Laboratory Research in Autonomous Sensory Perception
Abstract
This report documents the research performed in autonomous sensory perception through automatic sensor based decision making in a robot. The robot's task is to touch simple curved surfaces and estimate their geometric parameters in a completely autonomous fashion with inexpensive sensors. That is, after initial contact, it is to intelligently explore the object so as to understand its shape as quickly as possible. Equipment funded under the grant included several simple sensors and sensor parts, such as a wrist-mounted force/ torque sensor, a track-ball transducer, and data acquisition electronics, as well as the high-speed computer host which acts as controller. This equipment was successfully integrated with a Merlin 6540 industrial robot which was programmed to maintain force-controlled contact with miscellaneous curved objects. Also funded was one month of Principal Investigator time and three months of graduate assistant support. The system properly executed the exploration procedure, which is based theoretically upon the estimator uncertainty measure provided by a Kalman filter covariance matrix.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 30, 1991
- Accession Number
- ADA241753
Entities
People
- John S. Bay
Organizations
- Virginia Tech