Date of Award

1-1-2010

Document Type

Thesis

Degree Name

M.S.

Department

Electrical Engineering

First Advisor

Richard M. Voyles

Keywords

Active Tether, Emulated Synthetic Neural Network (ESNN), Soft Robot, Water Hammer Actuation

Abstract

This thesis presents components of an on-going research project aimed towards developing a miniature soft robot for urban search and rescue (USAR). The three significant contributions of the thesis are verifying the water hammer actuation previous work, developing an estimator of water hammer impulse direction from hose shape, and creating the infrastructure for distributed cognitive networks. There are many technical issues in designing soft robots, in terms of perception, actuation, cognition, power, physical structure and so on. We are focusing on actuation and cognition issues in this thesis. We investigated water hammer actuation as an alternative system which provides a continuously distributed form of actuation results from water hammer effect. It is special because it is a soft actuation method. We generated some comparison experiments and verified the benefits of the water hammer actuation, and also designed our soft robot to be hose-like in order to utilize the water hammer actuator. For the cognition part, we first addressed and verified that the shape of the hose-like robot has impact on impulse direction from the water hammer actuation. And then we implemented an emulated synthetic neural network (ESNN) to analyze the direction of the impulse from the water hammer actuation. Then in order to achieve the long-term goal, we distributed the emulated synthetic neural network onto many embedded system boards to achieve a distributed cognitive network. The distributed nodes in the network are using Bluetooth communication.

In the comparison experiments between the active tether system and passive tether system, we can clearly see the benefits of active tether in momentum transfer and friction reduction. For example, in the drag test, with the water hammer actuation the burden that the tether can pull was increased by about 1.6 times. For the distributed cognitive network, we successfully built an emulated synthetic neural network on distributed embedded system boards. With the shape information as the inputs, the difference on outputs from the ESNN and the experimental results is less than 3%.

Provenance

Recieved from ProQuest

Rights holder

Xiaoting Yang

File size

78 p.

File format

application/pdf

Language

en

Discipline

Electrical engineering

Share

COinS