World-first simulation

to test AV safety

Project

The development of CAVs offers an opportunity to dramatically redesign the way we approach transport in urban and rural environments across the UK. Drawing on experience from the public, private and academic sectors, OmniCAV’s objective is to accelerate the development of safer, smarter mobility by providing an end-to-end framework for the testing and certification of autonomous vehicles (AVs).

OmniCAV was awarded funding as part of a competition run by the Centre for Connected and Autonomous Vehicles (CCAV). With expertise across the country and from multiple disciplines, the ten-strong consortium, which is jointly funded by industry partners and Innovate UK, will spend two years developing a world-first simulation platform that can be used by accreditation bodies, insurers and manufacturers to accelerate the development and certification of AVs.

On a broad scale, the project addresses one of the ‘Grand Challenges’ the UK Government set out in its 2018 Industrial Strategy - the future of mobility. Having started in December 2018 the consortium will work together until November 2020, when the project will culminate in testing an autonomous vehicle via the entire end-to-end OmniCAV testing programme in a public demo, from simulation, to proving grounds, all the way through to public roads.

How does it work?

Infographic

Innovations

OmniCAV’s world-first innovations will accelerate the development, deployment, and adoption of safe autonomous vehicles.

Representative spectrum of UK road environments

Efforts related to the development of AV and smart mobility solutions have so far been heavily focused on urban areas. However, there are huge potential benefits to bringing these innovations to rural areas as well. While urban residents already have access to a wide range of mobility options, those living in the countryside are often forced to own a car for lack of other options. Including suburban and rural areas in the OmniCAV road loop will establish the differences in testing requirements between urban and rural settings and help  unlock the benefits of AVs for everyone.

Comprehensive scenario library

The OmniCAV project will build a library of scenarios against which AVs should be tested in simulation. Using scenarios in simulations is crucial to validating the safety of AVs, as it allows for particularly dangerous or complex situations, which occur rarely in the real world, to be included in the testing process. OmniCAV will create a best-in-class library by integrating three state-of-the-art methods of scenario generation: accident data, insurance claim data, and Systems Theoretic Process Analysis (bottom-up analysis of the vehicle’s control software to identify failures).

Smart actors for realistic dynamic contexts

The simulations used to test autonomous vehicles need to be faithful recreations of the real world. This realism must extend beyond the static environment – roads, trees, buildings, etc. – to include the dynamic context around the vehicle under test. The hardest problem to solve for AVs is how to interact safely with human road users, so realistically simulating the behaviour of the virtual humans that interact with the vehicle under test is crucial. OmniCAV’s simulation platform will include world-first Artificial Intelligence-based ‘smart actors’, working in conjunction with a traffic model for the road loop, calibrated and validated by traffic data taken from the real roads.

Trusted testing process

The simulations used to certify that autonomous vehicles are safe need themselves to be trusted and auditable. To help build trust in the integrity of the testing results, the OmniCAV project will develop concepts such as tamper-proof inputs and outputs and will be designed with due consideration to the need for accreditation. The consortium will benefit from experience in the certification of safety-critical systems in the aerospace industry.

High-fidelity geospatial data

OmniCAV will put in place best practices and establish new specifications and standards for collecting geospatial data, such as 3D features and network infrastructure. These will contribute to developing a digital copy of the UK in order to build a nationwide virtual test environment. As part of the OmniCAV project, a digital representation of the 32km road loop in Oxfordshire will be built, serving as an exemplar for future work on building a broader geospatial infrastructure.

Real-world validation

It is essential to ensure that a simulation used for autonomous vehicle testing is a realistic representation of the real world, and correctly predicts the outcomes of situations in the real world. The OmniCAV project will include tests conducted in the real world, both in controlled environments on proving grounds and on public roads. These tests will allow the simulator’s performance to be validated by recreating real-world scenes in the simulation and comparing the outcomes.

Project deliverables

Deliverables for the OmniCAV project are as follows:

A real-world validated simulation-based framework for certifying autonomous vehicles

Input into national and international policy and regulations on AV certification

Engagement with the public to grow trust in AV safety

Strengthening of the UK’s position in the emerging global AV ecosystem

Enabling autonomous vehicles in rural areas