Research on the Dynamic Obstacle Avoidance Method of Unmanned Surface Vehicle Based on RVO

The stability of collision avoidance is directly related to the safety of USVs (Unmanned Surface Vehicle). However, the uncertainty of perception about the velocities of dynamic obstacles seriously undermine the stability of collision avoidance. Thus, this research focuses on the problem of USV autonomous obstacle avoidance under uncertain conditions. First, the speed and heading deviation are calculated based on maximum likelihood estimation. According to the calculation results, the uncertainty is converted into certainty to establish dynamic uncertainty model. Based on the model, we improve closest point of approach method to advance the accuracy of threat measures. Then, using the gradient descent method to optimize avoidance strategy. To prevent jumping changes in collision avoidance strategy, an USV autonomous obstacle avoidance algorithm is proposed based on the RVO (Reciprocal Velocity Obstacle) algorithm, which is more conducive to increase the stability and safety of the USV system. Finally, the proposed algorithm is test on the simulation platform.