by Mike Millikin, editor, Cqtimes
This post is part of the ‘Think Further’ series sponsored by Fred Alger Management. For more “Think Further” content, please visit .”
The car has become the most computationally complex high-tech device with which the vast majority of consumers will ever come into —let alone own. The car has also become a focal point for the development of innovative and entrepreneurial technologies, services and business models designed to enhance and evolve not only the basic efficiency of vehicles, but also the way in which they are used alone and as a part of a larger multi-modal transportation eco-system.
There have been and continue to be a number of market drivers forcing this evolution: concerns over health effects, congestion, consumption of petroleum-based fuels, climate change, an ever more rapidly increasing population and what appears to be an inexorable movement toward large-scale urbanization. The auto industry has known for a long time that business-as-usual was literally unsustainable.
Globally, the automotive industry is being driven to provide more fuel efficient, less polluting and more electrified vehicles primarily by regulatory pressures, although consumer demand is playing its part as well. Within that broader context, however, are a number of technology areas which are not only helping automakers meet their regulatory requirements, but also changing the nature of the relationship between consumers and cars even during the short- to mid-term (5 to 10 years out).
Among the key technological enablers of this disruption in the automotive industry are microprocessors, software and sensors. A recent study by Roland Berger Strategy Consultants that current top-end vehicles have as many as 100 ECUs (electronic control units) and run on more than 100 million lines of code. For comparison, a Boeing Dreamliner 787 has about 15 million lines of code—about 10% of the amount expected in autonomous vehicles of the near future.
Helmut Matschi, member of the Executive Board of Tier 1 automotive supplier Continental, recently noted that “Software is developing into the single biggest enabler of innovation in the car.”
In-vehicle electronics—microprocessors, sensors, actuators, instrumentation panels, and controllers are critical in the design and operation of engine controllers; traction motor controllers; safety systems; advanced driver assistance systems; chassis control; measurement and diagnostics; navigation systems, emissions monitoring, communications, and, last but not least, entertainment.
Volkswagen unhappily provides a now-infamous example of the foundational role of processors and software in modern cars; its clever software defeat device monitored different sensors and inputs to determine whether or not its vehicle was being tested on a dynamometer or was driving on the open road. The software then adjusted its emissions control calibration accordingly.
Achieving regulatory standards for fuel consumption and emissions across the wide variety of operating conditions and loads experienced by a car requires increasingly capable—and increasingly intelligent—software control. Programming in hard “maps” to adjust engine and emission parameters has been industry practice for quite some time.
However, as the number of potential data inputs increases (variable valve timing, variable valve lift, variable effective compression/expansion ratio (Miller cycle, Atkinson cycle), fuel pressure, fuel injection timing, fuel injection rate, fuel injection events, fuel reactivity, turbocharger boost pressure / vane control, exhaust recirculation rate, combustion chamber design and exhaust aftertreatment system control) and as the output requirements increase (more stringent regulatory targets), the static map approach is impractical. Engineers are turning to model-based controllers.
Further, recognizing that the different driving styles and habits of different drivers have a tremendous impact on actual fuel consumption and emissions performance, researchers are now investigating ways of having the controllers adjusting their calibrations online in response to the different driver behaviors—in other words, a customized profile for each driver. (Earlier post.)
This is not just an issue for combustion engined cars; to deliver as much range as possible, electric cars need to manage their power consumption—a task also assigned to controls and controllers. Much of Audi’s work on its new all-electric SUV (e-tron quattro), for example, is focusing on the software control of the twin electric motors on the rear axle that give the vehicle the cornering performance of “a hunting dog after a rabbit”. (Earlier post.)
When it comes to implementing new types of more fuel efficient combustion models in advanced engines, the enabler is, again, software and controls.
Those are just a few examples. Overall, the proliferation of software and controllers is forcing a rethink in terms of overall vehicle architecture: electronic architectures need to be consolidated, while the interconnecting pathways—i.e., the computer network within each car—needs to be streamlined. As an example, the adoption of one-pair 100 Mbps Ethernet networking within new cars has been rapidly increasing. (Earlier post.)
One of the most perceptible shifts in the automotive world is the rapid movement to connected vehicles (enabled by these microprocessors, software and sensors)—and the advent of different levels of autonomous driving. The basic framework of vehicle-to-vehicle and vehicle-to-infrastructure communication enables an entire range of disruptive services from advanced driver assistance, to fully autonomous driving, to new models of mobility services that may change the fundamental usage patterns of vehicles that have dominated for decades.
Endowing the vehicle with the cognitive capabilities needed to deliver these functions requires what the industry calls “sensor fusion”—the ability for a central processing unit to take the inputs from the increasing number of sensors on the vehicle and process the data in real time to determine what needs to be done.
An outgrowth of that is the coming of Gigabit Ethernet (1000BaseT) to the car. Gigabit Ethernet will be used as a network backplane to connect the other computational/control domains of the car, as well as to provide direct, uncompressed high-speed connectivity for certain critical applications—high definition video for real-time object identification for advanced safety systems and autonomous vehicles, for example. Fabless semiconductor company Marvell has just introduced the first automotive Gigabit Ethernet transceiver, conforming to what appears to be the emerging standard (802.3bp). (Earlier post.)
Another outgrowth of this movement to the software-intensive car is the need for over-the-air (OTA) software updating. First brought mainstream by Tesla’s Model S, OTA has been gaining rapid adoption, with automakers such as BMW, Hyundai, Ford, Toyota and Mercedes-Benz now offering OTA software updates. OTA technology provider Movimento notes that this capability for “software defined cars” will eventually result in production versions of the driverless car and the other major advances rewiring the auto industry.
In short, vehicles are becoming more cognitive and responsive—and also more controllable. This shift from a dumb platform (essentially a powered wagon) to a smart and responsive platform gives us the ability to reshape the future of transportation into a more sustainable form.