Proposal of Horizon Variable MPC and Its Application to UAV Obstacle Avoidance

With advancements in hardware and computational algorithms, it has become possible to apply computationally intensive, sophisticated control methods to the control of high-speed dynamic systems. This paper proposes a new horizon variable MPC (Model Predictive Control) method. This method was developed to reduce computational cost while ensuring control performance. To verify the effectiveness of the proposed method, we have applied it to the obstacle avoidance control problem of a UAV as an example. In this case, the Dijkstra algorithm is used to derive the reference trajectory, and the proposed method is used to achieve optimal path control to avoid obstacles. Numerical experiments confirm that the proposed method works effectively for this problem.