AUTONOMOUS AGENTS UTILIZING PASSIVE OPTICAL SENSING FOR CONTROL

Abstract

The project addresses the problem of integrating a passive optical sensor onboard an autonomous unmanned aerial vehicle (UAV) and its operation in future missions against various targets/threats. The thesis develops the analytical and experimental software and hardware framework around a commercial-off-the-shelf quadcopter (COTSQ) for autonomous missions. The COTSQ features advanced autopilot capabilities along with an onboard optical sensor that is fully integrated into the control software. This research is applicable to any agent that has similar passive optical sensing capability. The work addresses the following key research questions and concepts: • Prove the sufficiency of the inner-outer loop design of a COTSQ autopilot to perform the autonomous sensing and detection of targets and threats in real time • What is the desired architecture of a high-level guidance controller to enable a single COTSQ and then multiple COTSQs to optimally search and automatically detect targets/threats? • How can the optical feed be used to facilitate detection and tracking of objects based on the real-time video feed from the UAVs? • How can this optical feed be utilized to advance the intelligent autonomy of the UAVs? • What are the fundamental limitations of achievable performance of the following components? o onboard instrumentation to enable robust execution of the search & detection mission and its robust scalability to potentially unlimited number of UAVs.