The first live trials of an autonomous vehicle fleet began in October in Oxford, with other tests set to get underway in London and a third major British city. Under the aegis of UK government R&D effort Project Endeavor, live on-road trials will run until Autumn 2021. They will demonstrate Level 4 autonomous driving in a range of urban environments and engage with local authorities and communities to prepare for the future of widespread autonomous vehicles in traffic.
In Level 4 autonomous driving, “High Automation,” the vehicle performs all driving tasks under specific circumstances. Geofencing is required, and hum,an override is an option.
After intensive advanced simulations using AI and deep-fake technology, a fleet of six Ford Mondeos, equipped by Oxford-based firm Oxbotica will make continuous ride-sharing voyages on a nine-mile round trip route rom Oxford Parkway station to Oxford’s main train station. Trials will run at all times of day and night, in a range of traffic and weather conditions.
Oxbotica’s technology combines radar, cameras and LiDAR and radar in a navigation and perception system controlled by Oxbotica’s software stack Selenium.
The company states that its multi-module localization system enables autonomy in on-road and off-road locations, and is not reliant on any external infrastructure. It can operate on its own or be fused with other location services driven by GPS, lidar or laser vision as part of Oxbotica’s modular and integrated approach.
Deep Fake
Oxbotica developed and deployed a deepfake technology that can generate thousands of photo-realistic images in minutes, helping to expose its autonomous vehicles to the near-infinite variations of the same situation, without potentially hazardous real-world testing.
Deepfaking employs deep learning artificial intelligence (AI) to generate fake photo-realistic images. Sophisticated algorithms reproduce the same scene in poor weather or adverse conditions, and subject its vehicles to rare occurrences.
The technology can reverse road signage or “class switch,” replacing one object with another. It can change lighting to simulate a different time of the day or season of the year, ensuring thatshadows or reflections appear exactly as they should. It uses the synthetic images to teach its software, producing thousands of accurately-labelled, true-to-life experiences and rehearsals.
Paul Newman, Co-Founder and CTO at Oxbotica, said: “Using deepfakes is an incredible opportunity for us to increase the speed and efficiency of safely bringing autonomy to any vehicle in any environment – a central focus of our Universal Autonomy vision. What we’re really doing here is training our AI to produce a syllabus for other AIs to learn from. It’s the equivalent of giving someone a fishing rod rather than a fish. It offers remarkable scaling opportunities.
Other Project Endeavor partners include urban innovator DG Cities and Immense, a transport simulation company; also the Transport Research Laboratory, the British Standards Institution and Oxfordshire County Council.
The last trio will focus on the development of a new safety assurance assessment scheme against PAS 1881 standard for public autonomous trials, helping inspire trust and define a consistent approach to safety that will enable future deployments to happen efficiently without slowing down the rate of innovation.
