Page 19

EETE JAN 2014

Amazing race toward autonomous car design By Junko Yoshida Chip companies are finding 2014 a critical year to drive their technology into semi-autonomous car platforms - currently in development by a number of different carmakers. Such platforms, says a Freescale Semiconductor executive, will ultimately become the basis for each car OEM’s own, branded, self-driving cars. Major car OEMs including General Motors, Nissan, and Toyota are all racing to develop their own, unique semi-autonomous architectures. While describing it as a “friendly race,” Davide Santo, Freescale’s ADAS (Advanced Driver Assistance System) microcontroller product line manager, said the competition intensified when Nissan this summer announced that its first cars using the autonomous car platform will arrive in 2020. Germany’s Daimler AG similarly announced plans to start selling a self-driving car by 2020. Self-driving cars are no longer just about Google cars. Carmakers aren’t pontificating or debating the pros and cons of self-driving cars, either. Daimler is already using self-driving as a way to differentiate from other luxury cars, as it competes with its German rival BMW. Just as much as car OEMs are under pressure to come up with their own autonomous car platforms, automotive chip suppliers such as Freescale, Infineon Technologies, and NXP Semiconductors are similarly feeling the heat. The second half of 2014 is a sort of consensus deadline for leading car OEMs to make final decisions on architecture and technologies for semi-autonomous car platforms. By then, Freescale says, it will be working closely with OEMs, contributing its ideas and making proposals, hopeful for design wins for key technologies on the platform. Carmakers are all “working toward” autonomous cars, agrees Drue Freeman, senior vice president for global automotive sales and marketing at NXP. “They are preparing roadmaps for self-driving cars.” But for now, the most visible competition among OEMs is the rollout of different sensor, camera, and radar technologies to enable ADAS. While different technologies help create a variety of ADAS features, ADAS, in essence, consists of two principles, explained Freescale’s Santo. First, you create a grid around a car and keep the car running within a lane. Second, you communicate where the car is, relative to other cars and the road infrastructure. Although it’s easier to think of self-driving cars as essentially built on a combination of different ADAS features, such a view may be an oversimplification. Hans Adlkofer, head of the system group at the automotive division at Infineon, explained that a variety of ADAS technologies, integrated in a vehicle, need to be able to run seamlessly on a single unit of underlying software adopted by each car company’s platform. In other words, ADAS features can’t exist totally independent of the semi-autonomous car platform where they will be deployed. Truth about autonomous cars Nobody believes that self-driving cars will be easy to develop. In fact, technology aside, the uncertainty of consumer behaviour is a major concern among car OEMs. Car companies are responsible for developing a solid solution that can guarantee driver and passenger safety in semi-autonomous/self-driving cars. In the 1949 Geneva Convention treaty, one section stipulates that every car must have a live driver behind the wheel. But with so many driving features getting automated in semi-autonomous cars, OEMs today know very little about whether human drivers will continue to pay attention to the road, or start doing something else while leaving the driving to the car. Carmakers designing a semi-autonomous car platform need to pay attention to “drivers’ fatigue and attention levels, while checking their driving styles,” said Freescale’s Santo. ‘Flight’ simulator for autonomous cars? In the next few years, car OEMs will be “collecting, evaluating, and analyzing a lot of data on drivers’ behaviour in semi-autonomous cars,” Santo explained. Some OEMs and tier ones are even running semi-autonomous car simulators, similar to flight simulators, said Santo, to see what happens to car and driver when complex situations arise on the road. Carmakers need to figure out a threshold level as to when the car should return control to the driver. In designing a semi-autonomous (or ADAS) platform, OEMs are trying to figure out interactions among different ADAS features within a system, especially to prevent any single point of failure. They’re emphasizing built-in redundancies wherever appropriate, says Santo. One could compare drivers in autonomous cars to pilots flying an aircraft on autopilot, but “cars can be much more dangerous than airplanes,” said Santo. First, there are more unexpected hazards on the road. But more important, rank-andfile drivers are not professional pilots. Because the vast majority of drivers are amateur operators, the development of self-driving cars at car OEMs and tier ones is driven by a team of experts from multiple disciplines ranging from psychology to fuel economy, Santo added. Junko Yoshida is Chief International Correspondent at EE Times www.electronics-eetimes.com Electronic Engineering Times Europe January 2014 19


EETE JAN 2014
To see the actual publication please follow the link above