Air ground coordination for image based control of an unmanned ground vehicle using aerial imagery
Abstract
This work presents a new framework for problem of motion planning and control of an unmanned ground vehicle (UGV) from an overhead camera or an unmanned aerial vehicle (UAV) using aerial imagery. Image based techniques for the control of a nonholonomic unicycle robot in the plane have been investigated and implemented. We have used ideas from projective geometry to formulate a mobile robot control problem in the image plane, thus expressing the robot control problem in image coordinates. We have obtained a working metric relationship between the velocities of the robot on the ground and the velocity of its projection in the image which enables us to calibrate cameras using a novel motion correspondence scheme. We have also looked into simplifying the control problem in the image plane. In order to do this we have used the idea of differential flatness, first proposed in the mid 1990s, with applications to the control of nonholonomic systems like cars and cars with trailers. We extend this idea to the image plane and express the feedback control problem entirely in the image plane. We envisage that the control of the ground robot will be done entirely in the image plane and there will be no need to have explicit knowledge of the world coordinates of the robot. Two typical mobile robot control problems have been addressed—(i) point to point control of a mobile robot and (ii) motion planning and path following strategies for a mobile robot. Further, we have also shown that these control strategies can be used not only for a static camera but also for the case of a camera that is attached to a moving overhead vehicle. The camera calibration routine and the controllers have been implemented in real time on a system consisting of the ER-1, a unicycle robot and a high resolution (1024 x 768) board level camera unit. Results from real time experiments are presented and we suggest some ideas for extensions of this work in 3D space as well as the problem of controlling multiple robots from multiple overhead cameras. ^
Subject Area
Engineering, Mechanical
Recommended Citation
Rahul S Rao,
"Air ground coordination for image based control of an unmanned ground vehicle using aerial imagery"
(January 1, 2004).
Dissertations available from ProQuest.
Paper AAI3152202.
http://repository.upenn.edu/dissertations/AAI3152202