Using Real-Time Stereo Vision for Mobile Robot Navigation

  • Authors:
  • Don Murray;James J. Little

  • Affiliations:
  • Computer Science Department, University of British Columbia, Canada. donm@cs.ubc.ca;Computer Science Department, University of British Columbia, Canada. little@cs.ubc.ca

  • Venue:
  • Autonomous Robots
  • Year:
  • 2000

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper describes a working vision-based mobile robot thatnavigates and autonomously explores its environment whilebuilding occupancy grid maps of the environment. We present amethod for reducing stereo vision disparity images totwo-dimensional map information. Stereo vision has severalattributes that set it apart from other sensorsmore commonly used for occupancy grid mapping. We discuss theseattributes, the errors that some of them create, andhow to overcome them. We reduce errors by segmentingdisparity images based on continuous disparity surfaces toreject “spikes” caused by stereo mismatches. Stereo visionprocessing and map updates are done at 5 Hz and the robotmoves at speeds of 300 cm/s.