Silicon Drives Autonomy Movement
Semiconductors that underlie the industry’s rapid transition are evolving rapidly
Silicon has joined iron and steel as a basic infrastructure material for autos—and the drive towards autonomy won’t slow that transition. Numerous tradeoffs will challenge automotive engineers as solid-state sensors and graphics processing units (GPUs) join the microcontrollers that have been the workhorses of automotive electronics.
Microcontrollers are the foundation for the industry’s gains in fuel economy and infotainment, as well as the emergence of automated-driving features They’re also the underpinning of the industry’s expanding use of software to provide new features and functions.
“Software is becoming a larger part of total costs. We’re seeing decisions being made on the basis of legacy software, (but) using it means they don’t need software development. Managing the software explosion that comes when you have 40 modules talking to each other is a big challenge,” said Amrit Vivekanand, Vice President of Renesas’ Automotive Systems Business Division.
Microcontrollers have to process all that software at breakneck speeds while managing all the communications between modules. They also have to simultaneously perform diagnostics and check the integrity of operations. Though clock speeds and throughput are necessities when automotive engineers choose microcontroller families, those aren't the only important parameters.
“When you look at the computing challenge, it’s not so much simply wanting a lot of computing power, it’s computing power within a set power level. That impacts heat, which impacts reliability. Software is where power gets consumed. That has a lot to do with how you architect the chips,” said Vivekanand, who earned an MS in Electrical Computer Engineering at Duke University.
When electronic engineers are establishing control architectures, there are always tradeoffs. Chip designers need to determine which functions can be transferred into hardware and where they want to retain flexibility with software. Doing everything in silicon yields low power consumption, but it increases chip size and reduces flexibility. Using software for every task shrinks die size, but power consumption pushes heat skyward.
Those silicon-level choices highlight the complexity facing OEMs, which are integrating scores of semiconductors and millions of lines of software into one huge, mobile electromechanical system. The auto industry in turn is addressing this complexity by linking multiple partners in ecosystems.
“The main reason for building ecosystems is time-to-market. When the OEMs want something like a hypervisor, they could do it, but it’s quicker for them to deal with a company that is deeply involved in that technology, or any other technology,” Vivekanand said.
These alliances can also agree on interfaces, something that’s an issue given the lack of standards and the number of variants in use. AUTOSAR and Adaptive AUTOSAR are helping; on the software side, the number of operating systems will make it difficult for smaller companies to create apps and maintain them.
“In something like safe operating systems, there are multiple vendors. Some companies think that being able to run apps on any of these will be easy. It’s not. If you’re a small third-party company, you can do one, or do three, but that’s about it. You can’t do them all,” Vivekanand said.
The transition to autonomy ripples out in many directions. It’s also changing views of safety. If an autonomous system on a driverless car fails, the vehicle may return control to humans.
“Today, people focus on functional safety, making sure systems fail safely. Going forward with autonomous-vehicle architectures, you may not want things to fail safely, but to fail operationally so functions are still available to the driver when the autonomous capabilities aren’t there,” Vivekanand asserted.
On deck: GPUs
The explosion of safety-related sensors has also created an opening for GPUs, which process images from cameras, radar and lidar inputs. These highly parallel microprocessors also are at the forefront of the artificial-intelligence revolution. The Renesas autonomy platform being developed with ecosystem partners includes GPUs and application-specific standard parts.
“We use GPUs for graphics. We do some in-house—they’re in our line—and we have a GPU license with Imagination Technology,” Vivekanand said. “For special applications like deep learning, we have some custom silicon IP. That’s the path we’re going down—we feel it’s very effective.”
The entire automotive supply chain is also grappling with connectivity and security. These intertwined issues impact most of the electronic control units in a vehicle.
“Connectivity is not just communication between wireless, the cloud and the car. It’s also the communication inside the car, between modules, providing the ability to get data from sensors, to reflash modules, to get data in and out of the car,” Vivekanand said.
The security that is paramount for connected cars goes down to the microcontroller level. Chipmakers are putting trusted zones and encryption modules on many chips. But adding trusted zones can make it more difficult to analyze faults.
“It’s harder to lock a chip down; when you do, it’s harder to see why chips fail. If a chip fails, you want to put it in a debug mode so you can see what happened. When things are locked down, it’s harder to debug. There are always tradeoffs,” Vivekanand explained.
Tradeoffs and transitions are occurring at all levels. Startups are beginning to play a more important role in the industry, adding a layer of complexity for OEMs. At the same time, there are changes in the large companies OEMs have relied on for years.
“There is a Tier 1 consolidation that makes it more difficult for OEMs to differentiate," Vivekanand said. "We’re seeing system integrators being called in more and more to tie in software that OEMs are getting from sometimes 20 or 30 companies."
Some automakers are responding by doing more work themselves. Now that software is becoming a differentiator for various makes and models, some coding projects are being brought in-house. So chipmakers like Renesas increasingly need to write some basic programs that make it easy for design teams to run reference designs through their paces.
“There’s a big change in software, OEMs are doing more of their own software," Vivekanand said. "Chipmakers are also doing more software. Much of it is written so we are able to measure the metrics carmakers care about. When we want to position ourselves for an in-vehicle infotainment system, we use it to show that our chip systems meet all the speed and communications paths so carmakers can see how the chips work."
University of Rochester Lab Creates New 'Reddmatter' Superconductivity Material...
MIT Report Finds US Lead in Advanced Computing is Almost Gone - Mobility...
Airbus Starts Testing Autonomous Landing, Taxi Assistance on A350 DragonFly...
Boeing to Develop Two New E-7 Variants for US Air Force - Mobility Engineering...
PAC-3 Missile Successfully Intercepts Cruise Missile Target - Mobility...
Air Force Pioneers the Future of Synthetic Jet Fuel - Mobility Engineering...
Leveraging Machine Learning in CAE to Reduce Prototype Simulation and Testing
Driver-Monitoring: A New Era for Advancements in Sensor Technology
Electronics & Computers
Tailoring Additive Manufacturing to Your Needs: Strategies for...
How to Achieve Seamless Deployment of Level 3 Virtual ECUs for...
Specifying Laser Modules for Optimized System Performance
Volvo CE Previews ConExpo 2023 Display
ArticlesManufacturing & Prototyping
Low Distortion Titanium in Laser Powder Bed Fusion Systems