Chipmakers, and design software providers are in a good position to leverage upcoming market opportunities in autonomous driving.
It’s not just car companies that are racing to build self-driving vehicles. An entire ecosystem is sprouting up around retrofitting existing vehicles with autonomous technology, despite the fact that the technology, infrastructure, regulatory and insurance issues are still not fully formed.
Uber already is using self-driving taxis, accompanied by a human driver, in Pittsburgh, Pa. And many in the automotive space believe the technology will see its first major disruption with ride-sharing services such as Uber and Lyft, which received a $500 million cash infusion from General Motors in January.
There are other examples, as well:
• Chinese search engine giant Baidu and Nvidia have announced a partnership to develop an artificial intelligence platform for self-driving cars.
• MIT-spinoff nuTonomy has self-driving taxis on the road in Singapore.
• Google also continues to expand the number of places it is testing vehicles, including the Phoenix metro area.
Common among them all is that none use native autonomous systems. Each was built, or is being built, using aftermarket technology.
“We are really at the beginning of something that is a once in a lifetime opportunity,” observed Elliot Garbus, vice president of Intel’s IoT Group and general manager of the company’s Transportation Solutions Division. “We’re at the beginning of what is a stunning transformation in the move toward autonomous driving. It is likely to be incredibly disruptive as transformations like this often are.”
The economic case for autonomous driving is estimated at a whopping $1.3 trillion in savings to the U.S. economy through less fuel, increased productivity, and reduced accidents, he noted. “In the U.S. alone there are about 30,000 people a year dying in automobile crashes—equivalent to a 737 falling out of the sky every day—and another 50 million injuries from auto accidents.”
To be sure, things are changing very quickly in the quest for autonomous driving. Less than a year ago most of the self-driving initiatives looked like science experiments run by non-traditional players, Garbus said. “The automakers were all very measured in their approach, to the effect of, ‘Universally, we are going to add a little more safety and capability, and sometime after 2030 or 2040, we will have a fully autonomous vehicle. Something happened over New Year 2016. All of that changed.”
From a technology standpoint, the automakers were very focused on trying to get as much computing as they could—air cooled, no fans, with a maximum of 10 watts, to fit under the passenger seat. Non-traditional players, meanwhile, were putting 2 to 3 kilowatts in the trunk of the vehicle. And then, suddenly, the race was on.
The momentum behind autonomous driving reflects what is happening with technology in general, said Rob Knoth, product management director for the DSG group at Cadence. “It’s not just in the automotive market. There is a movement of automation throughout different aspects of life. There are industrial aspects to it, there are automotive aspects of it. Even drones are autonomous systems. And in areas like agriculture, there is a huge impact of autonomous technology with self-driving combines, which have been in the fields for many years.”
He pointed out that it’s easy to look at ADAS and view it as a whole new car, but aftermarket add-ons are actually a bigger and more interesting trend to study because that is real. It can be used in existing cars, and it’s going to impact our lives a lot quicker.
“They are cracking the after-market ADAS problem in a fascinating way,” said Knoth. “They are looking at long-haul, and medium-haul trucks because they are hard on the roads, they take a lot of fuel, are horribly inefficient, and there’s a huge desire for increased safety. In addition, fewer people are going into truck driving as a career than in the past. Add those factors to the automation, retrofit it into big rigs in an economical way, and fuel costs can be cut, and safety increased.”
Otto has stated publicly that it would have self-driving fleets of trucks on the road by early 2017.
Using vehicle-to-vehicle communication to connect trucks, Peloton essentially adds after-market autonomous driving features also with the aim of improving safety and fuel efficiency.
Similar to cooperative adaptive cruise control, Peloton’s truck platooning system uses vehicle-to-vehicle (V2V) communication to connect the braking and acceleration between two trucks. The V2V link allows the lead truck to control the acceleration and braking of both trucks virtually simultaneously, reacting faster than a human or even radar sensors could.
What makes the aftermarket opportunity so challenging, though, is that it relies on the existing vehicle’s technology for much of the safety, command and control.
“If you look at the real promise, and where the bulk of native ADAS is focused, it’s the native integration of the intelligence and the sensing with the motors and with the operations of the vehicle,” said Knoth. “That’s where you are able to get a huge improvement in safety. With aftermarket, there is still a pretty big weak link, but it’s cool because we can realize it. It’s cool because you can get incremental improvements very quickly. But it does not address the core promise of real pervasive ADAS can bring.”
At the same time, aftermarket autonomous changes the ecosystem and leaves room for startups. One such startup, San Francisco-based Comma.ai said it will have a $1,000 autonomous driving kit by the end of this year. Cars must have electronic power steering, which means the steering wheel can be electronically actuated, along with electronic stability control. According to the company’s blog, the following are supported:
Definitely: Acura ILX 2016/2017 with Lane Keeping Assist System
Definitely: Honda Civic 2016 with Honda Sensing (all Touring edition)
Probably: All Honda and Acura with Lane Keeping Assist System
Comma.ai received a $3.1 million investment from venture capital firm Andreessen Horowitz in April.
In other recent activity, Delphi and Mobileye said they are partnering to jointly develop a self-driving system that automakers will be able to integrate into their cars starting 2019.
Computer vision, mapping, and machine learning specialist Mobileye developed the technology that powers Tesla’s Autopilot system. The two companies parted ways after the fatal Tesla crash.
BMW and Volkswagen also are racing to launch self-driving vehicles around the end of this decade.
In addition to opportunities for startups, aftermarket autonomous changes the ecosystem, and tool providers such as Cadence, Mentor Graphics and Synopsys are well aware of the changes.
“It used to be that the mechanical is what defined the vehicle,” said Wally Rhines, Mentor’s chairman and CEO. “Today, it’s more and more the electronics. And then, over time, it’s the software with the electronics. This applies to almost all of IoT, as well. When more of the value is coming from the software, the semiconductor suppliers either differentiate through their own embedded software or applications or cooperation with people who are creating it. Or they have a lot of trouble generating the value.”
Rhines noted that semiconductor purchases for the automotive sector grow about 6% per year. “For the amount of effort that is going into automotive electronics, the semiconductor companies are hoping to get an even bigger piece of the pie, and that means more and more software development that goes around the chips they provide,” he said.
Still, the automotive market is a different world for EDA vendors, whose customers traditionally have been chipmakers for computers and mobile phones. In those markets new features and time to market are critical, while extended reliability is not. The automotive industry is the exact opposite. Product cycles historically have been five to seven years. Most changes are incremental, and reliability is considered essential. Even in non-safety critical areas, product recalls are costly.
As the rate of change accelerates in this sector, it has caused disruption across the entire supply chain.
“If you talk about ADAS, whether it’s aftermarket ADAS or native ADAS, they are starting to freak out,” Knoth said. “You look at the explosion of LIDAR, huge growth of sensor processors, infotainment exploding all over the vehicle, and ADAS. The automakers are suddenly facing integration challenges, huge schedule pressure, the fact that they can’t keep designing the same part in 180nm. They have to start talking about 16nm finFET or 28nm. These guys have never dealt with signal integrity. They’ve never dealt with (semiconductor) reliability. They’ve never dealt with double or triple patterning. Likewise, from the software perspective, there is a huge integration challenge on top of things like security. These things present a huge design challenge, and even bigger verification challenge.”
Carmakers vs. aftermarket
This is one of the reasons carmakers insist autonomous driving technology needs to be designed in from the start. Ford recently said it would have fully autonomous vehicles for ride-sharing on the road in 2021. But Jim Buczkowski, technical fellow and director of electrical and electronics systems at Ford, maintains that aftermarket autonomous technology is not the way to go because the systems are so complex.
Others disagree. “I don’t see the traditional automotive companies talking about [aftermarket autonomous], or even preparing for that, which they should be,” said Joseph Vitale Jr., global automotive practice leader at Deloitte & Touche. “I’ve been beating the aftermarket horse with them for 10 years. It just amazes me that this industry has given up so much of the value added, whether it is insurance or the majority of financing.”
However, there are some automakers aggressively addressing aftermarket autonomous vehicles. “They are studying it, they are analyzing it, they are thinking about everything from how to be involved in the whole service side of [autonomous] because of the notion around fleets,” Vitale said. “If fleets become a significant portion of the market, they are going to be competing on price if they only do what they do today. There is a huge service component, and they can capture that. There is a huge retrofitting component associated with it. So they are now recognizing that they need to really look at, in a significant way, the aftermarket and service part of it. A big part of why they didn’t before was because of channel conflict with the dealers.”
Interestingly, when it comes to the design of autonomous systems themselves, the OEMs will have to pull that back from the Tier-1 and Tier-2 suppliers. “It’s way too integrated,” he said. “How do you determine who’s at fault for a warranty claim? Is it the software provider, is it the OEM? What’s going to be interesting to see is the chip manufacturers, the OEMS, the startup technology companies, truly collaborating in a significant way—in a very different way than they have historically done. The question is going to become whether the OEMs can adapt to that type of collaboration because you’re not going to control or dictate or take IP from them. The chipmakers have a huge opportunity here.”
What Can Go Wrong In Automotive
Experts at the Table, part 1: Security, diagnostics, standards and the future of autonomous vehicles.
Executive Insight: Wally Rhines
Mentor Graphics’ CEO digs into how the automotive industry and IoT security will change semiconductor design.
Auto Security And Technology Questions Persist
Fallout could be slower adoption of autonomous vehicles as ecosystem proceeds with caution.
The Higher Cost Of Automotive
Suppliers looking to enter this market pay a premium in design time, certification and verification requirements.
Lawyers, Insurance And Self-Driving Cars
Semiconductor companies have to contend with more than technology when it comes to the automotive market.