Optical Flow based Velocity Estimation for a Mobile Robot
Advisor Information
Prithviraj Dasgupta
Location
Milo Bail Student Center Omaha Room
Presentation Type
Oral Presentation
Start Date
8-3-2013 1:00 PM
End Date
8-3-2013 1:15 PM
Abstract
In this research, we address an important problem in mobile robotics – how to estimate the speed of a moving robot or vehicle using the technique of optical flow obtained from a series of images of the moving robot captured by a camera. Our method generalizes several restrictions and assumptions that have been used previously to solve the moving target velocity estimation problem – we use an uncalibrated camera, we do not use any reference points on the ground or on the image,and we do not make any assumptions on the height of the moving target from the ground. The only known parameter is the distance of the camera from the ground with the assumption that it is placed parallel to the ground; a case which holds for almost all robotic systems. In our proposed method we exploit the optical flow patterns generated by varying the focal length of the camera in order to pinpoint the principal point on the image plane and calculate the height of the moving target. This height is then used to calculate the speed of the target. We have verified our technique experimentally using a camera and a mobile robot and shown that our method proves to be a parsimonious yet effective irrespective the distance and the speed of the mobile robot.
Optical Flow based Velocity Estimation for a Mobile Robot
Milo Bail Student Center Omaha Room
In this research, we address an important problem in mobile robotics – how to estimate the speed of a moving robot or vehicle using the technique of optical flow obtained from a series of images of the moving robot captured by a camera. Our method generalizes several restrictions and assumptions that have been used previously to solve the moving target velocity estimation problem – we use an uncalibrated camera, we do not use any reference points on the ground or on the image,and we do not make any assumptions on the height of the moving target from the ground. The only known parameter is the distance of the camera from the ground with the assumption that it is placed parallel to the ground; a case which holds for almost all robotic systems. In our proposed method we exploit the optical flow patterns generated by varying the focal length of the camera in order to pinpoint the principal point on the image plane and calculate the height of the moving target. This height is then used to calculate the speed of the target. We have verified our technique experimentally using a camera and a mobile robot and shown that our method proves to be a parsimonious yet effective irrespective the distance and the speed of the mobile robot.