Input-output linearization using model reference adaptive control techniques

Based on the performance of a unicycle robot, the MRAC methodology is used to define update laws for estimation of parameters for the I/O linearization of nonlinear friction effects which occur in series with the linearized plant. A linear controller closing loops around the plant-nonlinearity concatenation is then able to control the system successfully. In implementation, lack of persistence of excitation allows the parameter estimates to drift, leading to imperfect I/O linearization; suitably bounding the estimates, however, ensures that stability and performance are maintained.< >