Model based Dynamic Obstacle Avoidance on Inclined Surface

Demonstration of a Gaussian process (GP) potential field method introduced in the ME5001 (master’s) project “Model-based Dynamic Obstacle Avoidance on Inclined Surface” by Xing Yan. In this video, we present two common issues in adjusting the GP-built potential field model, shown as the two failed examples. Additionally, we display a test without the use of the proposed potential field. We then provide two successful obstacle avoidance examples relying on our GP-based controller in two tasks of different complexity. To visually demonstrate the effect of the our method, we finally show the underlying GP potential field together with the positions of the robot (the red square) and obstacles (the black circles) through a successful run.