What’s under the hood of your self-driving car?

Google’s self-driving car is a battery electric vehicle. Is any one powertrain type best-suited to autonomous drive technology? Martin Kahl investigates

The idea that a tech company might challenge the traditional automotive industry by launching its own car is one that has been discussed for several years. Nonetheless, the mainstream automotive industry was caught by surprise when a Silicon Valley company that has only been in existence for several years went ahead and did just that.

Google Self-Driving CarUnveiled in May 2014, the Google car was not built by a mainstream car company; and although it is called a car, other than a stop/start button, it has no steering wheel or driver controls.

Google pitched its self-driving car as a low-speed, urban use vehicle developed specifically for providing mobility to people who could not otherwise use a car independently. Significantly, it is a battery electric vehicle, and this raised an interesting question: what, if any, is the most suitable powertrain type for autonomous cars? Megatrends put this question to several automotive industry experts.

“One of the key enablers for DAS [driver assistance systems] and semi-automation has been the electrification of the actuator system, which has enabled features like electric power steering,” says Andrew Whydell, Director of Product Planning for Global Electronics at Tier 1 supplier TRW. The first lane keeping system, explains Whydell, was introduced by Lexus in 2007 in the LS460. “The reason Lexus could do it before anybody else was that it had developed a hybrid version of the car with a 42V electric power steering system. It was effectively a step down from the 300V battery pack that actually ran the car.” This meant that Lexus had an electric system powerful enough even for a full-sized luxury car, because it was essentially running off a hybrid system.

Lexus LS460
Lexus LS460

“Being able to make electric power systems powerful enough has been a barrier until the last year or so. Now that we can do it with 12V systems, we are developing solutions for the technology. With Electronic Stability Control we now have the braking, and with EPS we have the steering. This means we don’t need to have an electric car.” Furthermore, explains Whydell, the use case for electric cars adds complexity. “Electric cars are generally used for low speed city driving. There’s a natural fit between speed and the level of automated capabilities. An electric car being driven in a city environment is in one of the most challenging environments for autonomous driving anyway. It’s clear why Google has taken that approach, building up capability at low speed, but depending on how complex the situation is, it may be that a driver needs to step in or the car will just slow down if it cannot figure out what it needs to do.”

In the last issue of Megatrends magazine, we looked at BMW’s highly automated driving technology which is being developed for internal combustion engine cars driving at expressway speeds – the exact opposite of Google’s self-driving, low-speed urban electric car. Another Tier 1, Continental, is also looking at low-speed semi-autonomous driving, with the intention of alleviating drivers of the chore of driving through slow-moving stop-start city centre traffic.

Prediction-based energy management

There is a link between powertrain efficiency and autonomous drive technology, says Vincent Charles, who works in Continental’s Innovations and Technology Corporate Communications department. The key concept, he says, is prediction-based energy management. Optimised combustion engines with an increasing level of electrification make a significant contribution to sustainability, says Charles, but a major factor in any vehicle’s ultimate energy consumption is the driver. “The driver decides on speed, acceleration, deceleration and gear changes based on his experience, knowledge of the car he is driving, and the road topology. Studies show that these decisions can easily affect energy consumption by as much as 20%.”

Such a wide variation in energy consumption has led Continental to focus on highlighting to drivers the impact of their driving style on fuel consumption, and providing them with opportunities to drive the vehicle as efficiently as possible.

Charles says Continental expects to improve efficiency potential by increasing the level of automation and networking of vehicles; ultimately, vehicles will optimise energy consumption using better information about their environment and the roads they are on than any human driver with the greatest foresight could ever otherwise have. Charles envisages a scenario in which a driver stuck in stop-start traffic can sit back, relax, and let the car drive through the traffic as energy-efficiently as possible using data gathered from monitoring the surrounding environment and through communication with other vehicles.

Continental Mobility Study 2013

The supplier is not only looking at low-speed solutions; Christian Senger, Continental’s head of research for automotive electronics, recently said, “The Continental Mobility Study 2013 has shown that motorists worldwide want automated driving on the freeway. Their needs match up perfectly with the development possibilities in the upcoming years.”

As such, says Charles, “prediction-based energy management” will become increasingly important in connection with growing engine hybridisation, co-ordinating the entire flow of energy in the powertrain as well as that of the fuel and electricity energy sources, lowering system costs and consumption.

Not radically different

Autonomous vehicles will most likely appear in megacities, as a form of taxi, believes Rob Rickell, Senior Vice President, Engineering at GKN Driveline. And, he adds, due to city centre emissions regulations, such a car will almost certainly be an electric car. “The Google car will be limited to a maximum speed. In terms of the drivetrain, there may be no steering wheel, but the steering will still work on a steering axle, and I expect the majority of such cars will continue to use a central electric motor, a single speed reduction gearbox and two driveshafts with CV joints driving the front or the rear wheels, depending on the passenger compartment.”

Such a car would, from an engineering perspective, not differ significantly from what is being done today with some electric vehicles, says Rickell, adding, “It’s hard to see why autonomous cars need to be radically different.” The big difference for a fully autonomous car, says Rickell, is the lack of passenger input. “There’s no steering wheel, and braking is done autonomously. In terms of the drivetrain, the most sensible things still apply, just as they would with a conventionally-driven vehicle. You could imagine in-wheel motors, but that adds cost compared to having a central electric motor, a simple differential transmission and two driveshafts.”

The ideal combination of efficient powertrain and autonomous drive technology would be likely to involve a central electric motor, says Rickell. He’s keen to emphasise, however, that the technology for autonomous driving already exists, but that further advancements are being slowed not by technology but by legislation, and the question of how much responsibility the occupant has to intervene and stop things happening when needed.

Detection, perception, decision and control

Nissan is one of the major global OEMs to have committed to bringing autonomous drive technology to the mainstream by 2020. Indeed, it already has an autonomous car, says Jerry Hardcastle, Vice President, Nissan Technical Centre Europe. Hardcastle, who is also General Manager, Innovation and Performance Projects in Nissan’s Global Marketing and Communications Department, explains, “It’s a Leaf. We showed it at an event called Nissan 360 last summer and we’ve demonstrated it widely in Japan.” The electric powertrain, which delivers instant torque, lends itself to autonomous vehicle applications, says Hardcastle, but this does not exclude other powertrains from the autonomous car race. “When you’re developing an autonomous car, you have four steps: detection, which covers all sensors. Then perception – that is, identifying what’s going to happen next; then comes decision, followed by control. And right now, all four steps are difficult.”

An EV simplifies the control element, explains Hardcastle. “But once you’ve developed the control mechanisms, it really doesn’t matter what you’re controlling. And for us, there are other aspects. At Infiniti, we’ve got Direct Adaptive Steering, which is easier to control than standard steering. You may remove the steering wheel, but you’ve still got to turn the wheels. Using the Direct Adaptive Steering system for autonomous vehicles will help.”

Hardcastle expects autonomous drive technology to be tested in EVs because, put simply, “it’s easier”. Once tested, the technology will be adapted to hybrids and internal combustion engines (ICEs).

Adding the software simply to control an ICE requires considerable coding. “Adding that to an autonomous car just makes it more complicated. Ultimately, we will develop the control systems for any powertrain. We’ll develop the autonomous drive technology for any powertrain. And then we’ll put them together. So it’s definitely not exclusive. It’s just easier at the beginning.”

Continental Mobility Study 2013

Given that the Google car has been pitched as offering mobility to people who were unable otherwise to drive a vehicle, developing autonomous car technology for a vehicle with a combustion engine would still require the manual refuelling of that vehicle. Here, too, EV technology lends itself to autonomous driving. GKN’s Rickell agrees, adding that the most logical refuelling – or in this case, recharging – method for an autonomous vehicle would be inductive (wireless) charging, with charging pads at parking spaces and traffic lights, for example.

Eliminating the manual refuelling of a gasoline tank is not the only reason to develop an autonomous car, says Hardcastle, “but it is one of the reasons. If you go down that path, then an induction-charged electric vehicle makes sense. The car will steer itself over the induction plate and charge itself.”

Feet off, hands off, eyes off

Another reason for autonomous drive technology development is safety, and the fact that the car can assist the driver. Nissan boasts class-leading levels of ‘safety shield’ technology in its latest generation Qashqai. “It’s got back-up collision protect, it’s got lane departure warning and it’s got electronically controlled braking systems,” says Hardcastle. “All of those are there now, trying to make the car safer. But if you give it more detection, more perception, then those systems can start to intervene in more difficult conditions. And you’ll see that developing, probably, before the full autonomous car.”
Hardcastle cites MIRA’s George Gillespie, who has described the development of the autonomous car as a three-stage process: first, feet off, with the car running in cruise control; second, hands off, with the car steering itself; and finally, eyes off – at which point, the car becomes fully autonomous. “Each of those steps takes another level of technology,” grins Hardcastle.

Markus Pfefferer, a Managing Director at Ducker Worldwide, does not believe any one powertrain type is best suited to autonomous cars. “It is obvious that the OEMs will use their existing technology to gradually mature their in-house hybrid and electric solutions,” he explains.

“Semi-autonomous technology, which typically includes lane-keeping, collision avoidance and cruise control features, is already available in vehicles like the 2014 Mercedes-Benz S-Class, which has an optional Advance Driving Assistance Package, and the 2014 Infiniti Q50. Both run on a regular powertrain.”

Pfefferer expects to see a rise in the number of cars with semi-autonomous drive technology over the next three to five years, all with conventional gasoline or diesel powertrains as OEMs begin adopting semi-autonomous systems in their regular line-up. “Current experimental fully autonomous vehicles, however, are either running on hybrid, such as the Ford Fusion and Toyota Prius, or on electric powertrains, like the General Motors EN-V, Nissan’s Pivo concept and the Leaf, and Google’s car. Therefore, fully autonomous vehicles which are expected to be ready for commercial production between 2020 and 2025 may use hybrid to start with and may switch to electric powertrains as electric vehicle technology matures.”

Develop from the ground up, or bolt on to existing technology?

The Google car is so heavily reliant on the specially developed computer that manages the vehicle and its electronic systems that a traditional combustion engine would seem out of place. “We’ve been bolting things on to existing cars for a long time, and started to realise that that is very limiting in what we can do,” says Jaime Waydo, a Systems Engineer on the Google Self-Driving Cars Project, in a Google video. The Google car has been custom-built from the ground up specifically for self-driving, and Google has the advantage of stepping into the game as an outsider, with no automotive heritage binding it; OEMs and suppliers from the traditional automotive industry, on the other hand, are effectively bolting things on to existing cars. As Ducker’s Pfefferer points out: “Google does not have a history of developing powertrains, so it’s not surprising that it has chosen an electric powertrain for its prototype, giving it independence from the OEMs.”

Nissan Leaf Autonomous Drive
Nissan Leaf Autonomous Drive

The answer to the original question still eludes us, however. As so often, the short answer is: it’s application-dependent. For now, at least, semi-autonomous drive technology is being offered as an optional comfort feature; drivers are expected, or expect, to retain ultimate control. Those vehicles will spend much of their time being manually driven, and their drivers will have expectations broadly similar to drivers of manually-driven ICE-powered cars. Users of fully autonomous cars have totally different expectations; through choice or necessity, those users will be passengers only, and for them mobility, convenience and cutting edge technology are of paramount importance.

Traditional powertrains remain compatible with semi-autonomous vehicle technology, and for now also with fully autonomous vehicle technology. But the traditional automotive industry must beware: whilst it bolts new technology on to existing products, the non-traditional players are custom-building mobility solutions from the ground up. Originally we questioned the compatibility of powertrains and autonomous drive technology; perhaps the question should run deeper, and question what we really want – cars as we know them, or mobility solutions?

This article appeared in the Q3 2014 issue of Automotive Megatrends Magazine. Follow this link to download the full issue.