A Learning Controller for Periodic Disturbance Rejection of Uncertain Nonholonomic Systems

A learning control approach is developed to tackle the robust stabilization problem for a class of chained-form non-holonomic systems with dynamic uncertainty and nonlinear parameterization. Different from the existing results, the additive periodic perturbations are allowed in the input channel, and a learning-based feedforward term is incorporated into the composite controller against unknown periodic dynamics with a known period. Moreover, a switching adaptive control strategy is constructed to get around the smooth stabilization burden associated with the nonholonomic systems due to Brockett's necessary condition. Numerical simulation results are presented to illustrate the performance of the proposed algorithm.