Visual Pursuit Control with Target Motion Learning via Gaussian Process

In this paper, we propose an observer-based visual pursuit control law which integrates target motion learning via Gaussian Process (GP). We consider two rigid bodies: a controlled rigid body with a visual sensor, and a target rigid body whose velocity is unknown. Furthermore, a vision-based motion observer which estimates the target motion is introduced. Then, we propose an enhanced vision-based nonlinear observer and visual pursuit control which employ target motion learning by GP, where the GP prediction is based on estimated relative rigid body motion. Then, we quantify the performance and prove stability by the notion of uniformly ultimately boundedness. Finally, we demonstrate the effectiveness of the proposed control law through simulations.