Dissertation Information for Jeffrey D. Will NAME: - Jeffrey D. Will
DEGREE:
- Ph.D.
DISCIPLINE:
- Electrical Engineering
SCHOOL:
- University of Illinois, Urbana-Champaign (USA) (2001)
ADVISORS: - John F. Reid
COMMITTEE MEMBERS: - None
MPACT Status: Incomplete - Not_Inspected
Title: Sensor fusion for field robot localization
Abstract: This dissertation addresses many practical engineering issues in the automation of an off-road vehicle with a concentration on sensor fusion for localization. The sensor platform in this research consists of a real-time kinematic differential Global Positioning System (RTK-DGPS), a fiber-optic gyroscope (FOG), an inertial inclination sensor, and a vision sensor. Error models for the RTK-DGPS are developed on two levels. The first model focuses on heading error and is based on vehicle speed and angular velocity. The second model is based on an autoregressive error model for RTK-DGPS position solutions as a function of the status of DGPS operation.
These sensors are used together in a Kalman filter to provide an estimation of the vehicle pose in real time. A framework for heading determination by fusing RTK-DGPS and a FOG is presented, as well as a Kalman filter for determining the full pose of the vehicle. Simulation results show that the position estimate error during coarse modes of DGPS operation can be reduced from several meters to less than 25 cm by this method. Further, this technique is readily applicable to code-phase DGPS receivers, for which a 30% reduction in error is demonstrated.
A side-slip-compensated model of vehicle motion is presented for use with the Kalman filter, which takes into account four different surface conditions and has been shown to reduce the level of systematic error without the addition of sensors or additional filter states. Systematic errors of 10% in the prediction of heading are reduced to below 1% in using this model.
A system for determining the three-dimensional coordinates of feature points in images taken from a single vision sensor of the vehicle is presented. Experimental results from both indoor and in-field tests show accuracies of ∼10 cm.
Finally, this dissertation presents an object-based software framework for implementing sensing, navigation, and control for autonomous mobile robots. Underlying software patterns used in this framework are identified. This framework is comprised of a generic, reusable, and flexible class hierarchy which has been demonstrated to promote parallel platform development and cooperative system development.
| |
MPACT Scores for Jeffrey D. Will A = 0
C = 1
A+C = 1
T = 0
G = 0
W = 0
TD = 0
TA = 0
calculated 2010-03-03 20:14:36
Advisors and Advisees Graph
generating graph, please reload |