A search vehicle robot is required to move on any obstacles ground to any place with adequate controlled behavior. This study aims at acquiring autonomous behavior for such a search vehicle robot by physically modeled simulation. The vehicle robot driven by six wheels is modeled by use of the physical engine (PhysX) in a virtual environment. We use neuro-evolution to have the robot vehicle acquire autonomous behavior, namely the robot vehicle behavior is controlled by an artificial neural network (ANN) and ANN is evolved to let it learn proper behavior by evolutionary computation (GA). Simulation results verify that the vehicle robot can acquire autonomous behaviors.
