There is an oft-repeated anecdote about autonomous vehicles: that if a self-driving car were crossing a bridge with a bend in it, around which a child was playing in the middle of the road, and if the car was driving too fast to react in time, its high level of intelligence would lead it to avert an accident by throwing itself into the abyss. Because it would understand that the life of the child was worth more than that of the adult in the car. Or, in another version of the story, if it had to choose between hitting a child or an elderly person, it would opt for the latter.
Proponents of this scenario argue that few would elect to buy a car capable of killing its own occupant. Or they deliberate over who would be held responsible in such an eventuality. Ultimately, this kind of legal, social and cultural debate is putting a spoke in the wheels of a technology that is already available and ready for use.
Little has been heard, however, about what Applus+ IDIADA’s Director for Innovation, José Manuel Barrios, has to say on the subject. “Technological efforts are underway to ensure that this scenario never comes to pass – that both the child and the elderly person remain unscathed. In the field of transport automation, the driver is the car itself – and it will be taught not to have accidents.” And, he goes on, if an accident does occur, “we will investigate the causes and work out fixes to put in place”.
Such differences of opinion are wholly indicative of the wealth of uncertainty and misinformation surrounding connected and autonomous vehicles. Applus+ IDIADA is a design,
engineering, testing and
homologation specialist for the international automotive industry and has just completed the research phase of several European Commission-backed initiatives, including GCDC i-Game and Companion.
The first of these established the future rules of engagement for the autonomous-vehicle industry. The GCDC (Grand Cooperative Driving Challenge) 2016 was an innovative demo event that took place on the A270 motorway between Helmond and Eindhoven (in the Netherlands), which saw ten European teams competing. The challenge combined vehicle automation (self-driving) with vehicle-to-vehicle and vehicle-to-infrastructure communication.
Companion, for its part, aims to define a coordinated system with which to create, maintain and dissolve – dynamically and autonomously – platoons of heavy-duty vehicles. And to do this via a decision-making mechanism that takes into account both infrastructure and the environment.
In the NEXTECH project, working alongside the Autonomous University of Barcelona (responsible for computer vision); Catalonia Polytechnic University (supercomputing); Miguel Hernández University (cooperation and communication); engineering company Pildo (positioning) and car manufacturer SEAT, we are testing an evolutionary autonomous vehicle that can take the wheel at the driver’s request. According to the Applus+ IDIADA director, “this is an exercise in researching, validating and testing new technologies since the vehicle is equipped with cameras, radar, artificial intelligence, etc.”.
In parallel, Horizon 2020’s C-MOBILE initiative, which kicked off in June, is looking into cooperative systems with a view to standardising communication between vehicles and urban infrastructure in eight European cities, including Barcelona, Bilbao and Vigo. “We are aiming for a permanent deployment of technology within the cities: technology to last,” observes Barrios. His company will be the project’s ambassador in the United States.
THE CURRENT CLIMATE
It is important to put the growth of such initiatives into context. “Transport in general is facing three big challenges: safety; congestion and traffic flow; and sustainability,” claims Barrios. The key difference between past and present is that today’s policy-makers have a clear solution at their disposal: automating and electrifying transport and promoting shared mobility.
“There are three developing trends: the connected vehicle, the cooperative vehicle and the automated or autonomous vehicle,” explains the Applus+ IDIADA director. “Practically all new models produced nowadays are connected, but this does not mean to say that we have achieved the pinnacle of connectivity.” Let’s compare this with mobile phones, a field in which few parallels exist between the early days and what we have today. “Car manufacturers are incorporating SIM cards into their vehicles, with their own proprietary platforms; mobiles connect with cars; cars connect with external devices; and integrated sensors alert us when, for example, tyre pressure drops.”
Cooperative systems look at the car’s communication with the outside world. Two different industrial strategies are being developed in this area. The first involves cooperation via a Wifi network. “The primary advantage here is that, being a local network, latency [response time] is kept to a minimum, which is ideal for safety-related applications,” says Barrios.
“The main drawback is that to connect with such a network, there must be a device known as an On Board Unit (OBU) in the car, which manufacturers could build in, as well as an external aerial, for which someone [ideally the municipal authorities] would have to pay.” The obstacle, therefore, is investment.
The second strategy makes use of mobile infrastructure – in other words, 2G, 3G, 4G and 5G networks. “The main benefit of this option is that, since the infrastructure already exists, there is no need for heavy investment,” Barrios suggests. “But there are other issues to consider, such as network coverage, network saturation and delays in communication.” In this regard, 5G shows promise, yet “expectations are high as to when and if these problems can really be solved”.
As far as the automated or autonomous car is concerned, it is helpful to look at the Society of Automotive Engineers (SAE)’s five levels of automation. “There will be a great leap forward any time now, with levels 3, 4 and 5,” explains Applus+ IDIADA’s Director for Innovation. These distinct stages of development are well known within the automotive industry as well as among telecommunications service providers and providers of artificial-intelligence and computer-vision solutions, who have been working on end-use applications for several years.
Level 3 involves having a button in the car enabling the ad-hoc transfer of driving responsibility over to the car. The problem is that the car can decide at any given moment, without warning, that it is no longer able to drive.
At level 4, Barrios explains, “the car is guaranteed to keep driving autonomously in a range of predetermined scenarios”. For instance, whenever it is on the motorway, no matter what happens, the vehicle will drive on its own. “And at level 5, the car will drive on its own in all circumstances.”
Traditional manufacturers see automation as an evolutionary process and, as such, are working through the levels step by step. But “there are new players in the sector that have come in aiming straight for level 5”.
What, then, is the main obstacle to making the autonomous vehicle a reality? “The legal framework exists, as does the technology. It is more a question of maturity in terms of developing, testing and validating the technology in all possible scenarios,” concludes Barrios, adding that “we have a shared responsibility in this, since user acceptance is also vital”.