Abstract
In this paper, we study an inverse kinematics problem in which optimum trajectories of walking robot toes are generated under the condition of minimum walking energy and the capability to avoid arbitrary shape obstacle. It is important for the walking robot that the motions of legs is determined to achieve the desired body operation. The trajectory of the toe is expressed by the multiple curves and the connection condition is used for reducing parameters. Neural networks are applied to generate the optimum trajectory parameters for an obstacle shape and a set of toe positions. Furthermore fuzzy clustering technique is introduced to reduce the duplication of data for learning of the neural network.
