Field Demos of Autonomous Collision Avoidance
Over the last four years, Maritime Robotics has collaborated with NTNU, DNV-GL and KONGSBERG in the Autosea project on sensor fusion and collision avoidance for autonomous surface vehicles. This June, NTNU and Maritime Robotics successfully demonstrated two collision avoidance algorithms on the Telemetron USV. In the below video, you can see the Telemetron USV avoid several ships in three collision avoidance scenarios: an overtaking scenario, head-on collision scenario, and one where both are combined.
The main goal of the Autosea project is to develop methods for guidance and navigation of autonomous ships. A central component of this is collision avoidance. The Autosea project has demonstrated complete collision avoidance systems in full-scale experiments involving autonomous and semi-autonomous surface vehicles both in Trondheimsfjorden and in the Netherlands, with Maritime Robotics’ Telemetron USV. The Telemetron is a ‘converted’ vessel, an originally manned vessel which has been adapted such that it can be run autonomously. It still has the original steering and control on board, which allows for a safety driver to be present, and an on-board ‘vehicle control station’ also allows anyone to operate the autonomous system from on board.
The collision avoidance algorithms, scenario-based model-predictive control and branching-course model-predictive control, were developed at NTNU by Dr. Giorgio D. Kwame Minde Kufoalor, Inger Berge Hagen, and Bjørn-Olav H. Eriksen. The algorithms operated on data about nearby vessels coming from either AIS signals or radar signals. Radar tracking was implemented by NTNU’s Erik F. Wilthil. The algorithms were tested on Maritime Robotics’ Telemetron USV, whose software and systems for the Autosea project have been implemented over recent years by Arild Hepsø, Thomas Ingebretsen, Andreas Misje, Kenan Trnka, Olav Gjestad and Stephanie Kemna.
The Autosea project has been very successful in its multiple collision avoidance demonstrations over the last years, building towards fully autonomous collision avoidance for unmanned vessels. It will be followed up by the Autosit project, which focuses on creating enhanced situational awareness for remotely operated or autonomous vessels. All these technology developments are fundamental steps for increasing autonomy of our unmanned surface vehicles; the Otter, Mariner, and Conversions, and our work towards creating autonomous passenger ferries.
Further reading (in Norwegian): https://www.adressa.no/pluss/okonomi/2019/06/19/P%C3%A5-Trondheimsfjorden-l%C3%A6rer-f%C3%B8rerl%C3%B8se-skip-hvordan-de-skal-unng%C3%A5-%C3%A5-kollidere-19243626.ece