Collisions between the mobile manipulator and surrounding obstacles during task execution can result in task failure or even physical damage to the robot. To address this issue, a novel varying parameter neural dynamics with penalty strategy (VPNDP) model is proposed for the motion planning of mobile manipulators. The VPNDP model formulates the practical problems as a time varying quadratic programming (TVQP) problem subject to equality, inequality, and boundary constraints. Inequality constraints are transformed into penalty terms in the objective function with penalty function method. The optimal solution is then obtained through a varying parameter neural dynamics method. Computer simulation results verify the effectiveness, practicality, and accuracy of the proposed VPNDP model.