Authors:
Sobers L. X. Francis
;
Sreenatha G. Anavatti
and
Matthew Garratt
Affiliation:
University of New South Wales@ADFA, Australia
Keyword(s):
Scene Flow, Lucas/Kanade, Horn/Schunck, 3D ,PMD Camera, Motion Detection, Velocity Estimation.
Related
Ontology
Subjects/Areas/Topics:
Control and Supervision Systems
;
Image Processing
;
Informatics in Control, Automation and Robotics
;
Mobile Robots and Autonomous Systems
;
Robotics and Automation
;
Vehicle Control Applications
;
Vision, Recognition and Reconstruction
Abstract:
This paper proposes a novel three dimensional (3D) velocity estimation method by using differential flow techniques
for the dynamic path planning of Autonomous Ground Vehicles (AGV) in cluttered environment. We
provide a frame work for the computation of dense and non rigid 3D flow vectors from the range data, obtained
from the time-of-flight camera. Combined Lucas/Kanade and Horn/schunck approach is used to estimate the
velocity of the dynamic obstacles. The trajectory of the dynamic obstacle is predicted from the direction of
3D flow field and the estimated velocity. By experiments, the utility of the approach is demonstrated with the
results.