We consider an autonomous vehicle platoon system consisting of N + 1 vehicles in the presence of modeling uncertainty. The uncertainty may be due to parameter variations, aerodynamics, external disturbances, etc., which is nonlinear and time-varying. Subject to the collision avoidance consideration, the original state is one-sided restricted. To resolve this restriction, we propose a state transformation to convert the bounded state into a globally unbounded state. Furthermore, motivated by the properties of artificial swarm systems, we incorporate the swarm system performance into the platoon system by treating it as a d'Alembert's constraint. By the Udwadia and Kalaba's approach, we obtain the analytic (closed-form) expression of the constraint force. Based on this, a class of robust controls for each vehicle (except the leading vehicle) is proposed to drive the platoon system to follow the ideal swarm model. Four major system performances are accomplished: (i) compact vehicle formation, (ii) collision avoidance, (iii) stable platoon system formation, (iv) global behavior.