Artikkelen tilhører Artificial Intelligence (AI), postet 3. mai 2016
In 2004, the American Defense Advanced Research Projects (DARPA) held its first challenge to make an autonomous vehicle. None of the participants managed to solve the problem. Today, Google’s self-driving vehicles have logged about 2.5 million kilometres and cars with autopilot functions are already on the market.
In 2004, the American Defense Advanced Research Projects (DARPA) held its first challenge to make an autonomous vehicle. None of the participants managed to solve the problem and claim the $1 million prize. Today, Google’s self-driving vehicles have logged about 2.5 million kilometres, and cars with autopilot functions are already on the market.
Due to large technological advances during the past few years, it is now possible to integrate supercomputers and sophisticated sensors into a car. Radar, ultrasonic sensors and laser scanners visualize the environment, while artificial intelligence and digital maps enable the car to navigate in traffic alongside pedestrians and manned vehicles.
- make individual transport available to anyone
- are safer and more environmentally friendly
- create new opportunities within urban planning
- challenge the role of humans in the traffic
While distracted drivers cause most traffic accidents, driverless cars never take their attention off the road and react immediately in the event of something unexpected. This may save many lives. Connected and coordinated cars that communicate in a network may also lead to a much more efficient traffic system.
WHEN WILL THE TRAFFIC BECOME DRIVERLESS?
In October 2015, Tesla issued a software update to 60.000 car owners, enabling their cars to manoeuvre on the highway without the driver doing anything. However, the future scenario of a fully automated traffic system is still many years ahead. Google claims it will have a fully automated car on the market within four years, while the European Road Transport Research Advisory Council estimates we will have to wait at least ten years, considering both legal and technological challenges.
Increasing degrees of automation will become more common in the coming years, where the car handles parts of the driving and summons the driver when needed. The table at the bottom of the page shows the degrees of autonomous driving. The current version of the Tesla Autopilot is at level 2, as the driver still is required to monitor the traffic.
Autonomous driving requires good weather conditions, and snow is especially problematic for the sensors. The cars are not yet able to solve all traffic problems faultlessly, but machine learning helps speeding up the development. When one Tesla with Autopilot has driven a stretch of a road, all remaining models will share the knowledge and handle the same distance a little better the next time.
OPPORTUNITIES AND CHALLENGES
Improved safety or less attentive drivers?
Human error is the cause of 90% of all car accidents. Taking humans out of the equation may therefore make road transport significantly safer. However, the transitional period with varying degrees of autonomy may lead to problems, due to drivers not knowing when the machine is in control or not. Based on this, Google insists that their vision of the fully autonomous car is the only option that is safe enough. Can information and training ensure that drivers are aware of their responsibility?
Digital vulnerability on the highway
The online car is susceptible to network breakdowns and unwanted access from outside the car. For instance, hackers have managed to take control of a Jeep via the online entertainment system in the car. Tesla maintains that its system is not possible to hack, and that physical access is required in order to control the car.
Opportunities for children, elderly and the blind?
Driverless cars will make individual transport available to anyone, regardless of their ability to drive a car. Increased access to mobility coupled with digital coordination may result in economic growth and increased productivity. On the other hand, automation threatens the jobs of 62.000 professional drivers in Norway.
Car sharing changes urban development
The taxi company Lyft and General Motors cooperate on autonomous technology. The technology allows for networked systems for “robotaxis” that drive continuously rather than being parked most of the day. Such a system would reduce the need for parking spaces, freeing up these areas for parks, pedestrians or new buildings.
Platooning allows for less distance between cars, reducing the space that they require on the road. While a driverless system may be more efficient, increased convenience may also lead to more cars on the roads.
Should the freedom to drive be revoked?
The biggest advantages of driverless cars come when all the cars on the road are digitally coordinated. In this scenario, there is no room for human drivers. For many people, driving a car is part of a lifestyle and a symbol of individual freedom – something that will be lost in the driverless future. In a system with autonomous cars, both companies as well as authorities will be able to track more of an individual’s movement. Is the increase in efficiency and improved safety enough to outweigh the reduction in privacy and personal freedom?
Artificial intelligence – actual responsibility
The UN Vienna convention from 1968 states that a car shall have a human driver. The Norwegian Road Traffic Act (Vegtrafikkloven) states that the driver always shall be in full control of the vehicle. Should we recognize the computer as the driver? Who will then be legally responsible when an accident happens?
How to program an accident?
Autonomous driving brings about new ethical dilemmas. Manufacturers will have to make crucial decisions on how the car shall react when something unexpected occurs. The outcome of an accident that will harm either a pedestrian or the person in the car must be decided in advance.
WHAT HAPPENS IN OTHER COUNTRIES?
In 2015, the government in Great Britain wrote an action plan for the introduction of autonomous vehicles. They have examined the British law, not finding anything that prevents the use of autonomous vehicles, as long as there is a person present in the car. The UK have made guidelines for testing, and trials have already started in Bristol, Coventry and Milton Keynes.
In Sweden, Volvo is cooperating with Gothenburg municipality and other authorities on the ambitious project “Drive Me”. In 2017, 100 customers will test cars equipped with autopilot on commuter routes in Gothenburg. At the same time, Volvo will also test 100 cars in China, where traffic and road conditions are different from that of Sweden.
In the US, certain states have permitted self-driving cars for years. Nevada was first in 2012, with several other states following shortly after. In February 2016, the National Highway Transportation Safety Administration decided that Google’s computer qualifies as a driver. This allows for making cars without steering wheels and pedals, and without an option for the passenger to take control.
The Netherlands aim to take the lead in testing of self-driving technology, and have opened their roads for testing both personal cars and freight transport. In April 2016, all of EU’s transport ministers signed the Amsterdam Declaration, promising to cooperate on rules and regulations for the development of automated transport.
On the highway between Norway and Finland, a project involving testing platooning of driverless trucks has started. Platooning can reduce fuel consumption as well as the need for labour, and can especially be useful in the transport of salmon to Asia via Helsinki.
Kongsberg municipality has decided to become a pilot city for testing of self-driving vehicles. Testing of a self-driving bus will commence in September 2016, with long-term plans of putting it to regular use.
The preliminary version of the National Transport Plan 2018-2029 recommends following Sweden in making a Norwegian Official Report (NOU) on autonomous vehicles. The transport plan states that autonomous driving can give gains in efficiency, and mentions ground vehicles in airports as a point of departure for using self-driving cars.