Alberto Broggi and his team have spent 15 years developing a true supercar: a vehicle that runs autonomously -- in any conditions. Is the future of driving really driverless?
Right now, a tiny Piaggio Porter minibus is making its way along the old Silk Road from Italy to China, a distance of around 13,000km. Made in a factory near Pisa, the Porter is at the more proletarian end of Italy's automotive spectrum: compact, cheap and cheerful. But what it lacks in style, this particular Porter makes up in artificial intelligence. Thanks to a network of cameras and scanners mounted on the vehicle, plus advanced GPS technology and complex software that can acquire, process and visualise all the input data in real time, this little bus isn't just circumnavigating half the planet -- it's doing so autonomously.
Having left its Parma base in July, the Porter was due (at the time wired went to press) to arrive in Shanghai in late October in time for the final weeks of the 2010 World Expo. The Expo's theme is "better cities, better life". The team behind this marathon "driverless" journey -- the VisLab Intercontinental Autonomous Challenge, or VIAC -- believes that autonomous vehicles could play a big part in that better future. "We've been working up to this for 15 years," says the project's director, Alberto Broggi, 43. "The main reason for embarking on such a gruelling trip is that we want to test the systems we've been developing all this time in the real world, in real traffic, in bad weather." Broggi has investigated applying artificial intelligence to four wheels since studying electronic engineering at the University of Parma in the late 80s.
Today he lectures there on computer science and AI.
He is also the director of VisLab, a company based at the university, which develops driving aids and perception systems, including blind-spot monitoring, collision mitigation, adaptive cruise control, pedestrian protection and traffic-sign recognition.
This is technology that has recently begun appearing on various mainstream cars, including Audis, BMWs and Volvos. But Broggi and his team are leading the way in applying it so that it works synchronously to deliver what he regards as the end game for mobility: the autonomous vehicle. "Autonomous vehicles will be everywhere in the future," he says. "But we need to think of it in a series of steps. Firstly, vehicles in a very structure situation -- such as an autonomous tractor moving around a field. "The next step would see cars moving autonomously in a controlled lane, in the same way that you see car-pool lanes on motorways. Finally, we would have vehicles in a busy urban environment. But then imagine autonomous taxis, delivery vehicles, buses..."
On the VIAC cross-Asia trip, the team's thinking is being put to the test. The focus of the project is on two vehicles, each of which has the same suite of sensors and identical control software.
The first vehicle carries out reconnaissance duties, conducts experimental sensing tests, makes decisions about the route and mapping, and collects data. The aim is for it to remain autonomous for as much of the time as possible. But, as it's the lead vehicle, human interventions are likely at critical junctures -- so it's manned at all times.
The second vehicle moves 100 per cent autonomously. It follows the Porter in front at a safe distance, using local sensing to alter its position, maintain a sensible speed and avoid obstacles.
If the lead vehicle disappears at any point, the one behind uses GPS co-ordinates to map out a route. The whole process is managed by Gold (Generic Obstacles and Lane Detection), a program developed in Parma. This is a stereo-vision-based hardware-and software system that detects road signs and traffic lights, as well as determining lane position according to painted lines on the road. "This was where we invested our greatest efforts, until the foundation of VisLab," Broggi says. The VIAC project is the culmination of two decades' work. Broggi and his colleagues' ambition was initially restricted by the limitations of contemporary computers: they simply didn't have sufficient processing power to provide a car with anything like a viable brain. In 1994, the group collaborated with researchers at Turin Polytechnic to create "Paprica" (PArallel PRocessor for Image Checking and Analysis), a batch processor with some 256 processors. As part of the European Commission- funded Eureka Prometheus project they demonstrated real-time lane detection using a converted Fiat Ducato van. "The vehicle was capable of gathering information from the outside environment, and relaying it to the driver," says Broggi.
Argo followed in 1998. This was a Lancia Thema saloon, which had been modified to travel 2,000km around Italy without human intervention. Argo drove itself for 94 per cent of the journey, using two small black-and-white cameras, and a modest Pentium processor for image acquisition, trajectory planning, driving actuation and data recording. "It was based on a system of cameras which were capable of recognising the lines of the road, but in some areas there were lines left behind by old building sites, which confused the software." The project's increasing profile led to the team being awarded funds by the Italian National Research Council. In 2001, it attracted the attention of giants such as Volkswagen, with whom VisLab collaborated on pedestrian-detection, thermal imagery and night-vision software, and the US Army. The latter commissioned the development of a system that would recognise pedestrians wandering unexpectedly in front of moving military vehicles.
In 2007 the team began work on what would become its masterpiece. "The idea was to build into a single vehicle the best technology developed through over 15 years of research," says Broggi. VisLab bought a Hyundai Sonata and began to integrate sensors and software.
The following year, the European Commission awarded Broggi's department a grant of €1.8 million, which it used to expand VisLab, increasing its team from six to 20 researchers. Still working with the Hyundai Sonata, the team rebranded the project BRAiVE (or brain-drive) soon after, and the following year the technology was presented in Xi'an, China, at the IEEE Intelligent Vehicles Symposium.
The BRAiVE Sonata features ten cameras, four laser scanners, 16 laser beams, GPS and an inertial measurement unit (which measures velocity, orientation and gravitational forces), as well as the full complement of driver-assistance devices. "The vehicle's strong point is its capacity to reconstruct, using sensors, an almost 360° vision of the surrounding environment, and base its calculations on that vision." That's why BRAiVE is one step ahead when it comes to overtaking an obstacle, crossing a junction or entering a roundabout safely. "We wanted it all, but it's clear that a complete solution, such as BRAiVE, won't be seen on our roads within the next 15years.That'swhywewanttopushthe Porters at the Shanghai expo," Broggi says.
After BRAiVE, the VisLab researchers were looking for the opportunity to show off their work more widely. "We have two different souls," Broggi says. "One is academic research. The other is about finding customers and selling this technology, because we need to eat. And funding research in Italy is really very difficult."
Around 95 per cent of the laboratory's revenue now comes from foreign agencies (such as the Pentagon and construction vehicle titan Caterpillar), and in Italy the team is working with electronics giant Marelli. The Shanghai Expo in 2010 isn't just a smart venue to demonstrate the technology to the biggest gathering of global thinkers for years; it also prompted the idea for a challenging journey. Broggi is delighted with the team's progress, and says there have been only minor glitches. "The most important thing is to be able to grow, to learn lots of things from this trip. If we don't have failures, it's not real," says the professor. "If we have failures, we'll identify them, fix them and by the end of the trip we'll have more knowledge." With an average vehicle speed of around 50kph, the team passed 3,000km on August 31 as they approached Kazan in Russia.
But what about the driverless future that so excites Broggi and his team? On this point, two questions remain unanswered. One concerns the increasingly rare pleasure many of us still derive from actually driving. Surely this is something that an Italian, of all nationalities, can still appreciate? Broggi laughs. "As far as taking the driver out of the car is concerned, well, there are cars and cars. If you're talking about a Ferrari, say, then no one would step away from it. But if we're talking about the car that takes you to and from work every day, for this kind of application the driverless car would be fantastic."
And what of the altogether thornier issue of how personal mobility is intertwined with issues of personal freedom? "In terms of your future vehicle, you could be willing to drive it, or there may be times when you are willing to let it drive you. It's your decision," says Broggi. "You might want to drive through the landscape of Tuscany, say. But when it comes to going to work and dealing with the traffic... well, then you switch it on. So no, I don't think it will limit your freedom. It's a tool. Use it or don't use it."
This article was originally published by WIRED UK
