On-Board Data Handling for Longer Duration Autonomous Systems on Expeditionary Missions

This STTR Phase I project will demonstrate the feasibility and effectiveness of novel biologically-inspired computational memory models for on-board exploitation of long-duration sensor data streams to enable autonomous missions in unknown environments. The key innovation in this effort is a computationally and space-efficient computational memory model that is able to: i) handle long-duration data streams; ii) identify informative features in data streams; iii) learn from unlabeled sensor observations; iv) adapt to new scenarios; v) store learned experience in short term and long term memories and their semantic associations; and v) perform prediction and inference using the observations and the learned models. The proposed model provides a framework for modeling and solving a large variety of autonomous learning and prediction problems that arise in UAV and UGV missions. The Phase I effort will include; development of proposed models, solution of UAV and UGV problems using the models, performance optimization for SWaP constrained onboard processing, quantitative and qualitative evaluation of the proposed technologies, and demonstration of proof of concept using real-world data from multiple use-cases. The project will benefit from the Ohio State University"s expertise in computational memory modeling and Novateur Research Solution"s experience in sensor exploitation and onboard processing.