GSA conference: Semiconductors drive automotive transformation
The automotive industry is facing enormous challenges, said Dominik Wee, a top automotive expert of consulting company McKinsey. Starting from the finding that the automotive industry is currently influenced by four disruptive trends (electrification, connectivity, autonomous driving and the diversification of mobility), Wee predicted that new business models that will emerge on top of the traditional car as the “mobility hardware” will expand the global automotive revenue pools by 35 percent by 2040, translating into a volume of $1.5 trillion of extra business with software, data and services. Even better: Most of this on-top business will be achieved as recurring revenue, in contrast to today’s business model based mainly on the one-time-sale of a (hardware) car.
Also the user behavior will significantly change, said Wee: Today’s predominant model of buying and owning a car to use it will dwindle in favor of a new paradigm: Car usage will be linked to a lesser extend to car ownership. Instead, consumers will increasingly lease vehicles on demand through sharing platforms. This does not appear to be good news for the automotive industry – after all, the mobile society will need fewer cars to be built. But at second looks, the news is not that bad: By 2030, 9% of all vehicles will be shared in some way or another, effectively reducing the market for privately-owned vehicles. Nevertheless, these shared vehicles will be used much more intensively than a private one, and this they ride more kilometers per year. And there is another aspect for car design, resulting from the changing ownership model. Today’s vehicles are typically a compromise between different tasks: commuting, business, leisure, or shopping. The new mobility model based on car sharing however will enable users to make less compromise – they will be able to rent or lease a vehicle exactly designed for the respective purpose. Which will lead to an even more pronounced differentiation in the car designs.
Definitively good news for data service providers is Wee’s prediction that the willingness to pay for data-driven services is increasing; while in 2014, only 21% of car users said they are willing to pay for connected car services, in 2015 this figure has risen to 32%, an increase of 53 percent points in just one year.
Is this good or bad news for the electronics industry, and in particular for chipbuilders? Basically it is good, said Wee. The market transformation described above allows semiconductor manufacturers to broaden their value proposition. Semiconductor expertise will be at the core of many of these new services, placing the chip industry in a favorable position to participate in the transformation of the mobility industry from automotive hardware to service-based models. However, it will be essential to acquire a deep understanding of the automotive domain. And the road to the future won’t be necessarily a comfortable one, Wee hinted. “Prepare for uncertainty,” he advised.
For automotive supplier (and to some extend, semiconductor manufacturer) Bosch, the mobility of the future is one thin in the first place: It is connected. Like Wee, Bosch Chief Expert Automotive ADAS Markus Tremmel pointed out that the individual mobility of the future will be a composition of multiple means of transportation, not necessarily dominated by today’s model of owning a private car. For technology development, this will have consequences. “The multimodal approach is only possible in a fully integrated electric environment”, he said. In particular the automated car needs to be connected, because it has to make decisions, “act like a brain in concert with other brains,” he sketched his scenario.
To drive in autopilot mode, the cars will have to be connected. Plus, they will need to have a broad range of sensors at their disposition – from long-range and mid-range radar to video cameras and ultrasonic distance sensors. Not to forget, Lidar. “Lidar is moving to mass production”, Tremmel said.
surround sensor equipment. (Source: Bosch)
But these “direct” sensors won’t be enough. The cars will also need to rely on “indirect” sensors – the sensors of other vehicles. “We need to look around the corner, we need to know what others know”, he said. The technology for this is already in reach: V2X (vehicle-to-x communications). Which is still not enough. The next ingredient necessary to complete the automated driving picture is automated decision making. Here, deep leaning is only one option; others are probabilistic reasoning and deterministic reasoning. “Deep learning will give you the possibility decide quicker than with conventional methods”, he said. However, the hardware for this approach is not available – not at the level of sophistication and performance needed by Bosch. “Current microprocessors are not suited to do it (deep learning algorithms) efficiently,” he said. “We need new microprocessor architectures”.
In the “macro view”, Tremmel currently observes two technology camps that prepare for the technology competition about automated driving. The “evolutionary” camp, represented by today’s traditional carmakers, and the “revolutionary” camp with contenders like Google and Apple. Both are moving into the same direction, but both are facing different challenges. Who will win? This remains to be seen, because, as Tremmel observed, “none masters the others field of expertise completely”.
Related articles:
Is LTE the better V2X technology?
How will deep learning change SoCs?
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
