The next milestone for autonomous driving: ADAS
In a recent report IHS distinguishes between active and passive Advanced Driver Assistance Systems (ADAS). Passive systems simply provide a driver alert in hazardous situations while active systems, connected to the respective actuators like power steering or electric brake, can take the control over the vehicle in such a situation, if the driver does not react to the warning signal. In active systems vision-based driver assistance systems are increasingly connected to steering and braking systems of cars.
Because of this distinction, developers and system integrators increasingly demand turnkey platforms that meet the tough requirements of automotive ruggedness and reliability as well as applicable functional safety standards, in particular ISO 26262. Chipmaker Freescale, Green Hill Software and Neusoft have recently devised such a platform.
Image Source: Strategy Analytics
By providing an integrated hardware and software solution for automotive-related image processing systems, the new development enables developers to significantly streamline the development of complex algorithms and porting these algorithms to the respective target hardware.
The solution to be provided by the alliance utilises the Apex Image Processing IP (ICP) from CogniVue which is sold through Freescale. Neusoft, a Chinese company, contributes ADAS image processing software which automatically adapts to the underlying silicon; Green Hill adds its safety-certified Integrity operating system along with its MULTI tool chain.
The result is a holistic, ready-to-use image processing solution custom-made for ADAS applications such as pedestrian recognition, traffic sign recognition or collision avoidance.
"If you want to bridge the gap between first-generation ADAS and tomorrow's autonomous cars, you need to utilise object recognition and object identification technology", says Freescale image processing technology expert Ray Cornyn. Freescale intends to sell this technology to automotive markets and therefore has licenced the Apex-642 ICP Core IP from CogniVue. The Apex processor technology enables users to extract application-specific information from an image and processes this data as required for decision-finding or other processes.
According to IHS principal analyst for automotive semiconductors, Luca DeAmbroggi, the rapid growth of active-control systems will significantly spur the demand for automotive chips because active systems require more semiconductors.
Image Source: www.businesskorea.co.kr
Notwithstanding the overall increase in demand ABI Research indicates that the use of standardised ICT technology in vehicle systems may represent a threat for the suppliers of traditional automotive ecosystems such as Freescale, NXP and Renesas.
On the other hand, it opens up new opportunities for chip suppliers such as Broadcom, Intel and Qualcom as well as startups such as Silicon Image to sell automotive-grade versions of their products into this expanding market. And there is another effect that drives the value of the semiconductors used in these systems: Chips for active-control systems need to be compliant with stringent safety certifications, such as Automotive Safety Integrity Level (ASIL), or ISO 26262. Since safety considerations are of paramount importance to active-control, semiconductors ASIL systems will require compliant chips that typically cost more than their standard counterparts used in passive systems.
ASIL-compliant (MCU) will enjoy even faster growth than overall semiconductors for active-control systems, with ASIL revenue growing from $70 million in 2013 to $450 million in 2020. In parallel to this growth in semiconductors, the demand for sensors to be used in such systems will also show strong growth. Active (as well as passive) ADAS typically utilise radar and video sensors - stereo cameras in most cases; recently automotive supplier TRW demonstrated a system based on a smart mono camera.
In general, even safety-relevant systems such as Autonomous Emergency Braking (AEB) can be implemented with just one sensor, but if a higher ASIL level is required, carmakers typically employ a combination of both radar and video sensors. It is expected that even affordable cars will be increasingly equipped with AEB systems in the future.
Image Source: www.semiconportal.com
In Europe, the New Car Assessment Program (NCAP) is encouraging OEMs to implement AEB for pedestrian detection and to help vehicles avoid collisions with other vulnerable road users, in particular cyclists. Given the European program's call to action, global semiconductor revenue for pedestrian-detection systems is projected to exhibit steady growth in the coming years; rising from just $2 million in 2013 to $400 million in 2020.