Sensor Data Integrity and Mitigation of Perceptual Failures
Abstract
This report describes basic research to analyze and investigate methods for automatic detection and mitigation of sensor and perception data. Specifically, the PI will investigate 1) Metrics for sensor performance in context using multi-modal redundancy and information theoretic approaches; 2) Detection, localization and mitigation of faults and failures using hierarchical causal reasoning for diagnosis; 3) Develop methods for using low-level (e.g., sensor data fusion filter) versus high-level (autonomous inference) processes for automated selection of data to use for perception.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 2012
- Accession Number
- ADA566335
Entities
People
- Thierry Peynot
Organizations
- University of Sydney