Adaptive and Intelligent Object Avoidance for Unmanned Surface Vehicles (USVs)

Automated object avoidance for Unmanned Surface Vehicle (USV) for both Navy (and DoD in general) and commercial applications has been the subject of much research and development over the past decades. The creation of a set of robust software tools for fully automated detection, classification/recognition and tracking of objects is vital in the development of an avoidance system for USVs. Without such tools the Navy operation of USV will require the existence of the man-in-the-loop (MITL) navigation. A multitude of current and future research and developments support the potential values and needs of autonomous systems. The use of truly autonomous USVs has been hampered by a lack of sophisticated and resource efficient obstacle avoidance systems. Current approaches have focused on either expensive active sensor systems or inferential processing techniques that are computationally intensive. In this proposal, UtopiaCompression and Defense Research Associate (UC/DRA) present an adaptive and intelligent system concept, using innovative algorithms in computer vision, mobile robotics, and computational intelligence, in particular Machine Learning (ML), which will facilitate true autonomous USV operations by providing a real-time, end-to-end solution tailored to Navy's sensor data and mission requirements using inexpensive imaging sensors and modest computational resources.