Temporal Pattern Recognition: A Network Architecture for Multi-sensor Fusion
Abstract
This paper proposes a self-organizing network architecture for the learning and recognition of temporal patterns. This architecture is used to perform multisensor fusion and scene analysis for ROBART II, a second-generation autonomous sentry robot. The ability to understand one's environment, the goal of a multi-sensor fusion system, is not governed solely by static pattern recognition. The order in which sensor firings occur can be as important as the events themselves, and an intelligent robotic system must be able to detect and understand this ordering. Thus the dimension of time allows access to a wealth of information about the current environment, past events, and expectations about the future. Thus the ability to incorporate time into information processing is essential for multi-sensor fusion, and is necessary for robotic tasks such as the recognition of sequences of events, understanding cause and effect, making predictions and planning.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1989
- Accession Number
- ADA205550
Entities
People
- C. E. Priebe
- D. Marchette