New test method paves the way for Autonomous Vehicles in the future

The Move_UK project has been ongoing for three years and this month completes its final phase of research, designed to accelerate the development and deployment of automated driving systems.

Direct Line Group is part of the consortium, led by BOSCH and joined by other key players in the motoring industry, Transport Research Laboratory (TRL), Jaguar Land Rover, the Royal Borough of Greenwich and The Floow.

 The consortium’s main objective is to develop a new methodology that will revolutionise the way Automated Driving Systems (ADS) data is captured in the future.

“Connected Validation” is a method that has been developed by the Move_UK consortium and which allows ‘events’ to be recorded when specific trigger events occur. For example, when sudden breaking takes place. This has allowed 8,500 hours of driving footage to be distilled into 450 short driving sequences where on board systems have detect a potential incident. This new methodology makes if far easier to analyse the data and really understand the risks associated with ADS.

The research also highlighted that the data can also provide a better understanding of human driving behaviour that will be instrumental in the development of Automated Vehicle technology, and which will help level 3 or level 4 automated cars understand and drive more like the human drivers they will share the roads with.

Traffic sign recognition (TSR) validation was another positive result that came out from the research. The project monitored where the car’s TSR recorded street signs across hundreds of locations in Greenwich. The data showed that in some cases the signs were not recognisable and in one case the sign did not register because vegetation had grown so much, the sign was barely visible. Having this information available enabled the team to react and get the vegetation trimmed to give the sign more visibility. In another case the street signs were pointing in the wrong directions. For example, the camera was looking at 20mph street sign on a road with a 30mph speed limit, but which related to an off-road where the speed limit was 20mph. By simply turning the street sign to face a different way, the car was able to respond to the correct speed limit. The learnings from this exercise proved that while road infrastructure plays a vital role in enabling AVs to operate on UK roads, this method of data validation makes it easy to discover where and when the infrastructure is failing. This will make our roads safer for AV and human drivers alike.

The new technique used by Move_UK significantly reduces the time it takes to process and analyse data and can help bring a roadworthy system to market. Using this methodology will also give us rich data plus a much better understanding of human driving behaviours, especially around certain areas. The recorded events can map out exactly where there may be areas of concern, for example around tricky roundabouts, zebra crossings, country roads or where people park on both sides. Having a methodology that can capture the information in a simple, comprehendible way will enable us to also understand the insurance risks associated with these new driving technologies and will help us develop insurance products of the future.

The process will now be developed for next generation of driver assisted systems and can help insurers, regulators and government in paving the way for widespread adoption of autonomous vehicles.

In our next blog we’ll explore Streetwise - a collaboration with FiveAI, Transport of London, Transport Research Laboratory, Oxford University and Direct Line Group which is testing autonomous cars on the streets of London, with an aim to get a better understanding of the AV systems used and how the technology responds in the real world.