Autonomous vehicle steering
Abstract
Controlling the steering of a vehicle using a single forward-looking camera is a difficult problem. This work presents a solution under certain conditions which has proven very robust after many hours of testing and which can be implemented cheaply. The solution is composed of three stages. In the first stage, a novel dual-correlation method is used to identify the lane boundaries and from these the position and orientation of the vehicle. Next, a model of the vehicle as well as some additional vehicle information is used to filter the estimated position and orientation from the first stage. In the final stage, constant state feedback based on the discrete linear quadratic regulator (DLQR) is used to determine the appropriate steering of the vehicle. After each stage is explained, experimental results are given which demonstrate the effectiveness of the proposed solution.
Degree
M.S.E.C.E.
Advisors
DeCarlo, Purdue University.
Subject Area
Automotive engineering|Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.