Working towards 'intelligent transport'

DSG is exploring technology that may lead to an easier commute, writesKarlin Lillington

The phrase "intelligent transportation system" will likely strike most people as either an oxymoron or an impossible dream.

But that's what a team of researchers are working towards at Trinity College Dublin, where the Distributed Systems Group (DSG) - the largest and longest-standing research group in the Department of Computer Science - is exploring technology that may lead to an easier commute.

The concepts may seem outlandish:

• "sentient vehicles" that would know where other cars are positioned on the road and would know how to move evasively;
• "smart roads" that could relay information about surface conditions and traffic;
• wireless-enabled buses and cars that could beam traffic information to drivers or commuters waiting for public transport.

But initial work is being done on all these areas and a wireless network, positioned in a corridor running from TCD to Media Lab Europe in the Liberties, is about to be put in place.

"We're interested in time and safety-critical applicatons," says Mr Raymond Cunningham, a PhD student and researcher with DSG.

DSG has already set up a system that takes traffic information via Dublin's existing traffic management system - which allows traffic light switch times to be adjusted according to traffic levels measured by magnetic sensors embedded in the street - and display the information in real time on a Web-based system.

"This allows them to visualise traffic volume as coloured sections of road," says Dr Vinny Cahill, head of DSG. Different colours indicate levels of congestion.

It's information that he says will eventually be available to commuters over the Web, and ultimately could enable journey time estimates based on geographical positioning systems (GPS) sensors in each vehicle (a feature that is increasingly common in new cars). That information could also be made available to mobile phones.

Take cars equipped with numerous sensors. Such a "sentient" car would need to know how to process and then understand incoming data about outside traffic, road conditions, unexpected objects on the road, driving conditions and more.

All the data from all the sentient cars might in turn be shared among the cars or sent on to a traffic control centre. Data might also be coming from traffic lights and intelligent roads.

Computer scientists call the methods of pulling together such information into a coherent whole, "fusion techniques". According to Mr Cunningham, there are numerous well-tried theoretical methods for fusing all this sensory information, and seeking the best technique for their applications is part of overall research.

Another PhD student, Mr Vinny Reynolds, asks: "If we had the hardware in place for sentient vehicles, what sort of programs would we need to have in place?"

He is working to create computerised models of how cars might take in and react to sensory data, using the concept of a "safety zone" that encircles the car. His research goal is to get the cars to behave in a co-ordinated manner on the road, but he also is exploring the question: "if vehicles could communicate, would it change the nature of how they travel?".

Sentient vehicles, he says, would have autonomy, could sense their environment, could co-operate and co-ordinate, share information and behave intelligently. It's a tall order, and in the animated model he has now, the little coloured blocks that represent cars occasionally run into and over each other.

Which raises the question of whether human drivers will be willing to cede autonomy to their car and let it decide whether it's time to apply the brakes, or not.

"It's not something you can just throw at people," he concedes. "More likely, the first systems will offer suggestions - 'this is what the vehicle thinks is the right thing to do'. It's going to be incremental steps."

But Dr Cahill is convinced autonomous cars are on their way. "There's lots of different pressures that are going to drive this," he says, prime among them basic safety issues.

If a car as intelligent as you is somewhere off in the distant future, a wireless network running through the centre of town, testing out experimental applications, is not.

The city of Dublin is about to place an initial cluster of 12 mini-computers, less than five inches square, that will act as wireless "nodes", or relay points for a wireless-access high-speed internet network.

The little PCs fit inside a box about half the size of a shoebox and will be mounted on the traffic signals running up Dame Street to Thomas Street. Mr Stefan Weber, a PhD student working on the project, says the lights are a perfect place to locate the boxes because the PCs can run off the lights' 230 volt power supply.

The boxes run on Linux - easier to configure remotely than Windows, says Mr Weber. He is interested in how "ad hoc" wireless networks might work.

In fixed internet networks, only select computers act as data handling centres that can process and direct internet traffic, such as email messages or requests for Web pages.

In an ad hoc network, each person's wireless device becomes a point on the network that can handle traffic data. Add another person to the network, and the bandwidth of the network increases. But the problem for reliable traffic data management is that people on wireless networks tend to be people on the move, who jump on and off a network.

So, says Mr Weber, ad hoc networks are quite flexible and powerful, but also have limitations. The wireless corridor, a project in which TCD is partnering with Media Lab Europe, Dublin City Council, Eircom, Ericsson and Intel, will be a testbed for seeing how the technical limitations can be addressed, the strengths exploited, and what kinds of application might use such a network.

"Out on the street, there could be loads of devices with their own wireless communication built in. Cars, buses, PDAs [personal digital assistants\], traffic lights, laptops - all would bring their own bandwidth," says Mr Weber. That means, eventually, a person waiting for a bus on Dame Street could check a mobile or hand-held to see when the next 16 will be along. Or an ambulance could automatically switch traffic lights to hasten its journey to hospital.

Mr Weber is looking at the protocols that might be used to route data through the network, and how to get the best quality of service.

This year, the network will be research only, but the plan is to expand it to 30 or 40 PC "nodes" and open it to the immediate community next year, and eventually, the general public. And then? Intelligent transport? We'll have to wait and see.