A comparison of continuous and discrete tracking-error model-based predictive control for mobile robots

Model-based predictive control approaches can be successfully applied to the trajectory tracking of wheeled mobile-robot applications if the process nonlinearity is considered, if real-time performance is achieved and if assumptions made in the control-law design are met when applied to a particular process. In this paper, continuous tracking-error model-based predictive control is presented. The controllers optimal actions are obtained from an explicit solution of the optimization criteria, which enables fast real-time applications. Due to its design in continuous time, its usage is not limited to the uniform sampling restrictions of a host computer, as is usually the case in discrete time design. Therefore, better performance is obtained in applications with non-uniform sampling, which is natural in many situations due to imperfect sensors, mismatched clocks, nondeterministic control delays or because of the unknown time of the pre-processing. The controller-design parameters are insensitive to the sampling time period, which contributes to simpler applications and greater robustness of the controller. We present a new continuous tracking-error model-based predictive control algorithm for mobile robots.Comparisons are made to our previous work with discrete design.Better performance, the design parameters are insensitive to the sampling time.Approach enables a non-uniform sampling which is natural in many applications.