Multi-robot Cooperative Obstacle Avoidance Based on Improved Artificial Potential Field Method
Fuyou Su, Chaoyang Huang, Jun Xu · 2022
Obstacle avoidance is important in the cooperative control of multi-robot system. Model predictive control (MPC) is used extensively in trajectory planning, in which a known global path is tracked with small tracking error. Due to the physical size and speed of the robot, dynamic obstacle avoidance should be included in MPC. This paper introduces an improved artificial potential field (APF) method to optimize the trajectory planning method based on MPC. Firstly, a fractional gravitational field is set at the target point to ensure that the robot reaches the target point. Secondly, a simplified repulsive force field is set up on the static obstacle, and it is segmented and linearized to make the repulsive force change smoothly and avoid deadlock due to the repulsive force balance at the target point. Finally, a communication mechanism between robots is introduced to realize real-time sharing of robot position information, so that a repulsive potential field of dynamic obstacles can be constructed for the neighboring robots. Experiments on simulation platform and robot platform verify the effectiveness of the method.