We won't be turning back to a single control computer in a car anytime soon. Cars will always have 20-50 micros in them. One of the biggest reasons cars went to the multiple ECU arrangment was to reduce the weight and cost of the wiring harness. Going to a single ECU would bring back long thick heavy wiring harnesses. Having to run separate wires for every sensor and control signal back to a single control ECU is cost and weight prohibitive. Placing little CPUs out near the sensors also improves measurements. I don't have to drive that analog signal back over 20 feet through an incredibly noisy electrical enviroment.
We will have a central control computer providing the automated driving functions. It will conect to the 20-50 modules already in the car to provide the control over a dedicated vehicle data bus. That bus currently is CAN. It will need to be upgraded to something else to get the thoughput needed for future.
Intel has the same problem here that they have in the battery powered markets. Their products draw way too much power. It's very hard to find an Intel CPU that can meet the Auto companies low current draw requirements.
I suspect this is one of the markets that Intel is "we did that, went out, now in, later likely out, maybe in the future in again)....they did exactly similar entries and exits to so many markets in the past...make up your mind Intel!
Intel is not new to the the automotive business, they developped the 8051 embedded CPU. I implemented it in numerous automotive applications in the 90's (car radios and infotainment). It was a great CPU for that purpose, then it became too expensive compared to Freescale and others.
It's a pity that Intel killed this product line by not adding more integrated peripherals to it and not reducing significantly its price. Another point: the 8051 had few or no Flash/EEPROM on chip, we used it as an OTP microcontroller for production (flash only in the dev phase). We were not able to upgrade the embedded SW once the car was out of factory.
Today a good FLASH is mandatory on such controllers and FLASH technology for automotive is of prime importance (think reliability). See how Spansion acquired Fujitsu automotive microcontrollers division.
I'm not sure about that. Most of the compute power required for ADAS & self-driving cars is environment analysis (vision algorithms & other such things)
Using a many-core approch for environment analysis (multi-threaded) and a medium-sized CPU for the decision part of the driving algorithm (single "big" thread) might give better perf/W and perf/mm² results.
Moreover, those 50-odd CPUs currently satisfy rigorous safety requirements, such as ISO26262 (which intel is just starting to look into. cf job offer on intel.com "We are excited to launch a new position at Intel, ISO 26262 Functional Safety Manager"). Intel is obviously far behind on these matters.
>> "They were right," said Steve, "There was no market for it...yet."
That was the mark of innovators. They create their own market. It is not a question that no one is using it, it is question of you imagining it and then creating it for the market. If the car firms do not change, TESLA and Google will put them out of business.
In my opinion, today's car electronics is a scattered approach where a multitude of independent modules have been implemented for various functions - Engine control, ABS, Climate control, Navigation, infotainment and so on.
For the car of the future, where we are talking about a totally autonomous driverless car, we need an integrated approach for all these functions and instead of those 50 or odd CPUs , we may opt for a powerful dual CPU kind of central complex to centrally monitor , manage and control all the car functions. And there , I think Intel will have an edge.
So, yes there are many angles and views for this new growing technique, I as well was in a kind of opinion years before when I heard of this initiative, but today when I see the results researchers and developers are getting because of technology it seems that this will really going to happen for positive outcome.
What are the engineering and design challenges in creating successful IoT devices? These devices are usually small, resource-constrained electronics designed to sense, collect, send, and/or interpret data. Some of the devices need to be smart enough to act upon data in real time, 24/7. Are the design challenges the same as with embedded systems, but with a little developer- and IT-skills added in? What do engineers need to know? Rick Merritt talks with two experts about the tools and best options for designing IoT devices in 2016. Specifically the guests will discuss sensors, security, and lessons from IoT deployments.