Date of Award

3-29-2017

Document Type

Thesis

Publisher

Santa Clara : Santa Clara University, 2017.

Degree Name

Master of Science (MS)

Department

Mechanical Engineering

First Advisor

Christopher Kitts

Abstract

Sensor based navigation techniques are those that alter the path being followed based on realtime sensor readings. They are often used in scalar fields, which are regions with a scalar quantity of interest, such as temperature or concentration level, associated with every point in the field. An example of sensor based navigation in a scalar field is using a set of measurements from a distributed group of robots in order to estimate the gradient of a field; driving in the direction of the gradient leads the group to the maximum value in the field, a point that often has practical value. In previous work, researchers in the Santa Clara University Robotic Systems Laboratory have proposed multirobot sensor based controllers to move to maximum points, minimum points, and along field contour lines. This thesis work contributes to this work, presenting work performed with other researchers in the Lab to navigate ridges, trenches, and saddle points. Simulations have verified that these techniques can track field features with a minimal amount of steady state error. The complete set of primitive controllers was then used as a basis for an application controller that used a state machine architecture to switch between the primitive controllers in order to execute complex tasks. Four different applications were demonstrated in simulation using this method: ridge control with recovery, moving from one local maximum to another, mapping the contours around a maximum, and traveling to a designated location while staying above a particular scalar value. These controllers were successfully demonstrated in simulation under a number of ideal conditions. Future work is proposed in order to extend the capabilities of this technique and to improve robustness.

Available for download on Wednesday, April 03, 2019

Share

COinS