The Leader/Follower formation control of mobile robots with uncertain dynamical models is investigated in this paper. A combined kinematic/torque control law is developed for the follower using backstepping in order to accommodate the dynamics both of the follower and the leader in contrast with kinematic-based formation controllers. Neural network is introduced to approximate the dynamics of the follower as well as its leader using online weight tuning. In order to allow the follower to navigate around obstacles while simultaneously tracking its leader, an obstacle avoidance scheme is introduced. It is shown using Lyapunov theory that the formation tracking errors for the follower are uniformly ultimately bounded, and numerical results are provided.