Sideways Fall Avoidance of Bipedal Robots Based on Capturability Analysis

Abstract

Humanoid robots should be able to keep balance for any direction of disturbance. N-step capturability framework was proposed for fall avoidance that involves a number of steps. But it did not consider collision between the support leg and the swing leg in its analysis. For this reason, the conventional capturability analysis cannot be applied for a case in which a robot is pushed in sideways. In this paper, we analyze sideways legged locomotion in terms of capturability. Based on this capturability, we propose the sideways legged locomotion for fall avoidance. The proposed method is validated through numerical simulation. The simulated robot was able to recover from predetermined sideways pushes. It was also found that robots can be able to recover from larger disturbance with variable step length.