Fueled by the high-profile efforts of technology giants like Google, major manufacturers like General Motors and Tesla, and new auto industry entrants like Uber and Apple, the driverless car revolution is here. Tomorrow’s self-driving cars will rely on a combination of familiar driver-assist features coupled with newly developed computer hardware and software, working together to make driving decisions based on input from sensors, cameras, and the surrounding environment. In much the same way, the legal issues related to autonomous vehicles will be addressed using a combination of existing law and new legislation. This article explores the steady climb of vehicle automation, the road that lies ahead, and an overview of the patent landscape impacted by this autonomous vehicle revolution. Part 2 will focus on safety regulations, product liability, data security, and other areas.
The Technology
Driverless v. Driver-Assisted Cars. There are two schools of thought behind the autonomous vehicle revolution. The first school—led by Google and Uber—maintains that truly autonomous vehicles cannot be achieved incrementally; instead, full automation is necessary. Human drivers in less-than-fully autonomous cars must constantly pay attention to be able to address sudden and unexpected situations—the very situations that easily distracted human drivers are ill-suited to handle. According to this view, the more control given to the computer, the less a human has to contribute, and the safer the car becomes.
The other school of thought—embraced by Tesla, General Motors, Mercedes-Benz, Audi, Volvo, Toyota, and Ford—maintains that autonomous technologies will continue to require driver assistance. In support of their view, these companies point to the successful integration of numerous autonomous technologies into today’s cars, ethical quandaries presented by fully autonomous cars, and the reluctance of drivers to abandon control of their vehicles. This perspective also has historical support: most other technological revolutions, including the industrial, electrical, computer, and Internet, have arrived in incremental improvements, not in lightning strikes.
At the heart of this debate is the lurking question: can, and should, sensors and a computer truly replace human senses and control? Hypothetical situations abound where the car’s computer could be confronted with difficult choices. Example: suppose a car is about to hit a group of pedestrians. Should the car veer off the road, into a median, or even into oncoming traffic? Proponents of driverless cars argue that such hypotheticals are contrived and unlikely in the real world, and that a human being would not have the time to make a judgment call anyway under such circumstances. Proponents of driver-assisted cars understand that, regardless of the ethical implications of such decisions, most drivers would not want to cede them to a computer. It is difficult to predict which school of thought will ultimately prevail, and federal regulation is required to avoid the patchwork of laws that will result as each state issues its own pertinent regulations. Indeed, the California DMV has reportedly issued draft regulations requiring automated cars to have a licensed driver positioned behind a steering wheel, tacitly favoring the “driver-assisted” school.
The Future Is Now. Examples of autonomous features currently on the market include:
• Automatic/intelligent braking or forward collision control senses an imminent frontal crash, warning the driver, adjusting the brake force, and/or engaging the brakes. The GMC Yukon and Sierra models’ “Intelligent Brake Assist Technology” senses an emergency stop and enables higher braking force before the driver fully depresses the pedal.
• Automatic headlight control senses ambient light levels and turns on or adjusts headlights. Toyota’s “Automatic High Beam” detects nearby light sources like tail lights or headlights and automatically switches between high and low beams.
• Parking assistance features range from sounding chimes when the car approaches an obstacle to autonomous parallel parking. Initially popularized in the U.S. by Toyota, newer, more advanced systems, for example by component manufacturer Bosch, enable the driver to leave the car and activate the feature using his or her smartphone.
• Blind-spot monitoring warns the driver when a vehicle enters a car’s blind spots. Volvo’s “Blind Spot Information System” produces a visible alert when lenses mounted on its doors detect a car in a driver’s blind spot as the driver switches lanes.
• Cruise control, as we know it, maintains a car at constant speed by taking over the car’s throttle. Adaptive cruise control uses radar or lasers to determine the location of surrounding cars to adjust speed to maintain a safe distance. BMW’s X5 now includes a traffic jam assistant where the car automatically speeds up and slows down in stop-and-go traffic.
• Traffic-sign recognition uses cameras to identify road signs, displaying the information to the driver or, as will be the case with Ford’s upcoming “Intelligent Speed Limiter,” adjusting the throttle to maintain a legal speed.
• Lane-departure warnings and auto-steering, like in Tesla’s Model S and X, use visual, laser, or infrared sensors to detect lane markers. If the car drifts out of its lane, the system either warns the driver or automatically adjust the car’s direction.
Looking Forward—Driverless Cars. Autonomous technologies make cars safer: fatalities involving 2011 model-year cars are nearly half as likely as fatalities involving those from 2008. As autonomous features become standard equipment, and the opportunity for unpredictable, distracted, and reckless human behavior narrows, these benefits should continue to accrue. Coordination between vehicles—vehicle-to-vehicle (“v2v”) technologies—promises to makes cars even safer. In a v2v system, all cars intelligently communicate with each other via short-range radio, and vehicle-to-vehicle collisions are ideally averted.
The autonomous car revolution promises other benefits as well, such as alleviating traffic. Where all cars play by the same rules and communicate, stop-and-go traffic will move faster. Where cars can be trusted to stay within lanes, highway lanes can be redrawn with narrower lane markings, thereby allowing additional lanes to carry more cars. Taken to its logical extreme, v2v technology may one day eliminate traffic congestion altogether: even a freeway packed to capacity would flow if all cars are able accelerate and decelerate nearly simultaneously.
Companies are making large investments in pursuit of the driverless car. Since 2012, Google’s fleet of driverless cars have together logged over 1.2 million miles, stopped at 200,000 stop signs, encountered 600,000 traffic lights, and “seen” 180 million vehicles. Uber recently hired 40 researchers and scientists from Carnegie Mellon’s National Robotics Engineering Center to work on driverless car technology in Uber’s cutting edge Pittsburgh research facility. Newcomer Faraday Futures, backed by a yet undisclosed investor to the tune of $1 billion, hired Nick Sampson, the head of vehicle and chassis engineering for the Tesla Model S, as well as Dan Reckhorn, Tesla’s former senior manufacturing executive. Faraday is actively looking for space to build its factory, with a high-end semi-autonomous car slated to hit the roads by 2017.
And Apple’s car development project, Project Titan—while still shrouded in mystery—recently received a boost with the hiring of Jonathan Cohen, who was previously the director of Nvidia’s deep learning program. Deep learning uses multi-layered neural computer networks to solve perceptual problems. Practical examples include vehicle, pedestrian and landmark identification for driver assistance. For these companies, it is the pursuit of a fully driverless car, as opposed to being merely driver-assisted, that continues to push their research and investment in autonomous vehicle technology.
Patents and Patent Litigation
It has been estimated that driver-assisted and driverless cars will be a $42 billion market by 2025. As mobile phone manufacturers know all too well, where an industry enjoys a boom in technological innovation and product evolution, so, too, may that revolution (and accompanying market opportunity) become a target of intense patent litigation. As driver-assisted and driverless cars hit the road, the question is not if patent litigation will follow, but who will be filing the cases and how hardware and software suppliers in the automotive industry can protect themselves.
Who Owns the Patents? Patent ownership in the driver-assisted and driverless space ranges from automotive mainstays (like General Motors) to newcomers (like Google) to non-practicing entities (NPEs). According to a 2015 Thomson Reuters report, Toyota, General Motors, and Hyundai file the most “autonomous car” patents, with General Motors having taken the lead in the past two years. Google is an important player as well and, according to a different survey, received the most patents between 2009 and 2014, followed by a non-practicing entity, Eagle Harbor Holdings. With respect to patents on more incremental autonomous features—such as those discussed above—Bosch leads the pack, followed by Audi, BMW, Continental, Daimler, Valeo and Volkswagen.
What Do These Patents Cover? Automation in cars is not new. Issued patents reflect that the core of many autonomous features being implemented today was invented and patented decades ago. For example, in 1975, General Motors applied for U.S. Patent No. 4,082,158, on a “[d]ifferential pressure power road speed control system,” i.e., cruise control. As another example, in the mid and late 1980s, General Motors applied for a patent on an adaptive cruise control system that uses video cameras to enable a vehicle to follow a second vehicle at a predetermined distance (U.S. Patent No. 4,987,357) and a patent on a lane detection system used for autonomous vehicle control (U.S. Patent No. 4,970,653).
Today’s patents continue to improve upon these concepts. Google’s U.S. Patent No. 8,473,144, filed in 2012, predicts the lateral distance between two passing cars well before they are adjacent to each other, and then adjusts this distance based on the characteristics of the cars. A car with this autonomous feature may leave more room when passing a larger truck than when passing a passenger car. As another example, Google’s U.S. Patent No. 8,195,394, filed in 2011, addresses not only the detection of moving objects near an autonomous vehicle, but using machine learning to identify the moving object, decide whether it is a car or a pedestrian, and control the vehicle based on this information.
Recent Patent Litigation. Patent litigation in the driver-assisted space is here. NPEs have sued on patents allegedly covering autonomous features against car manufacturers. For example:
• In July 2014, Adaptive Headlamp Technologies Inc. sued BMW on a patent directed to swiveling headlights “whose positions automatically change due to changes in steering angle and pitch of the vehicle after a threshold minimum is passed.” Adaptive Headlamp Techs., Inc. v. BMW of N. Am., LLC, 1:14-cv-962 (D. Del.).
• In December 2012, Cruise Control Technologies, a subsidiary of Empire IP, sued 15 car companies—including Porsche, Ford, BMW, and Hyundai—asserting patents allegedly covering cruise control systems. Cruise Control Techs. LLV v. Toyota Motor N. Am., 1:13-cv-00086-GMS (D. Del.). Notably, in June 2015, claims of the asserted patents were found invalid as anticipated by prior art, including a 1989 report by the National Highway Traffic Safety Administration (“NHTSA”).
• In mid-2012, American Vehicular Sciences sued Toyota on two patents for automatic crash notification systems. American Vehicular Sciences LLC v. Toyota Motor Corp., et. al., 6:12-cv-00404 (E.D. Tex.); American Vehicular Sciences LLC v. Toyota Motor Corp., 2:14-cv-13015 (E.D. Mich.).
Future Patent Litigation Landscape. Many in the traditional automotive industry have indicated that they have no intention of asserting their driver-assisted and driverless patents. Earlier this year, Toyota announced that it was making available over 5,000 hydrogen fuel cell patents on a royalty-free basis, and Ford and BMW indicated that they are opening up their electric vehicle patent portfolios for licensing. Similarly, Elon Musk, CEO of Tesla, announced in June 2014 that Tesla will apply an “open source” philosophy to its patents. 2012 saw the birth of the Auto Harvest Foundation, an IP consortium of over 250 manufacturing and R&D organizations from industry, government and academia, whose mission has been to facilitate patent open sourcing and cross-licensing deals.
In recent years, suits by NPEs against the automotive industry have spiked, increasing from 17 lawsuits in 2009 to 107 lawsuits in 2014. NPEs have presented a challenge to the car industry even as car technology evolved from its nascent days to today’s modern cars. In 1911, Henry Ford defeated the original automobile NPE—patent lawyer George Selden—when Mr. Ford prevailed on appeal against Mr. Selden’s combustion engine patent. NPEs will most likely continue to present a challenge as the technology advances from individual autonomous features to driver-assisted cars and, eventually, to driverless cars.
Potential Patent Defensive Strategies
Since the beginning of the 21st century, high technology products incorporating computer hardware and software have been a frequent target for patent litigation. There is no reason to believe that driverless and driver-assisted technology will be immune from such litigation. Fortunately, both long-standing and recent defensive strategies should work equally well in this arena.
Patent Eligibility Attacks Under Alice. One strategy car companies should consider is a patent eligibility attack under the Supreme Court’s decision in Alice Corp. Pty. Ltd. v. CLS Bank Int’l, 134 S. Ct. 2347 (2014). In the realm of driver-assisted and driverless cars, many technological advances focus on automating acts previously performed by a human. Accordingly, patent claims on this technology may boil down to little more than instructing a computer to perform a task that human drivers have performed for decades (e.g., changing lanes). Such claims may be vulnerable to attacks under Alice because they arguably do nothing more than implement an abstract idea on general purpose computer equipment.
Of course, the actual claim language will drive the analysis. Entities that do not engage in substantial research may attempt to prosecute and later assert patents listing desirable features of a driver-assisted or driverless car (“a method of organizing human activity” under Alice and possibly an abstract idea) and merely instruct one of ordinary skill to implement these features with a computer. Under Alice, the mere recitation of computer implementation is insufficient to confer patent eligibility on an otherwise patent-ineligible abstract idea. On the other hand, entities that engage in research and development should be able to flesh out and claim the details of how a new autonomous feature is implemented using, for example, a combination of sensors, software, and actuators. These claims would likely include substantive limitations in addition to generic computer implementation that may be sufficient to confer patent eligibility. Therefore, the application of Alice in the realm of driver-assisted and driverless cars has the potential to differentiate weak patents owned by non-practicing entities from legitimate patents claiming true technological advances owned by General Motors, Google, and the other pioneers in the field.
To illustrate, in the Cruise Control Technologies litigation mentioned above, claim 1 of the asserted patent, directed to a car’s cruise control system, requires only that it would be desirable for a “human operator” to be able to engage cruise control at a particular speed, and to see the speed at which it is set. It may be argued that this is simply “human activity”—an abstract idea—ripe for an attack under Alice, and that requiring the speed to be maintained “automatically” does not transform this abstract idea into patent-eligible subject matter.
Prior Art Attacks. Some autonomous driving technologies date back a decade or more. Carnegie Mellon’s first autonomous vehicle, “Terregator,” was demonstrated publicly in 1984. The Eureka Prometheus Project dates back to the late 1980s. More recently, the 2004 Defense Advanced Research Projects Agency (“DARPA”) Grand Challenge later demonstrated driverless car technology. Indeed, Ford recently invalidated enhanced cruise control patents by asserting prior art developed by the NHTSA in 1989.
Attacks on Method Claims Under Limelight. Car manufacturers should consider the Federal Circuit’s en banc and unanimous decision in Akamai Techs., Inc. v. Limelight Networks, Inc., 797 F.3d 1020 (Fed. Cir. 2015), holding direct infringement exists “when an alleged infringer conditions participation in an activity or receipt of a benefit upon performance of a step or steps of a patented method and establishes the manner or timing of that performance.” This holding may render it more difficult for manufacturers to argue divided infringement where a driver performs one or more steps of a claimed method (manually drives around an obstacle) while the manufacturer’s technology performs the other steps (analyzes data collected regarding the driver’s action to teach the car to avoid such obstacles in the future). Plaintiffs may argue that the steps performed by the driver are necessary for the driver to benefit from the autonomous feature and so should be attributed to the manufacturer to show direct infringement.
On the other hand, defendant OEMs (who manufacture sensors or other technology that perform one or more steps involved in an autonomous feature) may still be able to rely on the Supreme Court’s holding in Limelight Networks, Inc. v. Akamai Techs., Inc., 134 S.Ct. 2111 (2014), that indirect infringement requires a direct infringer who performs all claimed steps. An OEM may be able to argue that a driver is not required to perform the allegedly infringing step (processing data received from the OEM’s sensor in the claimed manner) to benefit from other uses of the OEM’s product (the data received from the OEM’s sensor is also used by other, non-infringing features). Thus, the OEM may be able to argue that no indirect infringement exists because the plaintiff cannot attribute to a single entity the combination of steps performed by the OEM’s product and those performed by the manufacturer or driver.
Post-Grant Proceedings. Car companies in driverless and driver-assisted technology litigation can also take advantage of other “tried and tested” defensive strategies, such as post-grant patent challenges made available under the America Invents Act. For example, in both the American Vehicular Sciences and Cruise Control Technologies litigations, the defendants obtained dismissals aided by taking advantage of post-grant attacks that successfully invalidated patents asserted against the automotive industry.