Dynamic Obstacle Avoidance of Mobile Robot Based on Adaptive Velocity Obstacle

Abstract

The Velocity Obstacle (VO) provides a good solution for mobile robot collision avoidance problem. VO divides the velocity region into collision zone and non-collision zone. The robot can avoid collisions by selecting the velocity in the non-collision zone. However, if the robot chooses the critical velocity of the collision cone to minimize the time consumption, the error of the sensor will lead to the collision. To solve this problem, the Adaptive Velocity Obstacle (AVO) is proposed which has a good balance between safety and efficiency. The expansion radius of the moving obstacle is adjusted adaptively according to the relative velocity and distance between the robot and the obstacle. As the radius changes, the collision cone can also change dynamically so that the robot can maintain a safe path under the sensor detection error while ensuring real-time performance. After, the Dynamic Window Approach (DWA) is used to get the local path through optimizing the velocity outside the collision cone. Finally, we carried out comparative simulations. Results show that the AVO can reduce the chance of collision under sensor error.