Autonomous swarming unmanned surface vessels (SUSVs) — equipped with Johns Hopkins APL-developed hardware and autonomy software

Uncrewed Surface Vessel Perception

Our Contribution

International regulations for preventing collisions at sea require vessels to operate within certain distances based on the visual identification of other vessels. We have applied machine-learning techniques to teach autonomous vessel-piloting systems how to identify and then safely respond to other ships. Our team compiled an extensive database of images at diverse aspects, ranges, and environmental conditions, and used this data to design and demonstrate a vessel-recognition system accurate enough to meet international requirements for safe passage of vessels.