Concurrent algorithms for a mobile robot vision system

Conference · Fri Jan 01 04:00:00 EST 1988

OSTI ID:5521738

Jones, J P; Mann, R C

The application of computer vision to mobile robots has generally been hampered by insufficient on-board computing power. The advent of VLSI-based general purpose concurrent multiprocessor systems promises to give mobile robots an increasing amount of on-board computing capability, and to allow computation intensive data analysis to be performed without high-bandwidth communication with a remote system. This paper describes the integration of robot vision algorithms on a 3-dimensional hypercube system on-board a mobile robot developed at Oak Ridge National Laboratory. The vision system is interfaced to navigation and robot control software, enabling the robot to maneuver in a laboratory environment, to find a known object of interest and to recognize the object's status based on visual sensing. We first present the robot system architecture and the principles followed in the vision system implementation. We then provide some benchmark timings for low-level image processing routines, describe a concurrent algorithm with load balancing for the Hough transform, a new algorithm for binary component labeling, and an algorithm for the concurrent extraction of region features from labeled images. This system analyzes a scene in less than 5 seconds and has proven to be a valuable experimental tool for research in mobile autonomous robots. 9 refs., 1 fig., 3 tabs.

• Research Organization: Oak Ridge National Lab., TN (USA)
• DOE Contract Number: AC05-84OR21400
• OSTI ID: 5521738
• Report Number(s): CONF-880462-2; CESAR-88/06; ON: DE88005850
• Country of Publication: United States
• Language: English

Related Subjects

42 ENGINEERING
420200* -- Engineering-- Facilities
Equipment
& Techniques
CAMERAS
COMPUTER ARCHITECTURE
COMPUTERS
HYPERCUBE COMPUTERS
IMAGE PROCESSING
NATIONAL ORGANIZATIONS
ORNL
PATTERN RECOGNITION
PROCESSING
RESEARCH PROGRAMS
ROBOTS
US AEC
US DOE
US ERDA
US ORGANIZATIONS
VISION