Improving Performance for Multi-Agent Systems using Mixed-Type Feedback and Fuzzy Logic Tuning

Abstract

The synchronization of multi-robot systems has numerous applications, including search and rescue missions and collaborative manufacturing. In multi-agent systems, information transmission delays in the communication network can lead to the degradation of overall system performance. The work in this thesis leverages the existence of delayed information in the controller feedback and proposes a method of tuning the controller parameters online with fuzzy logic to improve the synchronization performance. Teams of direct current motors and teams of Euler-Lagrange manipulators are considered. The stabilities of the proposed linear and nonlinear control policies are analyzed by defining mixed-type error dynamics and using linear matrix inequality techniques and Lyapunov’s method. Simulations are conducted to observe the effect of introducing self-delayed states into the controller feedback for different networks and the effectiveness of the proposed fuzzy logic control method. Experimental testing of a group of Phantom Omni manipulators are performed to validate the proposed controller.