A Generic Decentralized Gait Generator Architecture for Statically Stable Motion of Crawling Robots

Abstract

The purpose of this paper is to present a generic architecture for generating gaits for crawling robots. The proposed architecture can be used to design gait-generators that output statically stable gaits for any crawling robot that only uses its legs for locomotion (No external support or wheeled locomotion). The problem of gait-generation is approached by decomposing it into sub-problems which are easier to model, implement and are in effect more human readable. Using hand-engineered neural systems to choose legs actuation order, their speeds and passing the output to robot' controller makes the whole system easier to understand and implement. The proposed architecture was evaluated by simulating a couple of robots in a virtual environment. It is found that in spite of a few shortcomings, the presented architecture can be used to generate gaits for a variety of robots.