Get email delivery of the Cadence blog featured here
Don't expect to see completely autonomous automobiles any time soon. But cars are already extremely automated and are getting more so, with dozens of electronic control units (ECUs) that handle everything from transmission and braking control to keeping you from drifting out of the lane.
All this takes electronic design, and that's why there was an automotive track at this year's Design Automation Conference (DAC 2014), held June 1-5 in San Francisco. In addition to the papers and panels in this track, and a few automobiles on the exhibit floor, a "dual keynote" was given by James Buczkowski (upper right), Henry Ford fellow and director of EE systems at Ford Motor Company, and Jim Tung (lower right), fellow at MathWorks, a provider of model-based design tools.
For the dual keynote, both of the presenters were on stage at the same time, and they took turns talking and presenting slides and short videos. It all worked pretty smoothly and it added up to an informative and entertaining hour-long presentation. Here are some of the main points that emerged.
Buczkowski—It's not the technology, it's the experience
Buczkowski emphasized that what's important to customers is having an experience. "You will see a constant theme in my messaging that it is not just the technology—the technology is an enabler for a great experience, and that's what it's all about for those of us in the automobile industry." Experience includes what you can touch, feel, and see.
"My call to action to you in this [EDA] community begins with a focus on the experience," Buczkowski said. It's not just silicon that creates experiences—"we're looking end-to-end at silicon and applications all the way through."
Tung—"They just work"
There may be 50, 75, or even 100 ECUs in a late-model car, Tung noted. Functionality includes engine management, transmission management, stability systems for the chassis, traction control, air bags, suspension systems, anti-lock braking, adaptive cruise control—the list goes on and on. These are not simple, isolated systems. Adaptive cruise control, for instance, can anticipate when another vehicle is moving into your range, apply the brakes, and manage the engine. This requires the "confluence" of several different systems.
"These systems are successes because we can forget the fact that they are there," Tung said. "They just work. We ignore them and we rely on them." Of course, they need to be well designed in order to meet regulatory requirements and create a good consumer experience.
A big new trend in automobiles is sustainability—that is, using less fuel and causing less pollution. This is resulting in new electronic subsystems that did not exist with traditional internal combustion engines. These systems may include an electric motor, battery management, transmission management, regenerative braking, and new sensor technology. They require data and algorithms to manage data.
Buczkowski—Cars and cell phones are not the same
"We can't design cars like cell phones," Buczkowski said. "Our timing is a lot different." It takes 24 to 48 months to develop a new car from the ground up. This includes a new chassis, a new suspension system, and a new body interior, and possibly as many as 20,000 parts. Automakers design cars for a lifetime of about 10 years, because that's how long many customers expect to drive them.
Systems on chip (SoCs) and software need to be designed two or more years before production. Given the time lag, cars do not use the latest semiconductor processes and technologies. "The compressed time to market while still meeting quality requirements is a challenge," Buczkowski said.
Cost is also an issue. Cars and trucks must be affordable. Moreover, Buczkowski said, margins in the automotive business are "pretty slim and are not like those in consumer electronics." Regulatory requirements, often not understood by consumers, add to complexity and cost.
Buczkowski—Connected and autonomous cars
Buczkowski zeroed in on two themes—assisted/autonomous driving, and fully connected mobility. Assisted driving already exists, with features like adaptive cruise control, lane departure warning, forward collision warning, and blind spot monitoring. The key to autonomous driving is the proliferation of sensor technology. It will also require radar and cameras. A step in this direction is a new "traffic jam assist" feature that will keep drivers in the lane without having to steer at 25mph or less.
"We will introduce these [semi-autonomous] systems first and then build on those systems until we get to the automated vehicle," Buczkowski said. There are plenty of challenges. If the system breaks down, it will need to switch to driver-assisted mode seamlessly.
Fully connected mobility indicates that "cars will just become another machine on the Internet. They're going to be connected all the time and they need to be connected optimally," Buczkowski said.
The key to automotive electronic design is communicating and testing with models, Tung said. "Model-based design means you start very early in the process. You have to conceptualize the requirements and understand how the electronics, the software, the materials, the driver, and the road all interact." Through this process, designers can make sure they'll meet the requirements, and do exploration and optimization across different domains.
For example, consider a requirement to minimize fuel consumption while meeting performance objectives. This may call for an automated transmission that shifts efficiently from gear to gear. We need to model the transmission, the vehicle, and the transmission control elements as the car accelerates, de-accelerates, and merges into traffic.
Another aspect of model-based design, Tung noted, is that it can automatically generate code. This minimizes bugs due to human error and allows further optimization. He noted that MathWorks, which provides the MATLAB algorithmic development tool, collaborates with EDA vendors to develop integration with analog/mixed-signal and digital simulation tools. (A slide in the background showed that MathWorks offers integration with Cadence Virtuoso AMS, Incisive, and PSpice simulation tools).
Buczkowski—The bottom line
"This is a great time to be in the auto industry," Buczkowski concluded. "We have a lot of opportunities going forward. But in the end, customers will love their cars only if they can say, it just works."
Related Blog Posts
Gary Smith at DAC 2014: How System Design Is Changing Electronics
DAC 2014 Keynote: Imagination CEO Charts New Opportunities for Semiconductors