Improved Scene Understanding through Semantic Reasoning and Online Learning

This SBIR Phase I project will develop statistical relational learning (SRL) based framework that combines semantic reasoning and relational learning with deep convolutional neural networks for robust scene understanding. The proposed model reasons about the whole scene while incorporating the entire corpus of available information and their relationships with each other. The Phase I effort will focus on implementing the proposed framework and enabling technologies and demonstrating their effectiveness, and technical and operational feasibility for real-world autonomous ground vehicles. The project will benefit from Novateur’s experience in the areas of target detection and classification, multi-sensor data and decision fusion, UGV perception system, statistical relational learning, deep learning and adaptive context-aware systems for scene understanding and inference.