The University of Edinburgh -
Division of Informatics
Forrest Hill & 80 South Bridge

MSc Thesis #9601

Title:Gait Selection in Hexapod Robots
Date: 1996
Abstract:Locomotion is a skill that biological creatures do without conscious thought. What neural mechanisms do such creatures have to enable "thoughtless locomotion"? The answer to this question is generally now known: however research has established theories on possible architectures of the underlying mechanisms. This thesis concerns the implementation of one such theory, that locomotion is generated in decentralised nerve centres. Coordinated locomotion results from the couplings between these centres. The implementation is in the form of a heterogeneous neural network which generates rhythmic locomotory patterns for the control of a hexapod. The network, which is essentially 6 oscillators and the interconnections between these oscillators, successfully exhibits the two main gaits employed by insects when walking at the two extremes of speed. This is achieved by the variation of a single parameter. Further, the transition between gaits corresponds to that observed in insects. Also presented is a simple mechanism for effecting turning in the decentralised control system.

[Search These Pages] [DAI Home Page] [Comment]