A Biologically Inspired Four-Legged Walking Robot (Robo-Dog)

Harold Ear

The Robo-dog was designed and implemented by a Shiqi Peng, a PhD student in 2006. The dog incorporated many control systems such as Reinforced Learning, Subsumption Architecture and a Central Pattern Producer. The dog since 2013 has been in an inactive state, the dog now is used to educate students on various concepts such as debugging, electronics and programming. A variety of issues stem from the dog and are not limited to the lack of documentation, missing equipment and components, broken and faulty hardware components and a limited functioning program. The objectives of the project are: restore various capabilities of the dog, moving the dog from Point A to Point B, implementing a control system utilising pre-existing sensors on the dog. To begin attempting to manoeuvre the dog, various components such as tilt-sensors, foot-sensors and microcontroller communication needed to be addressed as the components were not in a working state. Documentation has been updated or created to reflect the changes that have been made to the components of the dog. Moving the dog from Point A to Point B was achieved through various walking techniques, by adapting and adjusting the limb movements in the walk gait for a dog and then applying the concepts to the robo-dog. The implementation of the limb movement and the walk gait focused on maximising forward movement per cycle incorporating a sequential approach where each movement were dependant on one other. A control system was implemented in the form of Subsumption Architecture; the behaviours implemented are to determine the balance of the dog, walk sequence of the dog and to determine whether the dog is walking on a ramp. The implementation goal is to provide an easy-to-expand control system so that future students can add additional behaviours to the system.

A Biologically Inspired Four-Legged Walking Robot (Robo-Dog)

Files and links (1)

Abstract

Details

Metrics