Research on Cooperative Obstacle Avoidance Method for Multi-Rotor UAV Swarms Based on Adaptive Sliding Mode Control

Abstract

In the context of complex urban environments, the demand for dynamic obstacle avoidance and formation control in multi-rotor unmanned aerial vehicle (UAV) swarms has become increasingly prominent. This study addresses these challenges by proposing a collaborative control method that integrates lidar perception with adaptive sliding mode control. An enhanced YOLOv8 algorithm is employed to achieve real-time detection and trajectory prediction of dynamic obstacles, thereby constructing a constrained space for safe avoidance. Subsequently, an adaptive sliding mode controller is designed, which, combined with an adjacency graph communication topology, enables distributed cooperative control within the swarm. This approach effectively mitigates the chattering issues commonly associated with traditional sliding mode control. Validation through hardware-in-the-loop simulations and actual flight experiments demonstrates that the proposed method significantly reduces the response time for obstacle avoidance in dense obstacle scenarios while maintaining formation position errors within a low range. The results indicate a marked improvement in both operational safety and efficiency for UAV swarms engaged in urban logistics and emergency rescue missions, offering a practical and reliable solution for real-world applications.