‘Affordable autonomous electric cars in 2-5 years,’ say Cambridge Consultants
The UK motorist will be able to buy affordable electric cars with self-driving capabilities before 2025, and probably by 2022, says the head of transport and infrastructure at Cambridge Consultants, Thomas Carmody.
According to Mr Carmody, the technology required is almost up to speed, but the legislation – not least insurance – has yet to be agreed. In addition, the sedate pace of change in the car industry isn’t helping.
“The automotive industry hasn’t seen this type of change before,” Thomas explains from the Science Park headquarters of the company where, it seems, every employee is a genius. “We’re asking for changes in a couple of years but, in time, it will be obvious that robots will be safer than humans.”
But with 1.35 million humans killed on the roads every year, why are so many convinced that robot cars are dangerous? Maybe the car trade hasn’t been pushing for change quickly enough – and time is running out.
“Automatic windscreen wipers, seat belts – it took years to adopt that stuff,” notes Thomas, adding that “there’s no one blueprint for building a self-driving system”.
The race to design a definitive self-driving system plays well for Cambridge Consultants.
“As a company Cambridge Consultants looks for markets that are encountering disruption,” says Thomas. “Last year automotive was the third largest R&D spend in the world – self-driving and electric is happening all at once is why that is, so all suppliers are fretting that they’re not up to speed.”
Advanced driver-assisted systems (ADAS) is the fastest-growing technology segment in the automotive market, worth an estimated $24billion in 2018 and predicted to reach $92billion by 2025. And there has been progress, of course – amazing progress - and a de facto standard for autonomous vehicle technology is getting close.
“The nuts and bolts of the system are fundamental,” says Thomas Carmody, “but the value proposition is because we understand the system – my technical colleagues understand how the whole system should run. We are a technical authority, so we understand all the components of an ADAS system. We have all the skills to build a complete driverless system for a customer.
“We provide IP technology ownership to the customer so it doesn’t look any different from any other technology they develop in-house.”
Cambridge Consultants may have six to eight AV projects running at any given time. One of the technologies the Science Park-based company has developed is Sharpwave.
“A robot driver depends on good information coming in from the cameras even if it’s foggy, raining, or snowing – it needs a clear picture of what’s coming. To do this we interpreted the effects of the weather, so for instance if it’s snowing, the software interprets the effects of bad weather, including any dirt on the camera, and it removes that and gives the robot driver a clearer picture which is needed to keep the car in its lane, at the right speed and seeing the signs.
“The software is one part of a self-driving system. What we wanted to do is remove the distortion of the system, because you need the adaptability to interpret distortion. Cameras, radar and LIDAR are accepted as the three ingredients you need, plus ultra-sonics for a complete capability of driving these elements, and an autonomous stack.”
That every car maker is developing a different system increases demand for Cambridge Consultants’ services.
“Tesla, for instance, is different because it didn’t like LIDAR,” Thomas says. Tesla uses radar, GPS, maps and other cameras and sensors. Indeed the luxury end of the car market - Volvo, Porsche, and BMW - does offer self-driving capabilities.
“The trick is to get the technology to a price point where you can get the technology into more Nissans, Vauxhalls and Fords, so we ask ‘how can we use fewer sensors?’ or ‘how do we reduce battery usage?’.
“We work project by project, and the projects tend to be time-critical. We develop technology often faster than the companies can do themselves. It’s breakthrough technology, so we receive assignments: typically the customer wants something it can do in two years to be done in one year.
“The autonomous technology era is a great process. I’d say first of all we’ll see taxis without drivers. The technology is so expensive it only works financially if it’s highly utilised. Waymo [formerly the Google self-driving car project] is already driving people around in self-driving taxis in California.”
In Cambridge, FiveAI is developing a service like this, which it has been testing on roads, albeit with human drivers still ready to take over.
“When you buy your next vehicle it’ll probably be electric in a couple of years’ time, even to the point where the technology is taking over the driving in some situations, such as heavy traffic or on the motorway,” Thomas continues. “You’d switch into auto mode. Regulation is one of the reasons why progress is slow. Insurance requirements mean you have to have complete control of the vehicle at all times and be able to take over at any point.”
As to the key question of how much a robot car depends on cloud-based systems, and how much is based in the vehicle, Thomas says: “Our focus is to deliver technology on to the device, where you have limited power and cost constraints. To achieve that ambition you have a relationship between the edge and the cloud, but it needs to be very robust.”
In terms of the driverless vehicle age, we’re after the point at which the car was built – German inventor Karl Benz patented his Benz Patent-Motorwagen in 1886 – but before the arrival of the Ford Model T in 1908. So not long to go and, with Cambridge Consultants behind the wheel, transport is fast becoming an unexpectedly exciting topic.