Abstract: An approach based on autonomous behavior agent of lunar rover motion planning and controlling is presented. First, based on the autonomous behavior agent, a group of behaviors is designed, and its theory is proven. Then, a behavior automata is used for behavior coordination. When the motion planning cannot guarantee the rover safety, a motion planning agent is invoked to learn behavior parameters, which are then memorized bya neural network. At the same time, the rover behaviors can meet both constraints of rover maneuverbility and lunar terrain transferbility. The approach can timely plan and control the rover in an unknown world. Simulation and experiment results are shown to demonstrate the effectiveness of the approach.