Automotive King Redefines Future
"We estimate from figures supplied to us by TU [Technische Universität] München that the total cost of a new car was 15 percent electronics in 1990, but will grow from 30 percent in 2020 to as much as 50 percent by 2030," Patrick Morgan, vice president and general manager of advanced driver assistance systems (ADAS) told EE Times. "And that doesn’t count the $400 billion that could be saved by preventing accidents, not to mention the 1.2 million lives saved worldwide and the 50 million injures prevented."
Today, the biggest safety systems in automobiles are passive systems like airbags and enhanced safety systems (ESC), such smart cruise controls, blind-spot detection, infrared night vision, collision mitigation systems, and rear-view camera systems all the way to advanced driver assistance systems (ADAS) like lane departure warnings and automatic braking to prevent rear-ending the car in front of you. But for the future, the name of the game will be "predictive safety," according to Morgan. "We will be selling sophisticated vision and image recognition systems—in addition to the radar where we already sell the large majority of systems with one million sold so far—that will be able to classify objects and take evasive action ranging from a simple alerts to the driver all the way to taking over control of the car to prevent collisions–especially with pedestrians."
The beating heart of Freescale’s ADAS is the S32V stereo video processor announced at this spring’s Mobile World Congress 2015 (March 2-5, Barcelona), but which is only now becoming widely available as production ramps up. The "V" in S32V stands for vision, which will lead to a new era of smarter of ADAS, according to Matt Johnson, vice president and general manager of automobile microcontrollers at Freescale.
"Freescale has already won a large majority of the market in [77GHz] radar, but looking forward, another big component will be vision," Johnson told us.
Freescale’s QorIQ processors are ideal for video-algorithm development, but according to Johnson, are too expensive for deployment in most cars. Thus Freescale created the S32V234, a 1-GHz quad ARM Cortex-A53 (automotive grade ISO functional-safety standard ISO26262 ASIL B) plus a Core-M4 realtime microcontroller, a 3-D graphics processing unit (GPU), ARM’s NEON vector single-instruction-multiple data (SIMD) technology and two Mobile Industry Processor Interfaces (MIPIs) for high-speed video input, two parallel video inputs and a Crypto Security Engine (CSE).
Fig. 3: The number of traffic fatalities (green line) has been steadily decreasing as more safety features have been added, from seat belts, to air bags, to automated driver assistance systems (ADAS) and could approach zero with next-generation predictive safety systems. (Source: Automotive Antenna Workshop)
"We provide the safety, reliability and security hardware including encryption/decryption and key management," said Johnson. "Competitors like mobile don’t protect their algorithms from hackers and reverse engineering in an open ecosystem."
The secret sauce in the VS32V234 is its dual APEX-2 image processing engines which can perform complex cognitive recognition tasks potentially realizing Morgan’s goal of predictive safety, which eventually will allow cars to drive themselves.
"The road to autonomous vehicles is by way of making the so safe they can drive themselves," Johnson told us. "We already have sensor fusion, autonomous emergency braking, lane departure correction and are perfecting pedestrian detection."
Fig. 4: The number of automated driver assistance systems (ADAS) will double over the next three-to-five years. (Source: FSL)
Autonomous vehicles will not all look like Google’s "Johnny Cab" but will be even safer with a driver behind the wheel.
Article courtesy of EE Times.
R. Colin Johnson is Advanced Technology Editor, EE Times