Robust walking of a quadraped robot

Abstract

A modular controller was developed to evaluate performance of robust locomotion control algorithms. An NUS developed quadruped was used as the testbed for this project. A low level controller was developed as two modules, the hardware interface and the fuzzy controller, which is attachable to any high level controller. Two high level controllers were developed - the behaviour-based and Central Pattern Generator (CPG) controllers. The behaviour-based controller is a modular controller based on interaction between various modules. Basic modules were developed to provide basic motion. Other modules were implemented to improve the controllera??s robustness. The CPG controller is based on the animal spinal systema??s coordination of locomotion. Pribe, Grossberg and Cohena??s Oscillator Model for CPG were implemented. The algorithms used in the behaviour-based controller were integrated to this controller to provide gait stability. Both static and dynamic gait patterns were simulated with online gait change.