A neural network for a walking machine

Hart, John Mark

Permalink

https://hdl.handle.net/2142/97994 Copy

Description

Title

A neural network for a walking machine

Author(s)

Hart, John Mark

Issue Date

1992

Director of Research (if dissertation) or Advisor (if thesis)

Uribe, Ricardo B.

Department of Study

Electrical & Computer Engineering

Discipline

Electrical & Computer Engineering

Degree Granting Institution

University of Illinois at Urbana-Champaign

Keyword(s)

• Electrical engineering
• Robotics
• Robotics--Control systems
• Robotics--Motion
• Neural networks (Computer science)

Language

en

Abstract

This research project was created to combine walking machines with neuron-based processing. Presented here is a design of a robotic walking machine with its controller based on the study of living organisms. A six-legged design was chosen due to the diverse population of organisms using this base for locomotion and also because of its inherent stability. A biological study was concentrated on the cockroach because of the wealth and diversity of documented research to uncover the neurophysiology of its control systems for the walking process. A physical robot was designed and constructed utilizing the resources of the Advanced Digital Systems Laboratory. Pneumatic cylinders, used for the movement of the physical parts, were selected for their strength in supporting the robot and its energy source, for the future autonomous capability of the robot. The organization of the neural control architectures are being implemented using a network of microcontrollers. An independent microcontroller for each leg has been used to provide the control signals necessary for movement and also for interaction with other microcontrollers. The nodes of the network mimic the local neural activity which takes place during the walking process in living organisms. The network has no hierarchy or central controller - the six nodes, with sensor feedback, interact together to form the walking pattern. This network was designed to propose the microcontroller as the nerve cell of future robots. During the time of the robot's initial walking demonstrations, it won the Lockmiller Award in Robotics and Artificial Intelligence from the Department of Mechanical Engineering at the University of Illinois.

Permalink

http://hdl.handle.net/2142/97994

Copyright and License Information

Copyright 1992 held by John Mark Hart

Owning Collections