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.