Home » Podcasts » Lunch with Leon » Lunch with Leon episode 34 – Rav Babbra and Chess Stetson
In a trans-Atlantic lunch – with one guest in California and the other in the UK – Leon Daniels meets with dRISK.
Chatting with CEO Chess Stetson in California and Business Development/Programme Manager Rav Babbra in the UK, they talk about their work with the government-owned Centre for Connected and Autonomous Vehicles (C-CAV).
We hear how they have been tasked by the UK government, through the C-CAVS to build the ‘ultimate driving test’ – not for human drivers, but to test self-driving cars.
Their test will determine that autonomous vehicles (AVs) are not simply ‘as safe’ as those driven by humans, but much safer than human drivers.
We learn that the test is being used with current AV technologies, to establish how we can get to a ‘vision zero’ (for accidents) in the UK.
Leon asks why it is acceptable that 1,500 people a year in the UK are killed by human drivers, yet critics say that AVs must have zero accidents?
He discusses why people are not as outranged by existing UK road deaths as they should be, given that 1,500 deaths a year wouldn’t be acceptable in industry or commerce, for example.
Chess opines that the UK’s ‘zero vision’ towards accidents is a “World-leading vision and a much more holistic approach to safety than many other places.”
He explains how the driving test for AVs will be used by the DVLA/DVSA when an AV is to be certified for use in the UK.
Like the UK driving test for humans, the AV version will present a range of scenarios to test the AV.
We learn how, using Transport for London (TfL)’s 1,000 traffic CCTV cameras, dRISK has been capturing ‘edge cases’ – high-risk incidents – and re-creating them in the simulation environment to see what the AV will do.
And, he poses the oft-asked questions: Will an AV stop for a child running out from behind an ice-cream van, or will it kill it? Other questions discussed include telling the difference between a real human and a dummy.
Chess explains why an AV will be able to recognise that a child dressed in a costume as a green traffic light, is recognised as a child crossing the road, not a green traffic light, even though it’s a situation the AV has never come across before.
“AVs predict what’s going to happen, not what has happened – and they can do that now already – for example, how to deal with the guy in the chicken suit at a level crossing,” he says.
The wide-ranging conversation turns to the ‘naked highways’ – ones without physical traffic signs and traffic lights – as the AV ‘talks’ with the infrastructure and receives instructions.
The thorny subject of cyclists and their interaction with AVs is mulled over. Says Chess “Cyclists and human drivers never get on; AVs will be able to negotiate their way around them.”
Then they look at how automation will develop. Will it be a progressive change, or an overnight switch from manually-driven cars to AVs? This leads onto to the increasing automation of existing new vehicles, with systems such as lane keeping and emergency braking already mandatory.
The move to full automation in commercial vehciles is examined, and the reasons for hub-to-hub trunking with trucks being the most likely first-adopters.
The wider applications for automation are explored, with its use to de-risk the entire transport network.
We discover that already dRISK’s software is used to re-route human driven vehicles to reduce risk. What would be the positive effect on fleet managers if by rerouting, they could reduce their accident claims by 10% in a year? How artificial intelligence (AI) uses data to plan routes that are safest at certain times of day, is explained.
Rav discusses how the integration of AI into transport is no pipe dream, and talks about how dRISK re-purposed its ‘object detection’ software for Covid, to take feeds from TfL’s roadside CCTV cameras.
In this new guise, AI counted people and vehicles, to see the effects of lockdown and monitor social distancing. The results are fed back to councils which can establish if the measures it put in place, for example barriers for pavement widening or space for tables and cafes, are working or not. It can also identify hotspots, for example at pedestrian crossings, where reducing the wait time reduces pedestrian congestion.
The conversation concludes with Rav talking about two surveys that are currently open about fleet managers’ opinions on the risks of vehicle routing, and the public’s opinion on AVs.
Visit the ITT Hub 2022 show site