Dynamic obstacle avoidance scheme based on model building and dimension reduction
Chunxi Zhu, Yingjiang Zhou, Runxian Cai, Can Wang · 2022
Aiming at the current obstacle avoidance strategy for dynamic obstacle avoidance is difficult to achieve both accuracy and speed. Opencv library is used to process the depth images obtained by Intel D435i in combination with the attitude and position information of Intel T265, and the future movement trend of obstacles is obtained. Obstacle information and car information are summarized into a four-dimensional data and compressed into binary images. The improved A* algorithm is used to obtain the binary image and achieve excellent control effect. The speed and accuracy of dynamic obstacle avoidance are achieved.