Contributed by Bob Chabot
Making Mobility Work
Smart vehicle architecture is the key to autonomous mobility
By 2020, vehicles are expected to exchange enormous amounts of data in the blink of an eye. How will they handle all this data and power distribution? By being built on a radically optimized smart vehicle architecture that acts as the brain and nervous systems of an automobile, according to Aptiv.
As new vehicles become smarter and advanced driver assist systems (ADAS) evolve from semi-autonomous to fully autonomous, automobiles are quickly becoming rolling supercomputers. The architecture that allows these data-devouring computers, transceivers and sensor technologies to send, receive and process billions of bits of data in order to see and make decisions to move vehicles and their human passengers safely and reliably, all without a human driver, is being built by companies such as Aptiv, NVidia and others. So, what’s driving the evolution?
In the 1950s, on vehicles such as the ’57 Chevy Bel-Air, wiring was simplistic, comprised of 12-volt systems, ring terminals as the primary connection method. (All images — Aptiv) (All images — Aptiv)
Vehicles Need More Computing Horsepower than Ever Before
“Consumer demand for safety and software-enabled features is increasing at an unprecedented rate. This increase in software-enabled capability impacts infotainment, user experience, active safety and connected vehicle services, and paves the way for the ultimate application, autonomous driving.
The 1957 Chevy Bel Air was not only a beauty on the outside, it was a paragon of simplicity on the inside — at least when it came to its electrical system. But as vehicles have evolved over time, so has the electrical architecture. Jump ahead to 2018 where vehicles need to transfer 15,000 pieces of data in the blink of an eye. And within just a couple of years, by 2020, that will jump to 100,000 pieces of data.
When applications like Level4/Level5 autonomous driving emerge, the amount of computing power necessary to enable the advanced vehicle features involved, in real time, will increase exponentially. Let’s express it in computing terms: a “flop” refers to the number of complex operations calculated each second. Current vehicles are completing less than a teraflop (or 1 trillion) operations a second — approximately equivalent to the computing power of one iPhone 7. An SAE Level4/Level5 autonomous car will require more than 200 teraflops to be calculated each second —the computing power of more than 500 iPhone 7’s. So, it’s not a stretch to say the vehicle you’ll drive to work will be a supercomputer.
An Effective Separation of the Hardware from the Software is Critical
“Vehicles will continue to need to be safely and securely connected to the ecosystem around them,” shared Glen De Vos, CTO for Aptiv. “Our secure gateways will provide all the computing power necessary to enable high-speed data networking, while also meeting the most stringent functional safety and cybersecurity standards being developed for the industry.”
“Critically, as computing platforms become more powerful, separating the underlying hardware from the software more effectively will enable vehicles to become more like home PCs — platforms that allow users to install lots of different types of software,” he added. “This is important because it allows the development of software independently from hardware, which takes a lot longer to bring to market. Think of it like the speed of developing an app vs. the time it takes to develop and release a new smartphone.”
As the vehicle becomes the largest source of data worldwide, it is critical that the appropriate data be analyzed in real-time in order to enable autonomous vehicles, mobility on demand services and the emerging data marketplace. By being connected, vehicles are beginning to interact with our world in a way that will transform our lives.
Smart Mobility Architectures Need to Support All Self-Driving Components
Full autonomy is a reality today, but to move forward, electrical architectures must also undergo a radical transformation. While today’s vehicles require hundreds of sensors to operate safely and the data generated by those sensors needs to be analyzed in real time, the advanced capabilities necessary to provide power, signal and data for the increasingly autonomous, electrified and connected future of mobility adds complexity and an increased chance for system failure. These architectures need to operate in limited space, be cost effective and ultimately deliver on safety.
“This is where technological evolution comes into play. Fail-safe and fail-operate systems are crucial for autonomous driving,” noted Markus Kerkhoff, Global Engineering Director for Aptiv Connection Systems. “They must be able to tolerate a failure and adapt in order to continue to operate properly. In other words, the system must be able to learn. Smart vehicle architectures will include electrical centers designed to include multiple layers of redundancy — backup that would be redundant in a fail-free environment, but provide safety nets for possible failure in complex systems.”
“What are some of the changes you can expect?” he posed. “Next-gen printed circuit boards will soon transition to fully-protected solid state electrical centers that will serve as the building blocks for autonomous mobility. Power supplies will have to be redundant to safeguard critical autonomous features. Both current and voltage monitoring of advanced driver assistance systems (ADAS) components will be essential, with reporting faults of ADAS system part of the system. All of this new technology and content will add cost, so providing function and safety at the most affordable price will be crucial.”
Updating a car's wiring is a lot like performing neural surgery. Every wire is part of an intricate and increasingly complicated network. How intricate? Today's premium cars have 2.5 miles of wiring and more than 100 ECUs. And both those numbers are growing larger and more complex every day.
Rethinking the Vehicle's Nervous System
Updating an automobile’s onboard wiring is a lot like performing neural surgery. Every wire is part of an intricate and increasingly complicated network. How intricate? Today's premium cars have 2.5 miles of wiring and more than 100 ECUs. And those numbers are growing larger and more complex every day. Writing the software code behind the “brains” of today’s increasingly sophisticated vehicles is a daunting task. But so too is re-thinking the architecture — its nervous system — which allows these brains to act.
OEM architects have the challenge of finding newer and better ways to help enable their burgeoning functionalities. Aptiv says resolving this architectural challenge requires three distinct, but interwoven approaches:
- Optimization — While a car’s commercial cycle usually lasts approximately six years, Aptiv engineers break that timeframe into two three-year segments and enhance a car’s features as much as they can without rewiring new technologies or parts. This allows the vehicle to stay top-of-the-line in terms of functionality, without taking on too much unnecessary rewiring.
- Evolutionary — This approach takes into account the reality of legacy cars. Often OEMs will have plans for a new car, but they have a lot of carry-over parts from their existing models. Aptiv’s evolutionary approach consists of keeping 50 percent of the parts as carry-over and allowing 50 percent for new components. This approach allows for the continuation of legacy cars, while we ensure the architecture is as up-to-date as possible.
- Visionary — This is where the real fun happens, according to Aptiv. The visionary approach allows a fresh start at square one. No legacy issues. No restrictions. It’s an innovator’s playground, where engineers can take truly revolutionary approaches to re-imagining how the world gets from Point A to Point B. These methods enable work that is not only efficient, but also respects the vehicle as a whole and avoids additional costs for the Tier 1 supplier and manufacturer.
Whatever blended approach is chosen, the vehicle’s nervous system must deliver the fast, safe and reliable distribution of data and power that the car’s brain demands. Smart vehicle architectures will be the key enabler.
Examples include components that have resulted from those explorations, such as enhanced energy management like stop-start functionality and personalized features like memory function, enhanced infotainment and diagnostics. For instance, advances in enhanced energy management are already reversing the growth curve in alternators for Level4/Level5 with significantly less control units, full redundancy for power supply and high-speed data exchange.
[Editor's note: Visit MOTOR.com for the latest diagnostic and service insights.]