Reactive Collision Avoidance Using the Velocity Obstacles Concept in Polar Coordinates

Abstract

In order to autonomously navigate in an unknown environment, a mobile robot should perceive the environments precisely and generate fast-moving path without collisions. In recent years, as the operating environment is becoming more and more complicated, considering various factors such as multiple agents, moving obstacles becomes an important issue in autonomous navigation. Therefore, it is necessary to develop a collision avoidance navigation algorithm which is effective in a variety of situations. A centralized navigation system collects information of the environments and all robots, and decides trajectories of each robot. As the environment gets more complex, calculating the collision-free trajectory is difficult. A distributed navigation system which controls the robot individually cannot guarantee optimal path of the robot, but it is easy to apply depending on the situation. This dissertation addresses local and reactive navigation without centralized coordination or control. A velocity obstacle approach one of local and reactive navigation methods is re-analyzed here. Most of the conventional velocity obstacle approaches are analyzed in Cartesian coordinates. The proposed approach of this dissertation performs collision prediction and avoidance motion planning of a mobile robot with non-linear velocity based on robot-centered polar coordinates. By re-analyzing the velocity obstacles concept in robot-centered polar coordinates, obstacle avoidance process has been simplified. Depending on the direction of the robot and the moving obstacles, the robot occasionally selects oscillating velocity as a result of using the conventional velocity obstacle approaches. In order to overcome the oscillation, new strategy which decides velocity of the robot to avoid collision with the oscillation-free path is designed. The proposed evaluation function is containing the current status of the robot, the relation between the robot and the obstacle, and the distance to the destination. The evaluation function is used for the robot velocity decision. Numerous simulations have been implemented to validate the proposed approach as well as the conventional algorithms. The performance of the proposed approach is verified by comparing the traveling time, distance, and computation time, and the smoothness of the robot path with the conventional algorithms.