While the importance of wheeled robots is well recognized, the control of these systems need to be further investigated, because the current navigating methods of wheeled robots are difficult to achieve high-speed navigation in an environment with obstacles. A theoretical framework is developed to take the robot, environments and goals all together into account based on viability theory. At first, the viability boundary is constructed. Then, the viability boundary is applied for building the viability condition constraints. Finally, with these constraints, an objective function was developed for goal guidance. The simulation shows that the control method can overcome the complexity of robot navigation with nonholonomic constraints, and achieves radical high-speed movement.