Optimal walking pattern generation for a quadruped robot using genetic-fuzzy algorithm

In this paper, we described an optimal walking pattern generation using genetic-fuzzy algorithm that can assist walking robot avoid obstacles. In order to walk on an uneven terrain, a quadruped robot must recognize obstacles and take a trajectory that fits with the environment. In that respect, the robot should have two decision-making algorithms that will help its structural limitation. The first algorithm is to generate a body movement that can be related to the movement of the legs, and the other is to make legs' movements smooth in order to reduce jerks. The research presented in this paper, using genetic-fuzzy algorithm, suggests how to find an optimal path movement and smooth walk for quadruped robots. To realize such movement, a relationship between body path and legs trajectory was defined, and a rule based on genetic-fuzzy algorithm was made. From that rule, the optimal legs' trajectory could be determined and the body path generated. As a result, a quadruped robot could walk and avoid obstacles with smoothness.