Presentation Information

[10p-N106-6]Study on Artificial Kinesthetic Sensory for Ground Condition Perception and Gait Switching in An Amoeba-inspired Four-legged Autonomous Walking Robot

〇(M1)Hyoto Yamaguchi1,2, Seiya Kasai1,2 (1.RCIQE, Hokkaido Univ., 2.IST, Hokkaido Univ.)

Keywords:

Autonomous walking robot,Four-legged robot,Reservoir computing

We have developed an amoeba-inspired four-legged autonomous walking robot that successively finds gaits to go over the obstacles. At present, the locomotion efficiency is poor because it searches for possible gaits step-by-step depending on the ground condition. To improve the locomotion efficiency, a possible approach is to implement the capability to select an appropriate walking mechanism depending on the ground condition. In this study, we investigate to acquire information of ground conditions without visual information, in terms of multimodal sensing technique. Here we consider the touch sensors that detect whether each leg touches the ground and three-axis accelerometer. We will shows the result of gait-switching based on ground condition perception using an artificial kinesthetic sensory mechanism together with machine learning.