An Algorithm of Automatic Stair Climbing for a Vision-driven Hexapod Robot

In this paper, an automatic stair climbing algorithm is proposed for a vision-driven hexapod robot. The vision system recognizes and locates the stairs based on yolov3, and then detects the relevant data of the stairs in real time. On the basis, the algorithm can generate the corresponding motion plan according to the stair climbing strategy. To validate our algorithm, a vision-driven hexapod robot was constructed and tested in physical experiments in the setting of multiple sets of stairs. The experimental results show that our algorithm has a good measurement accuracy and good environmental adaptability.