Navigant Research Blog

Cybersecurity for Self-Driving Cars Needs a Confidence Boost

— July 29, 2015

Highly detailed and accurate mapping data will be critical to the technical success of future autonomous vehicles. However, in order for consumers and regulators to accept vehicles that pilot themselves to a desired destination, they will need to have a great deal of trust in the technology. That trust is currently in serious danger of being eroded by an ongoing series of computer network attacks, including one demonstrated recently on Wired.com. The need to bolster automotive cybersecurity is one of the factors driving Mercedes-Benz, Audi, and BMW to jointly acquire Nokia’s Here mapping division.

Nokia was an early leader in the field of bringing high quality maps to mobile devices with its 2007 acquisition of Navteq, but the world of mobile cartography has shifted dramatically since then. With mapping apps from Google and Apple joining incumbents such as TomTom and Garmin, along with the rapid development of autonomous driving capabilities, the expectations for map data has increased exponentially. Cartographic data needs to be kept continuously updated through fleets of camera and sensor-equipped vehicles, in addition to crowd sourcing for real-time information. Unlike traditional automotive navigation systems that might get updated annually at best, this fresh data will need to be pushed to automated vehicles as soon as it’s ready.

The big three German premium brands are all expected to be on the leading edge of introducing level 2 automation capabilities and are likely to ramp up automation as soon as  technology and the market allows. Navigant Research’s Autonomous Vehicles report projects that nearly 50 million vehicles with some form of autonomous capability will be sold globally by 2030. One of the key drivers for the move to automation is the desire to reduce accidents to near zero by taking humans out of the driving control loop.

Gaining Trust

Before that can happen, everyone will need a much higher degree of confidence in the security the software and electronic systems, something that is getting more difficult by the day. For several years now, computer security researchers have been demonstrating increasingly sophisticated cyber attacks against vehicles, with the most recent coming from Charlie Miller and Chris Valasek. Miller and Valasek were able to remotely take control of a 2014 Jeep Cherokee through its telematics system, manipulate the audio system, wipers, steering, and even shut down the engine as it was driven by Wired reporter Andy Greenberg. These attacks are not trivial and are not yet widespread, but as we’ve seen from recent attacks against the U.S. Office of Personnel Management and retailers such as Target and Home Depot, the more attackers learn about the systems, the more attack vectors they find.

Automakers are hiring some of these same security researchers and creating teams solely focused on securing their in-vehicle networks. When automakers outsource control systems or data such as maps to suppliers, they often get only a black box that they can hook into without access to source code. Recognizing that they will be increasingly liable for the performance of advanced systems, they are now bringing some of the work back in-house where they can control it. Daimler AG CEO Dieter Zetsche recently acknowledged that security concerns were one of the reasons his company was partnering with its chief rivals to purchase Here maps. Similar concerns have prevented numerous automakers that have been approached by Google from adopting its autonomous driving software and developing their own code instead. Unless Google is willing to give up control of its software to automakers, it may only get adopted by lower tier companies without the resources to develop their own autonomous systems.

 

University of Michigan Opens Autonomous and Connected Vehicle Proving Ground

— July 20, 2015

For decades, the University of Michigan in Ann Arbor has been one of the top training grounds for the engineers that power the American automotive industry. However, with the focus on Google and its self-driving driving cars in recent years, the center of gravity seems to have shifted westward to Palo Alto, California, and Stanford University. With the grand opening of the Mobility Transformation Center (MTC) in Ann Arbor, however, Michigan hopes to regain its place in the pecking order while driving automated and connected vehicle technology forward.

Navigant Research’s Autonomous Vehicles report projects that nearly 50 million vehicles with some form of autonomous capability will be sold globally by 2030. Those vehicles will need to function reliably in a broad range of environments and coexist with the existing human-driven fleet as well as technologies from many different companies.

The Solution

The MTC, also known as Mcity, is a 32-acre dedicated proving ground built on the university’s North Campus, which was formerly the site of pharmaceutical giant Pfizer’s Ann Arbor R&D center. Mcity will be operated by the University of Michigan’s Transportation Research Institute (UMTRI) in partnership with more than a dozen automakers, suppliers, insurers, and the U.S. and Michigan departments of transportation. The facility includes road surfaces paved with a variety of materials along with unpaved roads and features such as signs, fire hydrants, crosswalks, roundabouts, overpasses, and tunnels. Movable building facades will enable the testing of a variety of real-world scenarios.

Partner companies and students of the university will all have access to the facilities to develop and validate new transportation technologies. Among the participating companies are Toyota, Honda, General Motors, Ford, Robert Bosch LLC, Delphi, Qualcomm, and State Farm Insurance. Each of the manufacturers have R&D centers and test facilities of their own in or near southeast Michigan, but those facilities are closed to outsiders. At Mcity, the companies will have a common ground where they can test individually and collaboratively to ensure that their respective systems can coexist.

And, unlike in California, where weather is rarely an issue that test drivers have to contend with, Mcity will also provide a platform for testing under all of the environmental conditions faced by drivers, including winter snowfall and road salt. Most current automated driving systems do not function as well or, in many cases, at all if the weather is less than ideal. This is a problem that will need to be addressed before systems are deployed to customers.

Preparing for V2X

Navigant Research’s Connected Vehicles report estimates that more than 80% of new vehicles sold in North America and Europe will be equipped with vehicle-to-external (V2X) communications capability by 2025. While initial deployments are expected to focus mainly on vehicle-to-vehicle transmission of basic safety messages, the potential to expand the data transfer to pedestrians, cyclists, and infrastructure could have a significant impact on reducing congestion and accidents. Among the features of Mcity are stationary and motorized pedestrians and cyclists that can be equipped with V2X transponders and that can also be used for testing the sensing capabilities of automated vehicles.

The collaborative efforts at Mcity will also include the development of performance and reliability standards for communications and automation systems. While much of this work will likely become the basis of new SAE industry standards, it could also feed into future federal motor vehicle safety standards, which do not currently address autonomous driving. Work at this new collaborative testing facility is scheduled to begin this summer.

 

Reliable Service Parts Critical to Autonomous Driving Future

— June 30, 2015

Short_Bridge_webThanks to advances in materials that increasingly avoid corrosion, modern engineering and manufacturing processes that improve build quality, and electronics that improve performance and efficiency, cars now last longer than ever. The average age of the more than 200 million cars on American roads today is nearly 11.5 years, and 20- to 30-year old machines are shockingly common. Despite how well-built vehicles have become, parts still eventually break or wear out and need replacement; this includes the sensors that control the vital systems in modern vehicles.

As cars become increasingly automated, the number of sensors has grown dramatically, and they need to be functional and reliable. This potentially poses a significant problem for vehicles after they are out of warranty or out of production. My friend Richard Truett, engineering reporter for trade publication Automotive News, buys older vehicles, repairs or restores them, drives them, and sells them before moving on to the next vehicle.

While most of Richard’s vehicles are older British sports cars that predate the electronic age, he recently bought a 1988 Pontiac Fiero with relatively low mileage that was in need of his TLC. As Richard went through the car from the wheels up, he attacked the engine control electronics that were keeping the car from running properly. In the process, he discovered issues that could pose serious problems for future automated vehicles. It’s actually not uncommon for people to manage to get around for months or years with the tell-tale “check engine light” illuminated, usually indicating some sort of sensor fault. For automated vehicles, that is less likely to be an option because of the dependence on sensors for basic functionality.

Lessons to Be Learned

Standard industry practice after a vehicle goes out of production is for automakers and suppliers to license the production of replacement service parts to third-party manufacturers. In many cases, these service part manufacturers will also reverse engineer the original parts and produce compatible replacements. What Richard discovered when trying to replace the oxygen sensors and spark plugs on his 27-year-old sports car was that compatibility and functionality were often not a sure thing. The electronic systems in the Fiero were comparatively primitive by 2015 standards, but brand-new components as basic as an oxygen sensor or throttle position sensor fail out of the box—that’s a bad sign, and these aren’t even safety-critical systems.

The sensors being used for automated driving systems are far more advanced, and the technology is evolving rapidly, so components are less likely to stay in production with the original manufacturer than they were 3 decades ago. It may not even be possible for third-party manufacturers to replicate the original parts, and if they do, they may not perform to the same standard, thus hampering the performance of safety-critical automated systems.

Navigant Research’s Autonomous Vehicles report projects that by 2030, 40% of new vehicles will have some sort of autonomous driving capability built in. Those vehicles will be totally dependent on sensors that must provide accurate and reliable information about the world around that vehicle in real-time. Before we become overly reliant on these systems to get us where we need to be on our daily rounds, manufacturers need to sort out solutions that will ensure a more robust and reliable stream of service parts. Perhaps this should even be part of the safety regulations that govern automated vehicles. There are still many fundamental questions to be answered before you can summon an autonomous Uber car from your wrist—and service parts is just one.

 

Uber-CMU Deal Highlights Transition of Autonomous Technology

— June 16, 2015

FT_Advisory_webIn February 2015, Uber and Carnegie Mellon University (CMU) announced a strategic partnership to develop autonomous driving technology. It caused a bit of a stir in the auto industry because until that time Google was the only non-automotive company that had put serious effort into advancing automotive technology. Uber, however, identified a technology that could make a big difference to its bottom line in the long term and had decided to become an active participant in the engineering.

Then, in May, an article in The Verge claimed that Uber had gutted CMU’s robotics lab. Within a week or two, more articles were published about how Uber had “poached” staff and that CMU was “in a crisis”—all very dramatic. It turns out that Uber had indeed opened its own engineering center in Pittsburgh, just up the road from CMU, and that a good number of engineers and scientists had been given lucrative offers to work there.

CMU, however, seemed somewhat less excited about the situation than the journalists. Gradually, more reasoned articles made it into print, pointing out that there are many other more positive aspects to this story. Yes, Uber had diverted a large chuck of the CMU staff to work full-time on its autonomous vehicle project, but CMU enhanced its reputation as the source of advanced robotic development work. Additionally, the City of Pittsburgh has the potential to challenge Silicon Valley as an alternative site for automotive innovation. (It probably doesn’t hurt that SAE International is also based nearby.)

Universities have always been at the forefront of the latest technology development, and their role is usually to then spin off and partner with commercial companies that can take concepts into production. This is precisely what is happening now in Pittsburgh—another indication that autonomous driving is close to entering production and becoming a reality. There are plenty of new challenges for CMU to explore, and the establishment of an advanced automotive engineering center in Pittsburgh can only be good for the future because it will attract talented researchers who eventually want to end up in high-paying jobs.

 

Blog Articles

Most Recent

By Date

Tags

Clean Transportation, Electric Vehicles, Policy & Regulation, Renewable Energy, Smart Energy Practice, Smart Energy Program, Smart Grid Practice, Smart Transportation Practice, Smart Transportation Program, Utility Innovations

By Author


{"userID":"","pageName":"Autonomous Vehicles","path":"\/tag\/autonomous-vehicles","date":"7\/31\/2015"}