The benefits and pitfalls of AI in traffic
Our transport network is increasingly reaching its limits. According to many, the solution isn't just more roads and trains, but artificial intelligence, aka AI. But how can computers and their algorithms protect us from traffic collapse?
It's just after five on a Friday afternoon. I'm sitting in my car trying to get out of town. Unfortunately, the traffic lamp I've been stuck at for what feels like twenty minutes only lets three cars through per green light. Why is that? Can't this be optimised? Do the drivers across from me suffer from the same problem?
I always ask myself these and similar questions when I'm on the road. Traffic light systems are constantly being optimised. But we're just starting to get AI involved.
What is artificial intelligence (AI)?
Artificial intelligence is a major field of computer science. There's still no universal definition for AI. American scientist Marvin Minsky has defined AI as «the science of making machines or systems do things that would require intelligence if done by men». The European Commission in turn defines AI as «systems that display intelligent behaviour by analysing their environment and taking actions – with some degree of autonomy – to achieve specific goals.»
AI isn't a technology, but refers to a variety of different approaches, methods and technologies. AI today is often divided into three categories: symbolic AI, data-driven AI, and future AI technologies. Symbolic AI involves systems in which humans develop a set of logical rules and convert them into algorithms so that machines can follow them. Machines can thus make decisions depending on the situation. Data-driven AI combines machine learning technologies with technologies used to search and analyse large amounts of data. Future AI technologies will include various developments in which AI can take on human traits such as intuition and creativity or even exceed human intelligence.
How can AI help optimise traffic?
Population growth and increased mobility are the main factors behind the large volume of traffic. According to federal calculations (in German), congestion on Swiss roads cost us CHF 1.9 billion Swiss francs in 2015. Traffic jams don't only cost nerves, but money as well.
AI can help make all modes of transport safer, more ecological, smarter and more convenient. It can be implemented in vehicles and infrastructure and thereby change how means of transport and their users interact. AI also helps evaluate trends, identify risks, alleviate traffic chaos, make transport more economical and manage it.
All the advantages of artificial intelligence come with challenges. These include ethical, social, economic and legal issues. Questions relevant to safety also arise.
How is AI used in road traffic today and what problems does this raise?
AI in traffic
There's a lot going on with AI in road traffic at the moment. Manufacturers are experimenting with self-driving vehicles for public and private transport. These vehicles are equipped with sensors and actuators as well as control units and software.
Some of these technologies perform only one or a few functions while driving, such as parking. Other technologies, on the other hand, should make the human driver completely superfluous. AI, which take over certain functions, are already common practice. Autonomous vehicles are currently being tested. Tesla has been testing completely self-driving vehicles for years. But such experiments can also suffer from setbacks. A self-driving bus (in German) project launched by the Schaffhausen public transport company was put on ice in June after the bus collided with an e-biker.
Uber (in German) is also testing self-driving vehicles. However, the company doesn't only rely on AI in that sector. Uber uses AI in all aspects of its services. From matching drivers to route optimisation.
But AI can also help in road transport. In truck platooning, for example. Several trucks drive one behind the other at a minimum distance. Only the front vehicle contains a human driver. The other trucks follow the leader using AI. This already works very well on the motorway, but in complex traffic situations, such as traffic lights, these technologies still have their problems.
Traffic management also benefits from AI. It analyses traffic patterns, traffic volume and other factors. On the one hand, this can help in planning a road system and reduce congestion. Traffic light systems can also be improved with AI.
All these technologies, in addition to making road traffic more comfortable, would also protect the environment. With liquid traffic, less congestion and optimized transportation, vehicles pump less exhaust fumes into the air.
Risks
Despite its many advantages, AI also carries risks in road traffic. Self-propelled cars give drivers free time; they no longer have to worry about driving. This circumstance could tempt many people to switch from public transport to self-propelled cars. This would destroy the progress made with ecology in transport.
Data protection also raises important questions. AI for self-propelled cars requires an enormous amount of data. They may be protected, but there is still the possibility that third parties could access the data.
Human error is the most common cause of traffic accidents. According to a study by the BFU (in German), 95 percent of traffic accidents in Switzerland are caused by human error. Without humans, the number of accidents could be drastically reduced. AI itself isn't yet immune to mistakes and accidents, as the bus in Schaffhausen and tests by Uber and Tesla show.
AI in transport also raises many ethical questions. How should AI react in an emergency? My colleague Philipp Rüegg will deal with this in another article for our theme week.
In the event of accidents, the question of liability arises. Who is responsible for an accident in any specific case? Liability must be defined according to the level of automation of vehicles. In the case of fully autonomous cars, is only the manufacturer liable?
AI in transport will change jobs, create new ones and also destroy jobs. Some professions such as taxi, bus or truck drivers will probably disappear in the future.
Header image: Image: Thomas Kunz
31 people like this article
Kevin Hofer
Senior Editor
kevin.hofer@digitecgalaxus.chFrom big data to big brother, Cyborgs to Sci-Fi. All aspects of technology and society fascinate me.
