Woolwich car trial takes a hands off approach to driving

Palms open and fingers spread, the hands of motorist Tom Watts hover beneath the car’s steering wheel as we glide along a busy dual carriageway through Woolwich in south London. But far from being a stuntman or erratic taxi driver, Tom is a calm and responsible automotive engineer demonstrating the capabilities of the latest autonomous vehicle that steers, accelerates, and brakes all by itself. 

His participation in the journey was, it turns out, negligible; he was there in case the need for intervention arose and to provide a degree of reassurance to his passengers. Throughout the 20-minute drive around two laps of a route populated with buses, lorries, cars and cyclists Tom’s hands could be seen in real time on a computer monitor placed behind the driver’s headrest – and not once did they reach out to touch the wheel. 

“I’m still in charge of the vehicle and will intervene if there is anything I deem to be unsafe,” he explained later after our faultless ride concluded. “But there was no need; that drive was 100%.” 

Tom and his colleague Nirav Shah sat alongside him work for Nissan, which is coming to the end of a three-year advanced autonomous drive technology project called ServCity that brings together five further consortium partners including Connected Places Catapult which leads on project management and economic modelling. The Catapult is now looking to use the project to create a blueprint for how a future ‘robo-taxi’ service could be deployed in a UK city. 

So far the vehicle has driven over 1750 miles around its south London circuit and recorded more than 83,000 ‘interactions’ with other vehicles and pedestrians with, it is comforting to hear, no accidents. 

A car packed with technology

The vehicle model on test is an electric Leaf and features an array of cameras and sensors strapped to the roof and discreetly attached inside the vehicle to the front and rear. Cameras on top are used to identify other vehicles and road users, and those behind the windscreen (telescopic and wide lens) are used for checking the status of traffic lights and for lane detection. 

Camera images combine with radar technology to ensure the car safely maintains an appropriate headway to the vehicle in front – known as adaptive cruise control – and four separate Lidar sensors determine the speed and trajectory of other vehicles. In addition, two global positioning sensors on the roof interact with local phone masts to triangulate the vehicle’s position to an accuracy of less than 10mm. 

But there’s more: an antenna on the back of the car receives radio signals from 270 roadside devices placed around the local area to provide advanced warning of obstacles ahead, such as buses waiting in the nearside lane. Cameras trained on the road and the artificial intelligence provided by the infrastructure are overseen by operatives sat inside the Smart Mobility Living Lab in Woolwich from where the on-road trials begin. 

This infrastructure-to-vehicle technology provides the autonomous vehicle with early information about what’s beyond a motorist’s line of sight, allowing the car to effectively ‘see’ around corners at what is happening much further up the road. This detail is useful, it was later explained, to allow the car to prepare to change lanes in good time to help it avoid getting stuck behind parked vehicles and to smooth the flow of traffic. 

Inside the Nissan Leaf, a screen beside the dashboard displays a series of squares of various colours and sizes that appear, move around and vanish again to represent other road users in front of us. For passengers sat in the back, monitors show not just the driver’s hands (as mentioned earlier) but an aerial map of the local road network indicating our position, with small red dots showing stationary objects. 

There is also a forward-facing graphic detailing the movement of our car through the traffic, accurately depicting what could be seen through the windscreen and out of the windows to each side. 

Interacting with other traffic 

Maintaining a speed of 30mph, or just below, the autonomous car smoothly passed alongside other traffic and it was easy to forget that the vehicle was driving itself. When a delivery van suddenly changed lanes and pulled out in front of us, the car responded promptly by slowing to maintain a safe headway. 

Approaching a small roundabout, it was clear that both our vehicle and another car travelling in the opposite direction were about to continue straight ahead and safely pass one another. The on-vehicle sensors successfully detected the other car’s speed and intended direction, allowing our vehicle to press on without having to slow down. 

On our second lap of the Woolwich circuit approaching the same site, another car was already on the roundabout and about to turn in front of us. We slowed to a halt as you would expect, to allow the other vehicle to pass. But there followed a momentary hesitation before the car pulled away again – dwelling for perhaps a second or two longer than a human driver may have done before hitting the accelerator. 

However, at a pedestrian crossing further up the road, the car slowed safely to a stop when all appeared clear. Was this another example of the car demonstrating a little too much caution? Not a bit of it, as suddenly from in front of a vehicle alongside us stepped a couple of pedestrians into the road. Would a car driven by a human have spotted those people so early? 

“We have tuned the vehicle so that it is suited to the local environment and London traffic,” said co-driver Nirav, adding that it is important the car finds a balance between being overly nervous to other movements and too aggressive. It was also pointed out that the vehicle must abide by the ‘unspoken culture’ among drivers: when to hold back and when to proceed. 

At the end of the journey, we were shown inside the vehicle’s boot, filled with a sizeable electronic control unit featuring six computers. There was certainly no room for bags of shopping or a set of golf clubs, but in time it is hoped that the equipment can be scaled down to more practical dimensions. 

Making driving safer, cleaner and more inclusive 

But why should we get excited by autonomous vehicles? Nissan’s Project Manager Robert Bateman explained that the technology promises improved safety, cleaner driving, and more inclusive transport. “Over 90% of road accidents are because of human error,” he said. “Autonomous vehicles only have one job to do; they are not distracted by changing channels on the radio, eating a sandwich, or thinking about what they will be doing in the office later. Our ultimate goal is to have zero fatalities on the road.”

In terms of a cleaner environment, Robert says the autonomous vehicle will know better than humans what is happening up ahead and so change lanes at the most opportune moment, helping to reduce congestion along with harsh acceleration and braking, thereby limiting the release of harmful particulates. 

A third benefit of autonomous vehicles is around inclusivity, he adds, assisting older motorists who may appreciate a period of rest from driving on a long journey. Autonomy could also give disabled people greater independence and allow them to rely less on other people to help them get around. 

The trial in Woolwich has certainly shown what autonomous vehicles are capable of and the potential benefits on offer. Other urban locations will no doubt be looking on with interest at what has been achieved.

ServCity brings together Nissan and Connected Places Catapult with specialists from research group TRL, Hitachi Europe which looked after artificial intelligence, the University of Nottingham which considered user experiences, and SBD Automotive which led on developing a smartphone app to help people book autonomous vehicles as part of a taxi service. 

The programme follows a three year HumanDrive project led by Nissan which trialled an autonomous vehicle on country lanes and culminated in a 230 mile autonomous journey from Cranfield in Bedfordshire to Sunderland. 

Read more about the ServCity project.