Learning to perceive objects for autonomous navigation

Jing Peng, Bir Bhanu

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Current machine perception techniques that typically use segmentation followed by object recognition lack the required robustness to cope with the large variety of situations encountered in real-world navigation. Many existing techniques are brittle in the sense that even minor changes in the expected task environment (e.g., different lighting conditions, geometrical distortion, etc.) can severely degrade the performance of the system or even make it fail completely. In this paper we present a system that achieves robust performance by using local reinforcement learning to induce a highly adaptive mapping from input images to segmentation strategies for successful recognition. This is accomplished by using the confidence level of model matching as reinforcement to drive learning. Local reinforcement learning gives rises to better improvement in recognition performance. The system is verified through experiments on a large set of real images of traffic signs.

Original languageEnglish
Pages (from-to)187-201
Number of pages15
JournalAutonomous Robots
Volume6
Issue number2
DOIs
StatePublished - 1 Jan 1999

Fingerprint

Reinforcement learning
Navigation
Traffic signs
Object recognition
Reinforcement
Lighting
Experiments

Cite this

Peng, Jing ; Bhanu, Bir. / Learning to perceive objects for autonomous navigation. In: Autonomous Robots. 1999 ; Vol. 6, No. 2. pp. 187-201.
@article{4d6adaff5cdd4ae1b8854a02ffbdb2af,
title = "Learning to perceive objects for autonomous navigation",
abstract = "Current machine perception techniques that typically use segmentation followed by object recognition lack the required robustness to cope with the large variety of situations encountered in real-world navigation. Many existing techniques are brittle in the sense that even minor changes in the expected task environment (e.g., different lighting conditions, geometrical distortion, etc.) can severely degrade the performance of the system or even make it fail completely. In this paper we present a system that achieves robust performance by using local reinforcement learning to induce a highly adaptive mapping from input images to segmentation strategies for successful recognition. This is accomplished by using the confidence level of model matching as reinforcement to drive learning. Local reinforcement learning gives rises to better improvement in recognition performance. The system is verified through experiments on a large set of real images of traffic signs.",
author = "Jing Peng and Bir Bhanu",
year = "1999",
month = "1",
day = "1",
doi = "10.1023/A:1008887511945",
language = "English",
volume = "6",
pages = "187--201",
journal = "Autonomous Robots",
issn = "0929-5593",
publisher = "Springer Netherlands",
number = "2",

}

Learning to perceive objects for autonomous navigation. / Peng, Jing; Bhanu, Bir.

In: Autonomous Robots, Vol. 6, No. 2, 01.01.1999, p. 187-201.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Learning to perceive objects for autonomous navigation

AU - Peng, Jing

AU - Bhanu, Bir

PY - 1999/1/1

Y1 - 1999/1/1

N2 - Current machine perception techniques that typically use segmentation followed by object recognition lack the required robustness to cope with the large variety of situations encountered in real-world navigation. Many existing techniques are brittle in the sense that even minor changes in the expected task environment (e.g., different lighting conditions, geometrical distortion, etc.) can severely degrade the performance of the system or even make it fail completely. In this paper we present a system that achieves robust performance by using local reinforcement learning to induce a highly adaptive mapping from input images to segmentation strategies for successful recognition. This is accomplished by using the confidence level of model matching as reinforcement to drive learning. Local reinforcement learning gives rises to better improvement in recognition performance. The system is verified through experiments on a large set of real images of traffic signs.

AB - Current machine perception techniques that typically use segmentation followed by object recognition lack the required robustness to cope with the large variety of situations encountered in real-world navigation. Many existing techniques are brittle in the sense that even minor changes in the expected task environment (e.g., different lighting conditions, geometrical distortion, etc.) can severely degrade the performance of the system or even make it fail completely. In this paper we present a system that achieves robust performance by using local reinforcement learning to induce a highly adaptive mapping from input images to segmentation strategies for successful recognition. This is accomplished by using the confidence level of model matching as reinforcement to drive learning. Local reinforcement learning gives rises to better improvement in recognition performance. The system is verified through experiments on a large set of real images of traffic signs.

UR - http://www.scopus.com/inward/record.url?scp=0032633274&partnerID=8YFLogxK

U2 - 10.1023/A:1008887511945

DO - 10.1023/A:1008887511945

M3 - Article

AN - SCOPUS:0032633274

VL - 6

SP - 187

EP - 201

JO - Autonomous Robots

JF - Autonomous Robots

SN - 0929-5593

IS - 2

ER -