Hierarchical structure of deterministic chaos in a chaotic neural networkmodel (CNN) is investigated in the view of application in robotics. Theresult shows a rich capacity of CNN in selectively controlling the synchro-nization of neuron outputs, and sensitively responding to external sensoryinputs, both being based on the intrinsic mechanism of the dynamics calledchaotic itinerancy. Choosing appropriate parameters, the simple designedrobot realized a chaotic search to the hierarchically selected directions. Themacroscopic drift preserving chaotic fluctuation was also derived by simplyadding weak external inputs to an intended direction. Obstacle avoidancewas simulated with the use of these properties.