Robust anytime tracking combining 3D shape, color, and motion with annealed dynamic histograms - Robotics Institute Carnegie Mellon University

Robust anytime tracking combining 3D shape, color, and motion with annealed dynamic histograms

David Held, Jesse Levinson, Sebastian Thrun, and Silvio Savarese
Miscellaneous, U.S. Patent 9,710,925 issued July 18, 2017, July, 2017

Abstract

Velocity controllers in accordance with embodiments of the invention enable velocity estimation for tracked objects. One embodiment includes a tracker controller, including: a processor; and a memory containing: a velocity tracker application; a state space describing relationships between measured locations, calculated locations, and changes in locations, where the calculated locations in the state space correspond to unoccluded points on the surface of the tracked object; wherein the processor is configured by the velocity tracker application to: pre-process the state space to identify a tracked object; estimate a velocity of the tracked object using a location history calculated from the measured locations of the tracked object within the state space and a motion model calculated from the state space; and return the velocity of the tracked object.

BibTeX

@misc{Held-2017-113062,
author = {David Held and Jesse Levinson and Sebastian Thrun and Silvio Savarese},
title = {Robust anytime tracking combining 3D shape, color, and motion with annealed dynamic histograms},
booktitle = {U.S. Patent 9,710,925 issued July 18, 2017},
month = {July},
year = {2017},
}