| The problem of adaptive control in the presence of input constraints is addressed. The new control laws are presented in the context of a holonomic mobile robot pushing and steering a wheelchair using a single manipulator. Adaptive control is employed to estimate, in real time, the inertia characteristics of the wheelchair and its occupant, and friction between the wheelchair and the surface that it traverses. The input constraints considered are actuator saturation and actuator dynamics. Saturation of control input limits the maximum input that can be provided to the system and input actuator dynamics dictate the speed with which the control input can be realized. If left unaddressed, these constraints can result in estimation windup, poor tracking performance and/or instability. Additionally, in this case the problem is further exacerbated due to the fact that only a single manipulator is to be used for both pushing and steering due to weight and cost considerations. The control designs presented here complement the mechanical design of the manipulator such that a holonomic mobile robot employing the manipulator can maneuver a wheelchair asymptotically along a reference trajectory. The reference trajectories can be generated autonomously or provided by a human operator. Apart from the significant improvements in the performance of adaptive control subject to input constraints, this work has the potential to provide greater mobility to residents of Long Term Care facilities, which are currently experiencing an acute shortage of nurses and an increasing number of inmates. In general it can be used for maneuvering wheeled platforms. Experimental results employing an RP-6 mobile robot platform and a wheelchair are presented to validate the theoretical results. |
