AVs on a Long Road to Consumer Acceptance

Only time will tell when it comes to building confidence in autonomous vehicles despite a recent pilot’s claim that most consumers are happy riding a driverless vehicle.

That’s the admission by David Hynd, chief scientist for the Transport Research Laboratory (TRL) whose recent StreetWise automated driving trials in London claims a huge backing by consumer who tried the technology. The project is part-funded by the Centre for Connected and Autonomous Vehicles (CCAV), delivered in partnership with Innovate UK and is part of the UK government’s £100M ($131M) Intelligent Mobility Fund, supporting the Future of Mobility Grand Challenge.

It involved an invitation-only group of participants to experience being driven autonomously on a busy, fixed 13-mile route in London. This included shared tramways, a variety of roundabouts, cyclists, pedestrians, T-junctions, signalised pedestrian crossings and a wide variety of vulnerable road users. Not only did Five’s self-driving system have to be high-functioning but one of the most rigorous safety cases had to be developed and monitored to assure safe operation.

Naturally, with the invite-only stipulation, the results of the consumer feedback must be take with a large pinch of salt particularly in the light of previous studies that show a large amount of distrust among consumers.

Nonetheless, of those who took part in the trials, 96% rated their overall journey experience as positive to very positive, with 86% stating that their expectations had been exceeded. Factors that drove these results included the system’s ability to keep a safe distance, perceive and maneuver safely around obstacles and hazards, drive like a human and manage roundabouts – common in Europe but almost entirely absent in the US – as well as participants’ trust in the diligence and professionalism of the safety driver in each vehicle.

However, Hynd admitted the technology has a long way to go to offer perfect safety in all conditions. Indeed on the data TU-Automotive has seen, if last year’s data released by Waymo into the need of its driverless vehicles to receive human intervention every 5,996 miles was extrapolated to the road mileage recorded on US highways of 3.22Trn, the annual fatalities could run into the millions instead of the 45,000 or so committed in human derived collisions.

Hynd defended the current test intervention rate, saying: “One of the things that somebody might be doing is picking challenging circumstances in order to develop their technology. In other words, you are going to except to take over because you are trying to learn, improve and develop your system. So that should be seen as a measurement on a development vehicle not a metric on a vehicle you would release to market.”

A comforting thought, were it not for the fact the bulk of AV testing, so far, has been in quiet rural Arizona locations or in slow-speed urban environments. Yet, Hynd says this is part of a slow transition towards automated technology.

He said: “There is a regulation currently going through the UN in Geneva and it’s called ALKS, or augmented lane keeping system, and it is essentially a motorway traffic-jam assist at lower speeds on dedicated roads. So if you constrain yourself to an operational design domain (ODD) that is relatively straight forward like this, there is potential to have a much lower rate of failure than in more complex environments. You build over time and increase the complexity and you increase the range of the ODD over time.”

Since our interview the state of Michigan has sided with any automakers’ long held belief that AVs require their own dedicated highways to ensure that safe ODD to which Hynd refers. He said: “You could have zones that where only automated vehicles are operated in an area analogous to a bus route and say ‘OK, that’s the AV route’ as a way of starting and demonstrating the safety. This will also build confidence in people to use the technology to bring us to a good place.”

Naturally, Hynd says the industry must keep future ambitions in sight to improve the technology with organic growth. He said: “The end goal is that vehicles would be operated on the open road and if we think about the proportion of the current collisions and the injuries and deaths that occur owing to speeding, drunk driving, distraction, inattention which construe the largest population casualty size, the automated vehicles [when they are ready to do so] clearly won’t do those things so you could reduce the casualty population by that size. However, they may also introduce other collisions and that’s the thing that needs to be understood properly. Where’s that balance?”

He pointed to the development where the Waymo One driverless ride-hailing service is currently operating in the Phoenix East Valley region of Arizona as an example of how the technology is progressing. Hynd said: “Waymo operating in very controlled areas, without a driver. Yet, just a year ago, they could not do that because the level of development of the system wasn’t adequate to do that. Now they can do that safely in those zones and those zones will expand over time. I think it’s a completely appropriate development sequence to expand those areas of usage without the safety driver as you get to that level of competence with the vehicle.”

— Paul Myles is a seasoned automotive journalist based in London. Follow him on Twitter @Paulmyles_