Stabilization and moving efficiency improvement by adjustment of moving speed in single locomotion

This paper shows improvement of mobility by adjusting moving velocity of the robot using extended locomotion selection algorithm. We propose an indicator based on stability and moving efficiency called locomotion reward in order to realize adjusting moving speed. Locomotion reward is extended to add explicitly state penalty depending on moving speed. The robot adjusts the speed so as to maximize locomotion reward. We analyze the behavior of moving speed by proposed method checking the relationship between stability and moving speed and differential of locomotion reward. It can be confirmed the usefulness of proposed method. Not only the analysis, we simulated the proposed method with bipedal walking in environment including rough terrain by OpenHRP3. The simulation result shows that the robot can improve moving efficiency on flat increasing moving speed and stability on rough terrain decreasing the one by proposed method. We can confirm that this proposed method has a high generality in decision-making problem.