Streetwise: developing self-driving vehicles for daily commute
Cars with a single occupant fill the streets of towns and cities in the UK. This leads to congestion and pollution and a lot of frustrating delay.
Many predict that self-driving vehicles could transform urban environments. Autonomous vehicles will travel the streets and travellers will share rides.
The UK government estimates that the global market for so-called connected and autonomous vehicles could be £907 billion by 2035.
UK-based FiveAI, a specialist in building self-driving-vehicle technology, including artificial intelligence, computer vision and robotics, is leading a project, supported by the Industrial Strategy Challenge Fund, that will look at the technical, regulatory, safety and social challenges faced by an urban commuter service based on shared self-driving electric vehicles.
Ben Peters, vice-president product of FiveAI, said: "Current public transport economics means aggregating passengers on to large buses, enabling operators to distribute the cost of a driver. This means many stops resulting in longer journeys or, at other times of the day, empty buses. If you remove the driver cost you can operate smaller vehicles even at busy times, making fewer stops. It encourages people to give up their personal cars. We then see knock-on benefits in terms of reduced congestion and pollution."
Project includes a demonstration service on the road
FiveAI is leading the 30-month £20 million Streetwise project to develop the technology and demonstrate how electric self-driving vehicles could provide a commuter service between Croydon and Bromley in outer London.
Ben added "Autonomous vehicles must recognise all the different actors in a scene and be able to do it in various weather conditions. Crucial to urban environments is being able to predict how these actors will interact with each other and the autonomous vehicle. We can already operate autonomously under certain conditions where traffic density is low, although we always operate with a trained safety driver in the vehicle who can intervene.
We think there’s approximately another 4 years of development before we’re able to remove the safety driver in complex traffic situations. Part of the project will involve further developing the technology, based on computer vision, lidar, radar, and artificial intelligence. It also involves gathering vast data from the route, including on cars, buses, trams, cyclists, and pedestrians and how they interact."
The trial service, with a trained safety driver, will happen around the end of 2019 and beginning of 2020 and will include feedback from commuters on whether it would make them give up their car and how they feel about being driven autonomously.
Consortium includes academics, insurer and safety experts
FiveAI has won £9.3 million through the Centre for Connected and Autonomous Vehicles from the Industrial Strategy Challenge Fund’s Driverless Cars Challenge to help it to lead the project and develop the technology.
A key partner is the Torr Vision Group of the University of Oxford, under computer vision and artificial intelligence expert Professor Philip Torr, which has won £1.1 million to support leading-edge work on how to continually add data to the learning systems at the heart of autonomous vehicles and on identification of vulnerable road users.
Insurance company Direct Line has £180,000 to support its work to understand the insurance implications of driverless vehicles. It is working with McLaren Applied Technologies (£149,000) and the Transport Research Laboratory (£1.1 million) on the safety case
Other partners include Transport for London (£407,000); experts in modelling, simulation and virtual testing of autonomous vehicles Claytex Services (£147,000); and the University of Warwick’s Warwick Manufacturing Group (£400,000), which has experience in testing of autonomous vehicles against specific scenarios.
Find out more about the Driverless Vehicles Challenge
Company looking to commercialise autonomous service
FiveAI has quickly grown from 13 employees at the time it was drawing up the project to now employing 150 people. It is focused on delivering the Streetwise project and then commercialising the results of its work.
It raised £14 million in private finance in 2017 and will be looking at further investment to help it to commercialise the platform as a service.
Ben added "Each new route we operate will require qualification. We will need to capture a lot of data to simulate those routes and make sure we are safe to operate across them. This project will build the tools and infrastructure that will enable us to do that and add more routes."