Anytime Trajectory Optimization for Multi-Drone Systems With Guaranteed Collision Avoidance

We present ATOMICA, Anytime Trajectory Optimization for MultI-drone systems with guaranteed Collision Avoidance, a novel algorithm designed to generate guaranteed collision-free trajectories for multi-UAV systems. Each UAV communicates with the others and treats them as dynamic obstacles within a receding-horizon guidance framework. Recursive feasibility is ensured by maintaining a safe backup trajectory at all times. The time-dependent collision avoidance constraints are efficiently handled using positivity certificates, eliminating the need for potentially unsafe time discretizations while enabling fast collision checking. The non convex optimization problem is solved using the convex concave procedure, which provides ATOMICA with anytime capability, allowing users to predefine the duration of each replanning step. We evaluate the algorithm through simulations, demonstrating a 22% reduction in mission duration compared to state-of-the-art methods. Additionally, we validate its real-time capabilities through real-world experiments.