Autonomous car expert says his enthusiasm for the technology was seen as ‘misguided’

Prof Nick Reed, TRL Academy Director
Prof Nick Reed, TRL Academy Director

One of the UK’s leading experts in the field of autonomous cars says his enthusiasm for the technology was seen as premature and misguided by some people.

In an exclusive interview with Robotics and Automation News, Professor Nick Reed, academy director at Transport Research Laboratory (TRL), says he has long believed automation was the way to go.

“Technology moves on at a rapid pace and there were certainly some commentators in the early days who felt that my enthusiasm for vehicle automation as an important topic of research was perhaps a little premature, or even misguided,” says Prof Reed.

Certainly there were some who viewed early versions of Google’s self-driving experiment with incredulity, but most would have thought of such vehicles’ adoption as a matter of personal choice.

However, as the technology develops and the market takes shape, particularly with the entrance of other tech companies and the big global automakers, the argument has shifted from “if” to “how”, not even “when”.

That driverless cars will take to the road in large numbers within the next few years is widely accepted. What is on the minds of many people is question of whether they will be safe.

Prof Reed is confident. “For me, with a background in studying all those factors that can impair driver performance, it was obvious that we can address many of them by automating the driving task.

“This is backed up by road collision statistics, which show that around 95 per cent of crashes have human error as a contributory factor.”

He indicates that, in his experience, driverless cars are so safe they’re actually quite boring.

“I have now been in many automated vehicles and, once the initial excitement of the vehicle driving by itself has subsided, the journey experience is relatively tame – and that’s exactly how it should be. Once passengers learn to trust and accept that the vehicle will do the right things, they can relax and engage in other tasks.”

Prof Reed joined what’s called the human factors and simulation group at TRL more than 10 years ago after his post-doctoral work in visual perception at the University of Oxford.

Prof Reed’s post-doctoral work was on the science behind how a cricket fielder knows where and how fast to run in order to be in the right place at the right time to catch a ball hit high in the air
Prof Reed’s post-doctoral work was on the science behind how a cricket fielder knows where and how fast to run in order to be in the right place at the right time to catch a ball hit high in the air

“My work on visual perception was about how humans use visual information to intercept projectile trajectories,” says Prof Reed. “Put more simply, this is how a cricket fielder knows where and how fast to run in order to be in the right place at the right time to catch a ball hit high in the air.

“We made some significant progress on this topic, including being published in Nature, with the ultimate aim to understand how the brain processes complex information and produces actions.

“However, probably the part of my work most relevant to TRL came during my time as a post-doctoral researcher, where we started to investigate how artificial neural networks could be taught to act in the same way as humans in the context of this behaviour.

“This ability to take complex datasets and enable a machine to learn how to respond successfully is the essence of machine learning, a concept that will play an ever increasing role in our lives, from facial recognition to naturalistic speech processing through to self-driving cars.”

One person can only do so much, but Prof Reed’s skills and knowledge in visual perception will almost certainly prove crucial in the development of autonomous car technology. Recently the European car safety watchdog – Euro New Car Assessment Programme (Euro NCAP) – has been conducting a number of tests into how autonomous car systems “see” pedestrians.

Euro NCAP’s secretary general, Dr Michiel van Ratingen, says: “These new tests are the first in the world to assess highly automated vehicle features and driver assistance systems from the pedestrian’s perspective.

“Many new cars now offer some form of AEB [autonomous emergency braking] system that can help prevent car-to-car collisions, but only some are also able to detect pedestrians. By checking the results on Euro NCAP’s website, consumers will be able to verify manufacturers’ safety claims and choose the right AEB option.”

Robot vision: Mine eyes have seen the glory of three dimensions

At TRL, Prof Reed has led a wide range of research projects using “full mission, high fidelity car and truck simulators”, which sound like several orders of magnitude more sophisticated than Gran Turismo on PlayStation and a whole lot more serious.

“The DigiCar simulator at TRL consists of a real car – a Honda Civic – surrounded by display screens,” explains Prof Reed. “High-definition images of the simulated road scenes are projected onto these screens. The simulator is fully interactive, so steering and pedal inputs made by the driver cause the vehicle to behave as it would in the real world. The scene on the screens then updates accordingly.

“We define the exact requirements of the 3D environment in which the simulator is to be driven: is it a motorway, urban or rural road; what signs, line markings, buildings should be present? And we create the traffic scenarios: what vehicles, cyclists and pedestrians should be present; how should they behave?

“We primarily use the simulator in three ways,” he continues. “Firstly, we look at changes to the driving environment. For example, TRL has completed numerous studies for Highways England to investigate how drivers adapt to motorways where no hard shoulder is present.

“Secondly, we look at changes to the vehicle. For example, we have studied driver behaviour with adaptive headlights, adaptive cruise control, and are now looking at increasing levels of automation of the driving task.

“Thirdly, we study factors that affect the driver directly, including fatigue, alcohol, drugs and distractions.

“As a result, a combination of research skills goes into the development, delivery, analysis and interpretation of simulator studies. These include programming, engineering, psychology, human factors and statistics. It is very rewarding to know that, by applying our scientific abilities, our work has helped save many lives and prevented tragic road crashes.”

His consistent work in the field of vehicle technology, and his understanding and expertise in vehicle automation, led to Prof Reed being asked to provide technical leadership to the Greenwich Automated Transport Environment (see video, above).

The-O2
Greenwich is home to the O2 Arena and the Meridian Line

The local government in Greenwich, a borough in south-east London, has in recent years enthusiastically adopted new technology, having installed an Internet of Things (IoT) network in the area to make such things as rubbish collection more efficient. Small IoT devices in bins can let the council know when each and every bin needs to be emptied. And an IoT network may end up playing a crucial role in an integrated transport system of the future which includes driverless cars, all connected to other things such as traffic lights and so on.

Looks like Greenwich has a plan. And so does the UK central government it seems. Secretary of State for Business, Innovation and Skills Sajid Javid recently announced the awarding of a £5.5 million grant to help fund further development of driverless technology. The money, distributed through InnovateUK, will see driverless technology trialled in real-world conditions on roads in Greenwich.

The three-year project will be run by a consortium of companies – dubbed MOVE_UK – which includes TRL, Bosch, Jaguar Land Rover, Direct Line Group, The Floow and the Royal Borough of Greenwich.

Responding to the government announcement, Rob Wallis, CEO of TRL, said: “TRL is building a strong, reputable portfolio of UK based projects in vehicle automation, and this is another great example of a ground-breaking project in this area. By creating a unique evidence base for automated driving systems, we will not only help to develop and speed up validation of these systems in the UK, but also guide future thinking around the development of virtual and physical testing approaches for years to come.”

But given that the government has had to step in to jump-start the autonomous car market in the UK, having already announced it has set aside £100 million to fund intelligent transport technologies, isn’t this evidence of the UK lagging behind other advanced economies in this crucial emerging technology and market?

Prof Reed disagrees. “I attend many international meetings on the topic of automated vehicles and it is pleasing to see that other nations are envious of the UK’s commitment to this topic.

“In particular, we hear positive comments about the development of the Department for Transport’s code of practice for testing automated vehicles, to which TRL contributed, and the Innovate UK funding to support trials in the UK.

“Yes, there is a huge commercial opportunity and automated vehicles will be developed irrespective of the UK government’s efforts. However, its support means that the safety, efficiency and productivity benefits will be enjoyed in the UK sooner.”

The government has already allocated £20 million of funds to autonomous vehicle projects and is planning a £6 billion science budget around the technology, possibly believing its own hype that the intelligent mobility market is set to be worth £900 billion by 2025.

Were it a passing fad then perhaps such government support would not be made available. But the general consensus is that it is no longer a matter of debate whether it is possible that autonomous cars will be a real component in an integrated intelligent transport system of the future – it’s a certainty. In fact, Prof Reed suggests that future generations will think of people who actually drive cars as something of a dangerous anachronism, one with a tragic history.

driverless car interior
It’s like having your own invisible chauffeur. Future generations might not drive any vehicles at all

“I am confident that, eventually, future generations will look back at driving with astonishment, wondering why we were prepared to let someone die every 25 seconds globally due to road crashes – primarily caused by human error,” says Prof Reed.

“However, this is many years away, as there are significant hurdles to overcome before we reach the situation where human driving can be consigned to history. These include tackling severe weather conditions and navigating the legal issues around the operation of automated vehicles.

“I would like to see an automated transport system that is safe, clean, cheap, accessible and equitable. The role of the government is going to be crucial in ensuring transport networks meet the needs of wider society and not just those of corporations offering automated vehicle services.”

Considering how fast the autonomous vehicle and intelligent transport systems market is moving at the moment, it may prove a huge challenge to government to manage the development in a way that balances the corporations’ greed for profit and the consumers’ need for convenience and safety.

The UK government has released a set of guidelines for testing driverless cars on the road, but a set of laws governing their actual – and almost inevitable – use on the roads is still some way off.

In the US, the discussion about the legal implications of autonomous cars has started somewhat ominously, with the National Highway Traffic Safety Administration (NHTSA) indicating that it has given in to Google’s argument that the artificial intelligence in its car would be considered human for legal purposes.

In a letter in response to Google’s argument, NHTSA Chief Counsel Paul Hemmersbaugh said: “NHTSA will interpret ‘driver’ in the context of Google’s described motor vehicle design as referring to the (self-driving system), and not to any of the vehicle occupants.

“We agree with Google its (self-driving car) will not have a ‘driver’ in the traditional sense that vehicles have had drivers during the last more than one hundred years.”

There goes humanity. After having evolved for hundreds of thousands of years, developing farming, civilisations, religions and philosophies along the way, it’s the legal system we developed that looks like it will finally sell us out to the machines. Science, of course, is oblivious to everything except itself.

But to be fair, the US legal system considers corporations in the same way, and treats them as human too. It’s debatable which is more questionable.