Hierarchical Model Predictive Control for Obstacle Avoidance in Airborne Radar Based Detect-and-Avoid Systems Operating in a Varying Environment

This paper presents a model predictive control (MPC) based obstacle avoidance trajectory generation in uncrewed aircraft systems (UAS). The system aims to generate desired trajectories for the avoidance of flying aircraft with constant velocity within the collision avoidance threshold. This approach presents a two-layered MPC architecture incorporating a collision cone technique to achieve real-time avoidance of dynamic obstacles in non-cooperative sensor-based beyond-visual-line-of-sight (BVLOS) operation scenarios. The efficiency of the presented real-time solution is achieved by using a fast analytic collision cone technique for path planning and effective initial guesses for path tracking.