Dor Raviv: Orca AI
We now need ‘eyes’ able to spot trouble in the most crowded waters, says Dor Raviv of Orca AI.
It started for Raviv at the age of 13, “during the night watch onboard my father’s catamaran off the coast of Croatia... I thought I saw something but there were so many confusing lights from the land, a pitch black sky everywhere else and the radar was too full of noise to make sense”.
It was a close shave: soon after waking his father it became apparent a huge commercial ship was bearing down on them “and we needed to take action very, very quickly”.
He’s never forgotten “being that petrified” and he knows others don’t always escape harm: “There are around 4,000 collisions a year, both major and minor - but they usually don’t take place far out at sea, but close to the port or on the approach, in congested waters. It’s where the usual navigational aids just aren’t that efficient, because you haven’t just got one nearby vessel to think about, you can have hundreds to deal with.”
But life had to lead him down a few, apparently discontinuous paths before he was in a position to do much about it. After staying on in the navy to head up a project investigating unmanned patrol vessels he left the maritime field entirely to become involved in image recognition: “Eventually I went to work for Orcam, developing a device that can clip onto glasses which helps visually impaired people to understand their environment – like a supermarket... and can even help them to read.” Very cool stuff.
However it all suddenly gelled during a chance meeting with an old colleague from his navy days. In a very short time the pair had hatched a plan to take radar, AIS, GPS inputs, add thermal cameras and tie it together with AI that can work on spatial recognition.
While the technology is presently being trialled on cargo carriers, it’s not just for the big ships. Patrol craft, dredgers, supply vessels are all candidates - it can be applied to a 120m container vessel or a 20m tugboat, he explained.
Despite the sophisticated aim, he’s against using very expensive equipment “as what we need is something that will do it at an affordable cost”. This has been enabled by both the plummeting price of sensors and thermal cameras along with the rise of machine learning which can process images from ordinary video sources.
But, although the idea with Orca AI is “that it will recognise what’s dangerous and what’s not, and alert the crew”, he’s convinced it won’t stop here.
Decision-making support, he said, “is one step along the route” to unmanned vessels. Though he admitted “the commercial industry is not quite ready” for fully autonomous ships, he believes the push will be gentle but inevitable as machine-learning becomes an integral part of daily life.
And our boats will start talking... to each other.
By Stevie Knight
Latest Press Releases
Dynamic Load Monitoring, specialists in the design and manufacture of Load Cell’s and Cable Working ... Read more
Dixon is a premier manufacturer and supplier of marine fluid transfer products for a wide range of a... Read more
Caspian Offshore Construction has recently taken delivery of a Damen Stan Tug 1606 ICE class in Kaza... Read more
On 2nd August, Damen Shipyards Hardinxveld hosted a naming ceremony for S. T. Marine Support’s new M... Read more
Damen Shipyards Group has recently signed a contract for the delivery of a pontoon to France-based c... Read more