Frame-Based Monitoring to Detect and Explain Multimodal Autonomous System Errors
Abstract
The PI proposes to develop a monitoring system that can detect and explain errors for multimodal autonomous systems. The problem arises due to autonomous systems that perceive, process, and explain information differently than humans do. When an autonomous vehicle fails, it will need to develop capabilities to explain its actions. The PI proposes a technical approach to develop frame-based monitoring system that will (1) synthesize multimodal information into a symbolic language and (2) detect and explain system errors via a monitoring system architecture. Thus, the novelty is in the application of a frame-based representation to a multimodal autonomous system composed of multiple parts for perception sensors, planning components, and actuators.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 06, 2025
- Source ID
- FA95502410149
Entities
People
- Leilani Gilpin
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of California, Santa Cruz