Horizon looks bright for autonomous ships

Blog post
Hannu Karvonen

Across many industries, companies looking to stay ahead of the competition are steering towards autonomous systems, and gaining time and cost savings as well as improving safety along the way.

In transportation, commercial drones and self-driving vehicles are to be joined by autonomous ships, which will one day navigate our oceans using artificial intelligence. But even unmanned vessels will need to be monitored and sometimes also controlled on demand by land-based professionals. This brings new challenges for autonomous ship navigation systems, which must be able to control ships in potentially critical conditions or, as they say, come hell or high water!

In the shift towards autonomy, equipment providers and ship builders widely believe the removal of the uncertain human element from on-board will dramatically cut the number of accidents at sea. But we’re not there yet and that’s where VTT comes in. For some time, VTT has been developing safe steering for future remote-monitored and controlled autonomous ships, and our new Autonomous Ship Handling Simulator Pilot (Apilot) in which autonomous navigation systems can be tested and developed, has become a high-profile part of our offering. Using Apilot for pre-field testing is a proven time and cost saver in developing automated features for vessels.

Smarter human interaction with intelligent systems

The Apilot simulator is also a perfect training tool for getting professionals to work better with their tools. In the past, maritime accidents related to automated navigation with autopilots have often been the result of human operators not interacting closely enough with their smart technologies on board.

The grounding of cruise ship Royal Majesty in Nantucket in 1995 is a case in point, where the crew on the bridge relied too heavily on their GPS-based navigation, even though it was not functioning and had shifted to dead reckoning many hours before. As a result, 32 000 tonnes of passenger ship headed towards the shoals, with personnel on the bridge still refusing all the signs around them that the vessel was miles off course. The crew reportedly also failed to recognise warning codes emitted by the electronic chart displays, indicating system failure.

Reliability and safety

Reliability and safety are core benefits of the Apilot navigation simulator, where VTT has a credible research record. This played to the advantage of my colleague, VTT Senior Scientist, Jussi Martio, in leading the development of Apilot as he was able to draw on our extensive research experience to establish test conditions that are as realistic as possible.

From a safety standpoint, it makes good sense to first work with a ship handling simulator in both developing and testing an autopilot system for manoeuvring a vessel. For example, it goes without saying that large autonomous ships need to have mastered collision avoidance before sailing into a crowded sea among passenger ships and pleasure boats.

Simulating and validating

Following simulator testing, the systems can graduate into the field to specially allocated testing areas at sea. In field testing, it is possible to combine the physical test area and Apilot’s virtual simulations when conducting safety validations. In this context, Apilot also works for the maritime industry as a data-gathering tool, providing clarificatory evidence of safety validation for partners and stakeholders, including local authorities. This is based on the expectation that autonomous systems algorithms must be sufficiently robust before putting out to real-world situations.

VTT specialises in this kind of simulation-based safety validation of autonomous systems, performing operational scenario-based safety testing and systematic evaluation against set requirements and acceptance criteria. These kinds of intelligent simulations are needed to prove concept viability and are right now causing a shift in thinking towards the renewal of entire industries.

Hannu Karvonen
Hannu Karvonen