vehicle and object. Then, a collision-free path is planned considering the dynamic characteristic of the vehicle to reduce and avoid the collision risk. To determine the feasibility of the proposed approach, simulations are performed, and the results are discussed. ？ ICROS 2018.