CAR (UK)

Autonomy in the UK: it’s here

The gent above is purely a minder – he is being driven around Milton Keynes by a computer. And sitting behind him is a nervous Phil McNamara

-

THE RANGE Rover Sport brakes suddenly and aggressive­ly a good four car-lengths from the roundabout entrance, then creeps forward. With the way clear, the SUV takes the plunge and pulls out, jerkily steering through the bend and then taking the first exit onto another dual carriagewa­y. Instantly it lurches sideways into the right-hand lane; I find myself twisting to check the blindspot. It’s thankfully empty, so – unscathed – we continue to our destinatio­n, Milton Keynes Central. I’ve had smoother rides to the station, but this trip is uniquely different: there’s no human driver. This is a demonstrat­ion, on open public roads, of the fruits of a three-year project to develop autonomous and connected car technology in the UK.

Jaguar Land Rover safety driver Jim O’Donoghue is in the driver’s seat but he’s barely ‘at’ the wheel: he only intervenes once during the four-mile journey, not on national speed limit stretches but when the robot Range Rover gets foxed by a car park junction.

This prototype has, to use the scientific jargon, Level 4 autonomous capability, meaning it can fully drive itself in a ‘geofenced’ area JLR researcher­s have mapped in detail. Level 5 – unconditio­nally autonomous vehicles without steering wheels – is the theoretica­l maximum. Audi is currently trapped in Groundhog Day, trying to get the Level 3 autonomous system it revealed last4

‘The selfdrivin­g Range Rover behaves like a learner with limited mechanical sympathy’

summer – which can only self-drive at up to 37mph in motorway congestion – approved for commercial use on the A8. That could take another 12-18 months, says Audi. This free-Range Sport has a far wider operating window, thanks to its bespoke ‘thinking’ software and ‘seeing’ hardware costing around £250,000. It’s a pioneer. And, onboard for 15 minutes, so am I. The demonstrat­ion starts with a clunking noise: from the passenger seat, autonomous software operator Niharika Bhargava has armed the system, which audibly takes control of the electric power-assisted steering. Electronic control units for brakes, accelerato­r and other systems have also gone self-sufficient. ‘The computer correspond­s with all the sensors, [creating a] perception from fusing their informatio­n on whether something’s in front, the curvature of the road, map position, traffic lights and so on. It sends that to a gateway which translates this autonomous language into car language, which sends commands to the ECUs,’ explains Bhargava.

How good is an autonomous driver?

Not very On the move, the self-driving Range Rover Sport behaves like a hesitant learner with limited mechanical sympathy: braking and throttle inputs are consistent only in their ham-footedness.

From the age it takes to pull away at traffic lights, I’m not convinced it would pass the 20-metre eyesight assessment that precedes a driving test. And while it’s magical watching the steering wheel twirl from side to side as if propelled by a poltergeis­t, it’s appropriat­ely scary too, as the car veers enthusiast­ically towards one bend’s kerb before yanking on more lock to avoid contact. Silky smooth it is not.

But then I’m judging it against a competent human driver. This is a scratch prototype; it doesn’t tap into existing cruise control and brake assist systems benefittin­g from years of refinement. ‘For Level 4, to get the architectu­re to be robust and for the car to control itself, we’ve had to go back to basics and teach it to drive,’ says O’Donoghue. ‘So it errs on the side of caution and is a bit lumpy in places.’

Indisputab­ly. But it’s still a mind-boggling achievemen­t, even allowing for the demo’s structured nature: JLR has given this Range Rover Sport the ability to see its surroundin­gs, compute appropriat­e moves and then control the vehicle systems to enact them and self-pilot the SUV to its destinatio­n. All while dicing with random human behaviour on foot or in cars. Small roundabout­s – where logic has to compete against hard-charging, non-indicating humans – have proven tough to navigate, and JLR has learned that sunlight can obscure green traffic lights for the camera, leading to hesitancy. It’s all vital field research into sensor capability and artificial intelligen­ce, made possible by the focus provided by the UK Autodrive study.

‘Any actuation on a car is a big challenge, and that’s before the safety aspect’ Mark Cund, Jaguar Land Rover

What has UK Autodrive done for us?

Tim Armitage has the satisfied air of Hannibal at the end of an A-Team episode: a plan made four years ago has come together. Armitage works for Arup, the global planning consultant­s, which mobilised 15 partners from the car industry, tech sector, academia, local authoritie­s and the legal world, to develop and test autonomous and connected vehicle technology, and examine its impact on civic infrastruc­ture and society. The government bought into the UK Autodrive consortium’s vision with £20 million of funding, and looped itself into the feedback process. Ministers dream of making the UK a hub for connected and autonomous vehicle developmen­t.

The autonomous Range Sport is just one technology testbed. The partners agree that one of the biggest breakthrou­ghs is connecting different cars from three different manufactur­ers – Ford, Tata and Jaguar – as well as linking them to highways infrastruc­ture, to enable the sharing of informatio­n to boost safety and efficiency.

The most futuristic-looking element is the egg-shaped, self-driving pods, designed to ferry people or cargo short distances. These real-world Johnny Cabs from sci-fi movie

Total Recall are closest to commercial introducti­on. Armitage reflects on the project’s biggest outcomes. ‘I think UK Autodrive has proven that the technology will work. It still needs to be refined and production­ised, and it needs to be much, much more affordable, but the technology will work. The next big task is to let people experience that technology, and work out how best to use it.’

Two cities – Coventry and Milton Keynes – joined the consortium as the playground for on-road testing. Coventry plays the role of traditiona­l British city: tight, complex and organicall­y spawned, with a ring road that demands rapid decision-making due to its flurry of junctions. Milton Keynes, with its metronomic grid structure, provides fast roads and myriad roundabout­s for testing. The wide

‘UK Autodrive has proved that the technology will work, but it needs to be re ined’ Tim Armitage, project leader

walkways of Milton Keynes proved ideal for running the driverless pods; Coventry’s council equipped six traffic lights with transponde­rs, to allow them to communicat­e their change sequence to oncoming connected cars.

The trials have highlighte­d some infrastruc­ture changes needed to unleash the potential of the autonomous and connected age. Airports, railway stations and hotels could ditch vast car parks for kerb management – areas where autonomous vehicles (AVs) can drop off and pick up. And connected traffic lights, dynamic road signs and real-time parking updates would be great. But who pays?

‘It’s a chicken and egg situation,’ says Ford’s automated driving supervisor Christian Ress. ‘Car makers will ask councils when they’ll put in communicat­ors, and councils will ask car makers when they’ll equip the cars. UK Autodrive is good because we can work on a common approach to bring this to market.’

And how might AVs affect traffic flow? More than 2000 specially installed sensors have logged the traffic in Milton Keynes to create a computer model, with AVs gradually introduced to simulate their impact. Initially they will slow things down – because of their adherence to speed limits and aversion to tailgating – but when they reach a critical mass, congestion should decrease.

How far off is that? Although the consortium has made progress with government, easing red tape around on-road testing and raising awareness of the cyber security threat, none of the participan­ts sounds confident about getting the tech type-approved. After all, the regulators are in the invidious position of having to green-light an all-new technology, and they need to conceive a new testing method to triple-check it’s safe. Audi’s experience shows it takes years.

David Hudson, Tata’s group chief engineer, says the Indian car maker joined in to kickstart its autonomous journey, and generate real data to build on in the lab: ‘Simulation is a critical part of future adoption. If you don’t get good at simulating use cases, it will be impossible to validate in future. If you can’t pre-prove, you might have to drive something like eight billion test miles to cover all the use cases!’

And my test ride shows there’s a long way to go before the autonomous technology is sufficient­ly refined and failsafe.

‘We’ve shown that Level 4 vehicles can operate autonomous­ly in geographic­ally specified areas. Commercial­ly these vehicles will be available in the next five to six years,’ says Arup’s Tim Armitage. And Level 5 autonomy? ‘A vehicle that could go absolutely anywhere – that’s decades away.’

Jumping junctions with a psychic Mondeo

We’re in a car park, playing a game of chicken, blind. Our Ford Mondeo hybrid approaches an intersecti­on obscured by parked cars. Although it can’t be seen, a Tata Hexa SUV is also approachin­g from the side: we’re on a collision course. What can you do but have faith in the technology? Driver Lovedeep Brar consults his instrument binnacle (below, left), which graphicall­y signals an approachin­g vehicle. But it also shows a green band on the digital speedo, between 27 and 50km/h – if Brar stays within that range, the Mondeo should pass safely. We sail into the junction and… nothing untoward happens; the Hexa has stopped.

It’s all thanks to dedicated short-range communicat­ions (DSRC). Wi-fi routers in each car emit radio signals every millisecon­d, allowing the cars to process informatio­n about connected objects in the vicinity. With the Mondeo having right of way, the binnacle advises us to proceed, while the Tata driver is instructed to stop to avoid a potential collision.

Here, the car-to-car communicat­ions are advisory, but in the autonomous future a connected car could automatica­lly brake if it’s in danger. There’s an efficiency benefit too: the researcher­s demonstrat­ed ‘intersecti­on priority management’, where the Mondeo, Hexa and a following Jaguar F-Pace all criss-crossed the junction like the Red Arrows, their speed and sequence determined from the DSRC. Time and fuel would be saved by keeping connected cars flowing. Ford, Tata and Jaguar are chuffed that they’ve got the EU communicat­ions protocol to work across their cars. The benefits are obvious, and rolling these features out as driver aids is far less fanciful than autonomy. But both have the same Achilles’ heel: adoption needs to be fast and far-reaching to have a profound impact. In the meantime, it’s back to the research labs, proving grounds and regulators, as the drive for connected and autonomous cars continues.

‘We’ve tested connected tech; now it’s over to product developmen­t’ Christian Ress, Ford

 ??  ??
 ??  ?? WHEEL SPEEDThis humble sensor is as valuable as any of the highertech parts. It measures how fast the wheels are turning to provide exact speed and distance data. By comparing this with other data, the computer gets an idea of slip, and thereforer­oad conditions.
WHEEL SPEEDThis humble sensor is as valuable as any of the highertech parts. It measures how fast the wheels are turning to provide exact speed and distance data. By comparing this with other data, the computer gets an idea of slip, and thereforer­oad conditions.
 ??  ?? Just popping down to the tip, dear. Got a load of electrical junk tothrow out
Just popping down to the tip, dear. Got a load of electrical junk tothrow out
 ??  ??
 ??  ??
 ??  ?? After four years of work, Tim Armitage and his teams have proved the technology can operate in the real world. Now they need to bring the costs down
After four years of work, Tim Armitage and his teams have proved the technology can operate in the real world. Now they need to bring the costs down
 ??  ??
 ??  ?? Expect to see more of these humming around airports, hospitals and city centres – soon without any drivers
Expect to see more of these humming around airports, hospitals and city centres – soon without any drivers
 ??  ??
 ??  ?? Green area of speedo shows how fast you need to go to avoid acollision
Green area of speedo shows how fast you need to go to avoid acollision
 ??  ??

Newspapers in English

Newspapers from United Kingdom