MPC-Based Obstacle Aware Multi-UAV Formation Control Under Imperfect Communication

We propose a distributed Model Predictive Controller to solve the problem of multiple fixed-wing UAV formation control and collision avoidance under time-varying communication constraints. Under a graph theoretic leader-followed based formation control pattern, we exploit the Artifical Potential Fields (APF) collision avoidance. We then investigate the model by simulating time-delays, missing linkages and noise to the existing topology, and seek to improve the performance using Kalman filters with Gaussian assumption. We introduce holistic cost function which incorporates formation, reference objective and controls, we are able to guarantee early convergence for the UAV system. We perform simulations in Microsoft AirSim, and analyze the results to study the validity and effectiveness of our implemented model.