Optimizing the Gait of a Humanoid Robot Towards Human-like Walking

Achieving a stable, human-like gait with a humanoid robot is a challenging problem. While several rather simple as well as more complex techniques exist to generate stable walking patterns, only little attention has been paid towards the resemblance to the human gait. Popular gaits, for example, apply the strategy to bend the knees and to swing the torso in lateral direction in order to ensure stability. As a result, the walking patterns do not look very human-like. However, human resemblance is an important aspect whenever robots are designed to interact naturally with humans. In this paper, we present techniques to optimize an initial, stable gait of a humanoid robot with respect to human resemblance. To acquire walking data of a human, we use a full-body motion capture system. We propose four different optimizing algorithms that work at joint angle basis and use the joint angle difference as measure of similarity. Experimental results carried out with a HOAP-2 robot in simulation demonstrate that we can adapt the robot’s initial gait so that it is significantly more human-like.