Dynamic Obstacle Avoidance of UAV Using Chance Constrained Model Predictive Control

Addressing the challenge of dynamic obstacle avoidance for drones in uncertain environments, this paper introduces an innovative algorithm known as Velocity and Position Chance Constrained Model Predictive Control with Control Barrier Functions (CCVP-MPC-CBF). This algorithm comprehensively takes into account the probabilistic constraints of relative positions and velocities between drones and obstacles, introducing the concept of collision probability and establishing corresponding thresholds. Furthermore, by incorporating Control Barrier Functions (CBF) as constraint conditions, the algorithm ensures the safety and stability of the system. This strategy effectively overcomes the limitations that traditional MPCs may encounter when dealing with uncertainties in real-world systems, maintaining the invariance of the system's set while achieving optimal performance under the premise of ensuring system safety. Additionally, the method has been extended to enable collaborative obstacle avoidance among multiple drones. Finally, through comprehensive simulation experiments, it has been verified that even in complex environments, this algorithm enables drones to effectively avoid collisions with multiple dynamic obstacles, demonstrating the effectiveness and reliability of this approach.