Integrating Visual and Range Data for Robotic Object Detection

DMCA

Integrating Visual and Range Data for Robotic Object Detection

Cached

Download Links

by Stephen Gould , Paul Baumstarck , Morgan Quigley

Citations:

38 - 3 self

BibTeX

@MISC{Gould_integratingvisual,
    author = {Stephen Gould and Paul Baumstarck and Morgan Quigley},
    title = {Integrating Visual and Range Data for Robotic Object Detection},
    year = {}
}

OpenURL

Abstract

Abstract. The problem of object detection and recognition is a notoriously difficult one, and one that has been the focus of much work in the computer vision and robotics communities. Most work has concentrated on systems that operate purely on visual inputs (i.e., images) and largely ignores other sensor modalities. However, despite the great progress made down this track, the goal of high accuracy object detection for robotic platforms in cluttered real-world environments remains elusive. Instead of relying on information from the image alone, we present a method that exploits the multiple sensor modalities available on a robotic platform. In particular, our method augments a 2-d object detector with 3-d information from a depth sensor to produce a “multi-modal object detector.” We demonstrate our method on a working robotic system and evaluate its performance on a number of common household/office objects. 1

Keyphrases

robotic object detection range data robotic platform depth sensor 2-d object detector robotics community much work cluttered real-world environment high accuracy object detection multi-modal object detector computer vision object detection robotic system common household office object sensor modality visual input multiple sensor modality 3-d information great progress

DMCA