Learning-Based Distributed Safe DET Formation Control for Uncertain Multiagent Systems With Dynamic Interagent Coupling Weights
Abstract
The safe formation control is studied for uncertain nonlinear multiagent systems (MASs) through a learning-based framework coordinated with control barrier functions (CBFs) and a barrier Lyapunov function (BLF). To mitigate chain reaction avoidance, a dynamic interagent coupling weight is developed, in which the coupling weights are adjusted according to obstacle proximity. Within the backstepping framework, a learning-based nominal virtual controller is constructed to compensate for unknown nonlinear dynamics. Then, two CBF-induced safety conditions are formulated for obstacle avoidance and interagent collision avoidance, and the nominal virtual controller is refined through a quadratic program to obtain a smooth safety-oriented virtual controller. To support the implementation of the safety constraints, the second-step error is confined within a predefined bound by the BLF. Furthermore, a dynamic event-triggered mechanism (DETM) based on BLF-related barrier information is designed to reduce unnecessary control updates. Forward invariance, uniform ultimate boundedness, and exclusion of Zeno behavior are established. The effectiveness of the proposed method is demonstrated through simulation results.