Multimodal Sensing and Information Integration for Multiple Object Tracking
Abstract
The funded research designed methods to improve multiple object tracking using measurements from multimodal systems. The methods combine sequential Bayesian filtering to estimate the time-varying parameters of physics-based models and Bayesian nonparametric modeling to infer and learn information directly from the measurements. Integrating these methods resulted in robust learning and increased performance when compared to current state-of-the-art methodologies. Multiple challenging scenarios were considered: time-varying number of moving objects, unknown measurement-to-object associations, time-varying environmental conditions, and multiple statistically-dependent measurements.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 18, 2020
- Accession Number
- AD1144427
Entities
People
- Antonia Papandreou-suppappola
Organizations
- Arizona State University