Object recognition for the HERMIES robot

Technical Report · Sat Oct 01 04:00:00 EDT 1988

OSTI ID:6660026

Bowyer, K W

The HERMIES-IIB robot in the CESAR lab at ORNL is used in research related to vision, multi-sensor integration, coordinated arm movement, path planning, parallel algorithms, and learning. This report describes the development of a visual object recognition system for HERMIES. This system is implemented on an NCube system, an MIMD hypercube parallel computing system. In general, the input to the object recognition system will be an image containing one or more instances of the known different types of obstacles, and the output will be the location(s) of the obstacle(s) relative to the cameras. Based on assumptions about the known geometry of the obstacles, the viewing conditions, the relevant range of viewing distances and the image resolution needed, it is proposed that the obstacles can be recognized by the presence of roughly rectangular outlines in the image. The object recognition system uses Hough transforms to find rectangular outlines in the image, and then uses a sequence of constraints to eliminate those outlines which could not be due to obstacles. 7 refs., 4 figs.

• Research Organization: Oak Ridge National Lab., TN (USA)
• DOE Contract Number: AC05-84OR21400
• OSTI ID: 6660026
• Report Number(s): ORNL/TM-10935; CESAR-88/55; ON: DE89004259
• Country of Publication: United States
• Language: English

Related Subjects

42 ENGINEERING
420203* -- Engineering-- Handling Equipment & Procedures
99 GENERAL AND MISCELLANEOUS
990220 -- Computers
Computerized Models
& Computer Programs-- (1987-1989)
CAMERAS
COMPUTER ARCHITECTURE
COMPUTER CODES
IMAGE PROCESSING
N CODES
NAVIGATION
PATTERN RECOGNITION
PROCESSING
REMOTE SENSING
RESOLUTION
ROBOTS
SPATIAL RESOLUTION