A model-based fault tolerant control design for nonholonomic mobile robots in formation

Abstract

In this paper, a new model-based fault tolerant kinematic/torque control law is developed by using backstepping for leader-follower robots in formation. In comparison to the kinematic-based formation controllers, the proposed control law design considers the dynamics of the robots and the formation. Initially, the control law is developed for leader and follower mobile robots under normal operation (i.e., no faults); the stability of the formation is verified using Lyapunov theory. Later, an online model-based fault tolerant design is introduced in the presence of a fault, where the fault could be incipient or abrupt in nature. In other words, the fault is mitigated by adding an extra term into the existing control law, which is a function of the unknown fault dynamics, recovered using a neural network. Using Lyapunov theory, the stability of the fault tolerant control law and the formation errors are guaranteed to converge asymptotically in the presence of faults and system uncertainties. Finally, numerical results are used to verify the theoretical results presented in the paper.