Enhanced IED Detection Using Ontological Knowledgebase with Fuzzy Reasoning

In Phase I, we have developed a suite of algorithms with capabilities including anomaly detection, false alarm mitigator, intelligent target grouping for minefield display to the human operator, and minefield detection. An ontological knowledgebase has also be developed, which provides a unified environment for the operation and coordination of individual algorithm modules. Test results using actual airborne composite images have shown that we are able to detect minefields with accuracy of 100%. In Phase II, we propose to add these technologies to Army vehicular system, which include intelligent IED minefield display, dynamic estimation of parameter values suitable for various operation conditions, interaction with operator feedback through speech recognition, and ontological knowledgebase management. System developed in Phase II will be deployed in a multiprocessor parallel computation platform for near real-time operation. With the new Phase II system added to Army vehicle, in addition to visualizing the target and its GPS coordinates, the human operator will also be able to view the corresponding IED minefield and providing his or her feedback. The combined system can also perform fuzzy reasoning and fuzzy learning to continuously improve the overall detection performance.