Fault-tolerant multi-agent formation control using distributed nonlinear MPC with discrete-time super-twisting sliding mode fault estimation

Abstract

This paper presents a novel Fault-Tolerant Distributed Nonlinear Model Predictive Controller for Formation Control of Agents with Fractional-Order Dynamics (DNMPC-FCFO). The proposed approach enhances fault tolerance in multi-agent systems, particularly in environments with obstacles, by implementing a discrete-time fractional-order sliding mode fault estimation (DTFO-SMF) technique. Unlike conventional integer-order systems, this method leverages fractional-order dynamics to achieve more accurate fault estimation. A discrete-time sliding mode is employed for fault detection in distributed fractional-order systems, addressing the lingering effects of faults on system dynamics. By introducing a fractional-order sliding mode, our approach ensures precise fault estimation for each agent, accounting for disturbances indirectly during the estimation process. Building on previous research, we incorporate contractive constraints and a Lyapunov function into the optimization framework, ensuring system stability. The fractional-order system design is integral to developing a controller that prevents agent collisions and enables obstacle navigation, even under limited communication among mobile robots. In scenarios where an agent experiences a fault, adhering to predefined constraints becomes challenging. However, the proposed fault estimation mechanism supports the continued proper functionality of affected agents. Simulation results demonstrate the effectiveness of our approach in maintaining formation control and obstacle avoidance, highlighting its potential for practical applications.