Online Gait Adaptation of a Hexapod Robot Using an Improved Artificial Hormone Mechanism

摘要

Walking animals show a high level of proficiency in locomotion performance. This inspires the development of legged robots to approach these living creatures in emulating their abilities to cope with uncertainty and to quickly react to changing environments in artificial systems. Central pattern generators (CPGs) and a hormone mechanism are promising methods that many researchers have applied to aid autonomous robots to perform effective adjustable locomotion. Based on these two mechanisms, we present here a bio-inspired walking robot which is controlled by a combination of multiple CPGs and an artificial hormone mechanism with multiple receptor stages to achieve online gait adaptation. The presented control technique aims to provide more dynamics for the artificial hormone mechanism with an inclusion of hormone-receptor binding effect. The testing scenarios on a simulated hexapod robot include walking performance efficiency and adaptability to unexpected damages. It is clearly seen that varying of hormone-receptor binding effect at each time step results in a better locomotion performance in terms of faster adaptation, more balanced locomotion, and self-organized gait generation. The result of our new control technique also supports online gait adaptability to deal with unexpected morphological changes.