Cooperation Control of Under-actuated Mobile Robots with RBF-NN Approximator

The underactuated system adaptive control is a tough problem since its dynamic model limitation and nonlinear unknown parameters. In this paper, based on a two active wheels mobile robot kinematics and dynamics model, a Radial Basis Function neural network (RBFnn) was embedded into control system to approximate unknown terms. In mathematics aspect, the control algorithm is based on Lyapunov direct theory and backstepping method, which solving the states (position, orientation, velocity, etc) errors boundedness and convergence problem. The controller includes following aspects: 1.Driving robots states approach to predefined location, 2.Using saturation to avoid step signal disturbance, 3.Approximate unmodeled dynamic terms by RBF neural network. Moreover, a kind of cooperation control methodology was designed, to ensure several robots running in a specific formation. Finally, the simulation results performs that the system is stable and reasonable in the large modeling errors.