Navigating driverless cars

December 2014

By Emma Poole, Lawyer and Researcher, Melbourne, Australia

In September 2014, The Economist debated whether “completely self-driving cars” are “feasible in the foreseeable future”. The 32 percent of readers who voted “no” have obviously missed the news that this staple of science fiction has already turned into science fact.

Google’s self-driving car project has dominated coverage of the emerging driverless car sector. (Photo: Google)

Whether we call them driverless, self-driving or autonomous, these cars have navigated downtown Parma and driven from Italy to China almost unaided (a human being had to drive through Moscow and pay the tolls). A Mercedes-Benz S-Class travelled between Mannheim and Pforzheim without driver input in August 2013 and, most famously, the Google Self-Driving Car Project has now completed over 700,000 test kilometres. The prototype cars cannot always navigate potholes, see a traffic light with the sun behind it or drive in the rain, but driverless vehicles are more than feasible, you can buy them.

What is a self-driving car?

A car is self-driving if it can operate without the “active control and continuous monitoring” of a human being. According to the US Department of Transportation, this means that the car’s operation does not require driver input to control the steering, acceleration or braking. However, automation is really a question of degree. The National Highway Traffic Safety Administration in the US has identified five “levels” of automation (see box).

The five levels of automation:

  1. No-Automation - the driver is in complete control;
  2. Function-Specific Automation - a function assists the driver (electronic stability control or brake-assist technology);
  3. Combined Function Automation - two functions are designed to work together in certain scenarios – such as “adaptive cruise control … with lane centering”;
  4. Limited Self-Driving Automation - enabling the driver to give up control of the car in certain scenarios, with sensors to trigger the need to return control to the driver;
  5. Full Self-Driving Automation - the car performs all driving functions and monitors road conditions without any input; a person will determine the destination and then give up all control of the car.

Already here?

Fully automated vehicles are now commonplace in certain controlled environments. You may have already travelled on a segregated railway or guideway (also known as light rail) system in Vancouver, London, Singapore or between terminals at airports around the world. The Park Shuttle vehicles in the Netherlands use dedicated bus lanes and mining company Rio Tinto has a fleet of autonomous trucks operating at its Pilbara iron ore mine in Western Australia. At the same time, an increasing number of mass-produced cars now incorporate semi-autonomous, driver assistance functions such as assisted steering, parking or braking; drowsiness detectors; and devices to detect, and help avoid, potential collisions with other vehicles or pedestrians.

The next generation of semi-autonomous car technologies coordinate two or more functions. Examples include Mercedes Benz’s adaptive cruise control with steering assistance; Volvo’s traffic jam assistance, allowing cars to automatically brake and follow cars in slow moving queues; and Audi’s piloted parking. Many other carmakers including General Motors, Ford, Volkswagen, Nissan, Toyota and BMW are also testing advanced driver assistance systems (ADAS).

How they work

Exane BNP Paribas, an investment company, predicts that the tech and telecommunication sectors will see more benefit from the expected growth in the connected car market than the traditional automobile industry. The reason for this is simple – something is driving driverless cars and that something is software processing huge volumes of data.

Driverless cars operate by amassing information collected from cameras, sensors, geo-location devices (including radar), digital maps, navigation programming and communication from other connected vehicles and infrastructure. Operating systems and software then process this information and coordinate the mechanical functions of the car. These processes mimic the hugely complex task human drivers undertake when they monitor the road, the car and themselves in order to drive. Recent examples include Google’s patent on reading traffic lights or Tesla’s latest saloon, which adjusts its speed to comply with road signs.

Key benefits – access and safety

Fully autonomous cars will negate the need for driving restrictions relating to age and ability because the only prerequisite for making a journey may be the ability to program a destination. A six year old may take himself to school or an older person stay independent for longer. This increase in access to mobility should facilitate the active participation of the 22 percent of the world’s population who will be over 60 in 2050. To highlight the potential of driverless cars to assist people with disabilities a Google car has recently driven a blind man to Taco Bell for a take-out.

Since 2011, London Heathrow Airport has been running a fleet of
driverless electronic vehicles on a dedicated guideway. Designed
by UK-based company Ultra PRT, these autonomous pods ferry around
1,000 passengers each day between the airport’s Terminal 5 and its
car parks. (Photo: Ultra PRT)

Increasing road safety will be the most critical benefit of driverless cars. At the moment there are 1.24 million road traffic deaths worldwide every year (50 percent are pedestrians, cyclists and motorcyclists) and road traffic accidents are the number one cause of death for those aged between 15 and 29 years. Around 90 percent of all traffic accidents are caused by human error whether distraction by phone calls or texts; drowsiness; intoxication; or the effects of medical impairments or conditions. Fully automated vehicle technologies offer the potential to circumvent human driver error completely and combine robotically rapid responses with 360-degree perception. Equally, semi-automated, driver-aware vehicles could use sensors to detect changes in heart rate or skin temperature and then trigger extra safety or assistance measures. General Motors is already testing eye and head-tracking technologies to check for signs of drowsiness.

Coming soon

The first radio-controlled cars drove in the US in the 1920s but development in the sector has been slow until very recently. Research by the Boston Consulting Group suggests that R&D investment by the automobile industry generally (including into driver assistance functions) has rapidly expanded in the past four years.

As part of this pattern, purpose-built testing sites are popping up around the world: from AstaZero, Volvo’s 2 million square meter (21.5 million square feet) testing facility in Sweden to a custom-made test town built outside Ann Arbor in Michigan in the US. Testing of autonomous vehicles on public roads has been facilitated by amendments to the Vienna Convention on Road Traffic and the introduction of legislation in the UK, France and various US states (including California, Florida and Nevada).

Public-sector tests include the EU’s V-CHARGE consortium’s automated valet system that will park and charge electric cars, the three LUTZ Pathfinder “autonomous pods” due to explore the pavements of Milton Keynes in early 2015 and “vehicle platooning” by the Safe Road Trains for the Environment (SATRE) project, funded by the European Commission (enabling cars to connect to, and follow, a lead vehicle driven by a professional driver in a particular highway lane).

The G-word

Google’s self-driving car project has dominated coverage of the emerging driverless car sector. The New York Times outed the secret project in October 2010 and since then the resources of the world’s third-largest company have been deployed to put Google at the center of autonomous vehicle R&D (including hiring influential scientists and engineers such as Sebastian Thrun).

This project from the company’s Google X division forms part of Google’s broader strategy of investment in the early stages of new technologies, demonstrated by a huge surge in patent applications. In 2013, Google was awarded around 2,000 US patents, almost double the number of all its previous patents and the company has clearly learned the importance of owning intellectual property (IP) in the building blocks of new technological sectors from the smartphone and semi-conductor patent wars.

This early investment gives Google a number of strategic options. It can build the first mass-market driverless cars or license the technologies that underpin the sector to manufacturers (hoping that at least a few are adopted as standards). Equally the company could follow the strategy it adopted with the open access release of its Android mobile operating system and continue to harvest the data generated by eager users of its systems.

The big problem – code, ethics and liability

As driverless cars emerge onto roads there will be a shift in legal responsibility for driving from drivers to manufacturers (and their suppliers). As the CEO of Renault-Nissan, Carlos Ghosen, suggested “The problem isn’t technology, it’s legislation, and the whole question of responsibility that goes with these cars moving around.”

A car may end up in the wrong place due to an error in a digital map, a sensor malfunction, a glitch in the navigation software or a combination of all three. Another’s programming may brake to avoid a pedestrian, killing the traveler or the occupant of the following car (in a variant of the famous “trolley problem” and Isaac Asimov’s first rule of robotics). Yet another car may be subject to cyber-attack through an undiagnosed flaw in the underlying open-source architecture of the connected world (such as the recently discovered HeartBleed and Shellshock flaws).

Driverless cars will inevitably break, crash and hurt people. When they do, we will need to understand who to hold accountable: whether it be the traveller, the manufacturer, the various suppliers or the programmers who wrote the underlying code. This will be a question of software IP and as the recent Alice v CLS Bank decision of the US Supreme Court (see Alice v. CLS Bank: United States Supreme Court Establishes General Patentability Test) has shown us, the status of IP protection for software is vexed and in flux. While this question arises in relation to all kinds of digital architecture, driverless cars will be the one type of connected device where legal issues (such as the effects of collaborative creation, device interoperability, digital circumvention and the ownership of APIs) will almost always be life or death.

Same, same but different

The first industries and business models disrupted by driverless cars may be involved in point-to-point transport and delivery like public transport, taxis, car hire, couriers, trucking and logistics. Improved safety will reduce demand for all the sectors that assist after road traffic accidents – towing, vehicle repair, auto parts suppliers and even ambulances and emergency services. Governments may lose revenue from parking charges or fines for infringements but develop new revenue streams such as GPS-based road pricing (a charge for using particular roads). Drivers may no longer need collision insurance but, as John Villasenor at the Brookings Institution has pointed out, as travellers or operators of driverless cars, they may need expanded forms of product liability insurance.

If the systematization of driverless cars becomes a reality (think of a real life version of the MAG-LEV electrical/magnetic horizontal and vertical roadway system that automotive designer Harold Belker helped design for the film Minority Report) then vehicles could become the ultimate connected devices allowing “Smart Cities” with integrated networks and infrastructure to move populations en masse. Mr. Belker described the overall goal of his fictional system as “individual transportation within a mass transport system” and it is possible to imagine that automated cars in a connected network would reduce the number of vehicles required to meet transport needs. Existing car share platforms provide the software and online environments necessary to match available cars with waiting travellers. Uber’s digital architecture will clearly move seamlessly into this space and its economic model will only become more attractive as its most expensive input and liability, drivers, are phased out.

The industry that is likely to be most fundamentally disrupted by driverless cars is car making itself. Mass production of automated electric vehicles by additive manufacturing processes like 3D printing could transform the economics of car ownership, from financing, depreciation and fuel to insurance and maintenance. In that circumstance, it is not clear what design and branding elements will attract consumers to self-driving cars, except that they will affect how the car looks and feels to travel in, rather than to drive. Some may miss the romance of hot metal and the open road but as the Wall Street Journal memorably put it: “Give people a button that says ‘Home’ and I guarantee they will push it”.

The WIPO Magazine is intended to help broaden public understanding of intellectual property and of WIPO’s work, and is not an official document of WIPO. The designations employed and the presentation of material throughout this publication do not imply the expression of any opinion whatsoever on the part of WIPO concerning the legal status of any country, territory or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This publication is not intended to reflect the views of the Member States or the WIPO Secretariat. The mention of specific companies or products of manufacturers does not imply that they are endorsed or recommended by WIPO in preference to others of a similar nature that are not mentioned.