Dynamic Window Approach Using GOA Search Algorithm for Collision Avoidance Based on Obstacle Trajectory Prediction

Dynamic Window Approach Using GOA Search Algorithm for Collision Avoidance Based on Obstacle Trajectory Prediction

Blog Article

Mobile robots not only need to navigate autonomously in complex environments, but also need to avoid collisions with dynamic obstacles, especially pedestrians.This paper realizes dynamic obstacles detection and state estimation algorithm with LiDAR data, and proposes the path planning and obstacle avoidance algorithm based on dynamic obstacle trajectory prediction.The current planning methods of mobile robots are mostly applicable to static scenes and lack adaptability to pedestrian dynamics.

The core idea of the proposed algorithm is to guide the mobile robot towards the back of the pedestrian to reduce the psychological pressure on the pedestrian as well as to avoid interfering with the pedestrian’s mel axolotl movement and to achieve dynamic obstacle avoidance, which is in line with the idea of going around from the back in the Convention on the International Regulations for the Avoidance of Collisions at Sea (COLREGS).Firstly, the pedestrian detection method is studied on the basis of Lidar data, and the pedestrian motion state is estimated based on Kalman filter to avoid error detection.Secondly, the dynamic window approach (DWA) method is improved and pedestrian dynamic is added with carefully designed algorithm for path planning and obstacle avoidance.

Furthermore, to improve the sequal eclipse 5 battery computational efficiency and flexibility, the GOA(Grasshopper Optimization Algorithm) is used to improve the DWA search schema.Finally, the effectiveness of the proposed method is proved by comparing the existing methods with simulation and experimental tests.The research of this paper can improve the safety of the robot in the human-machine crowd scene.