Development of a vision-based situational awareness capability for unmanned surface vessels

Abstract

The current generations of unmanned surface vessels (USVs) are reliant on the human operator for collision avoidance. This reliance poses a constraint on the operational envelope of the USV as it requires a high bandwidth and low latency communication link between the USV and control station. This thesis adopts a systems engineering approach in identifying the capability gap and the factors that drive the need for a USV with autonomous capability. An algorithm employing edge detection and morphological structuring methods is developed in this thesis to explore the feasibility of using a computer vision--based technique to provide a situational awareness capability, which is required to achieve autonomous navigation. The algorithm was tested with both color video imagery and infrared video imagery, and the results obtained from processing the images demonstrated the viability of using this information to provide situational awareness to the USV. It is recommended that further work be done to improve the robustness of the algorithm.