Once upon a time, our vision of the car of the future was a flying car. We have largely given up on that and replaced it with a vision of a self-driving robotic car, one that lets us be as distracted as we wish as we motor—on the ground—to our destination. The difference in these visions is that the robotic car is actually practical and could be deployed in a little more than a decade from now, according to experts at Continental.
The evidence they point to is that many precursor technologies such as electronic stability control (ESC), lane departure warning (LDW), adaptive cruise control (ACC), and autonomous emergency braking (AEB) systems are fielded and working today. While offered as options—and current take rates are low—the influence of the insurance industry with car safety ratings such as the European New Car Assessment Programs (Euro NCAP) will drive adoption of these advanced driver assistance systems (ADASs). This emphasis on safety through ADAS and technology will inevitably lead to robotic cars, in their view.
“Technologies such as AEBS will become standard on 100% of new cars, at least in Europe, by 2020,” predicted Markus Schneider, Head of Business Development & CBS/ADAS Business Unit for Continental. He was speaking at a company press conference held in Alzenau, Germany in October. This will be driven by the desire of OEMs to achieve Euro NCAP five-star crash ratings. He was speaking to the fact that in 2014 features such as LDW and AEB systems will begin to be phased into Euro NCAP’s rating system. “We saw similar [adoption] with ESC. ESC first became part of the Euro NCAP rating and after about 7 years it became legislated [on all new cars],” he said. “We think we will see the same thing with [current] ADAS systems.”
Insurance industry weighs in
The roadmap to automation for safety is real. Matthew Avery, Head of Research at the U.K. insurance industry facility Thatcham, presented data that showed cars with five-star Euro NCAP ratings reduce risk of injury by 15% and fatalities by 79%. Why is Euro NCAP phasing in ADAS? “Ninety percent of crashes have some element of driver error, many due to distraction,” he explained at the same Continental press conference. “ESC is an established life saver; other ADAS systems show potential.”
One of the first AEB systems marketed was a LIDAR based system (supplied by Continental) dubbed City Safety, which was standard on the MY 2008 Volvo XC60. “Since 2008, we have been able to monitor European and U.S. crash claims [for the XC60] and we see a 27% reduction in rear impacts vs. other similar 4X4s,” he said. “This is important,” he stressed, “because this is the first time we have evidence that AEB systems are actually working.”
He presented other data that showed automatic braking is more effective than a simple warning to the driver that a crash is imminent, which is referred to as Forward Collision Warning (FCW). Robotic intervention through AEB is best, and the XC60 is proving the point.
However, he stressed further development is required. For example, the Volvo City Safety uses only LIDAR and is generally effective in the tests Thatcham devised for speeds up to 20 km/h (12 mph). Faster than that in the Thatcham tests, and the system cannot act fast enough to avoid crashes, though it mitigates crashes through prebraking. Other technologies do better. He presented data that showed the passive stereo camera EyeSight system that equips Subaru Outback systems for AEB is more effective in avoiding crashes below 50 km/h (31 mph). “LIDAR is stopping around 30 km/h, while [the tested] stereo system is effective up to about 50 km/h,” he said.
Other functions that will be important for achieving five-star Euro NCAP include pedestrian protection, LDW, and LKA (lane keeping assist.)
Near-term technologies—bridge to the future
Feedback such as this from Thatcham is giving direction to Continental’s efforts in future developments. For example, to extend front crash avoidance to 72 km/h (45 mph), Continental will add a multifunction camera to a LIDAR system, the resulting sensor fusion extending the range and therefore the reaction time.
Looking out to about 2018, the general theme of sensors as Continental describes them is shown in the accompanying table.
“Fully automated driving will not be a revolution,” says Schneider, but a development of new functions (think LDW or braking for pedestrians) that piece by piece will build all functions needed for robotic driving. He noted that computers and algorithms still need development. “Functions need to work 100% of the time.” As one commentator from the company remarked, the issues around legal liability of autonomous driving may actually be the biggest hurdle in development.