Predicting the future for video camera systems

Jo Versavel, Managing Director of Traficon, talks about near-term trends in video camera systems

Jo Versavel starts by making one thing clear: long-term forecasts as to what the future holds for video-based traffic monitoring are to all intents and purposes meaningless. The state of the art is developing so fast that in reality it's impossible to say where we'll be in 10 years' time, says the Managing Director of 5574 Traficon. In his opinion making firm predictions even five years out is too ambitious, whereas three is just about manageable.

It's a bold and refreshingly honest opening gambit. It's also realistic; although a man with an acknowledged vested interest and an unashamed advocate of the greater use of video-based systems, Versavel stresses a need for prudence when it comes to deployment.

Technology versus training

"Too many jurisdictions look to the increasing use of technology as a panacea," he says. "The reality is that while technology can significantly improve the performance of traffic operators, we should start by asking what they really need to be able to do their jobs properly. That training needs to be improved too is often overlooked.

"That's an unpopular judgement, not least because training is time-consuming and often complex, and people are a major ongoing cost which weighs heavily on the balance sheet." The observation is an especially important one, he notes, given the centralisation of control which digitisation and always-improving communications capabilities have brought about.

"There is a trend towards individual operatives being given greater and greater numbers of sites to monitor and control. I doubt that we'll ever see a situation where the traffic in New York City is controlled from a remote operations centre in India or the Far East, for instance, but the communications possibilities are increasing all the time.

"It's a question of balance. Local knowledge is crucial to effective network operations, particularly when it comes to incident management. I can think of an instance of a recent tunnel training exercise where two-thirds of the local traffic operators had never even been into the area they were supposed to be controlling. How can we reasonably expect them to be able to effectively manage the site if they haven't got an intimate knowledge of the lie of the land?"

Decreasing vulnerability

On the policy front, Versavel identifies a greater recent emphasis on the needs and protection of vulnerable road users such as pedestrians and cyclists. In a multimodal context, he adds, this is entirely predictable but it highlights the efficacy of video for wide-area observation, and the ability that confers on operators to make immediate appraisals of a scene and decisions on action.

"The main advantage of video is the amount of information it provides," Versavel says. "Incident detection and management is much improved. The next step is to look at how we can fuse different technologies and improve things still further."

Technological developments

The integration of video and thermal cameras is a specific area which Traficon has been looking into: "The advantage of thermal cameras is their lack of a need for additional illumination. Obviously, where there is existing lighting, you'd look to use it but even if it's absent there is still the ability with thermal cameras to send back data to network operators. Experiments in Belgium, the US and other places have already yielded decent results. The big issue at the moment is price acceptability. But I think that we can expect to see deployments happening within two to three years from now." An advantage of combining different camera technologies is that siting requirements tend to be similar: the necessary lines of sight for video and thermal cameras, for instance, tend to be the same. Versavel sees less merit in trying to more closely combine video with other types of detection and monitoring technologies such as radar and laser.

"Radar is great for exact speed measurement and monitoring global traffic conditions but when using it for measurement purposes one would normally position it differently than video. Thus far we've not been looking at combining anything like this at Traficon, although whether we do in the future of course depends on market needs."

Relative values

The fall in relative cost has driven the rise of video. In applications such as hard shoulder running, which is seeing increasing implementation as a means of gaining increased capacity out of existing road infrastructure, Versavel feels that video-based automatic incident detection is the only realistic option.

There is, though, a continuing role for the induction loop.

"It's less about competition and more about application," he continues. "The same can be said of the increasing interest in floating-vehicle or -person data-gathering systems, which are probably the future when it comes to travel time calculation.

"We can and should look at how we combine all these sources of data. We should be looking for those 'extra' sources but we can't solve problems with statistics - we need accurate information and the rapid ability to interpret, and that I see is video's role.

"There is a general tendency towards greater video coverage and if that results in a 'Big Brother' perception then I think we just have to find a way of living with it. Certainly, when it comes to safety and security I think there is a positive case to be made.

"However, the political issues, such as the length of time that data is held, need to be revisited on an ongoing basis because of how quickly the technology is evolving."

The future

Of more direct importance to traffic operations is the filtering of information and the avoidance of operator overload, he feels. It makes the refinement of a video system's operating algorithms all the more crucial.

"That should be another permanent and ongoing objective. Moore's Law applies as much to cameras as it does to chip development. We're seeing a doubling in camera capability every 18 months or so. The same can be said of communications and interpretive capabilities.

"Go back 10 years, and the state of the art in terms of interpretation was real-time analysis of just a few hundred pixels. Now we look to interpret the whole image - and that's against a background of ever-improving cameras with much higher resolutions, better night sensitivity, less blooming, no more smear and colour capabilities. That's why I think predicting 10 years ahead is just meaningless.

"Even three years from now, I think that high-resolution colour will be the norm. Image compression standards are also developing continually, to the extent that even keeping up with the names of the new standards is a challenge in itself.

"I think, though, that within three years we'll see the ability to distribute images to a great deal more people. I'm thinking beyond 'just' traffic operators and talking about the supply of real-time video to road users that'll allow them to see what's around the corner ahead. Technically, it'll certainly be possible. Again, though, the question remains of who will pay for the implementation." Delivery of real-time video to mobile devices will follow the trend of delivery to vehicles, Versavel says.

"That also ties in with both the general trend towards improving the safety of infrastructure and the increasing emphasis on the protection of vulnerable road users. We've barely seen the beginning in terms of what's to come in this respect. Once you've identified that the road user is there, you can then keep them better informed. From there, you can look to increase the levels of interaction with road users of all types."

Hard sell

Traficon's Automatic Incident Detection (AID) system has been applied for hard shoulder running applications since 2006. A Traficon-supplied solution has been in place in Belfast, Northern Ireland since 2006. At the end of 2007, hard shoulder running monitored by AID went live in Grenoble, France, and this was followed by the latest implementation, on the A73 in Germany, in February this year.

Hard shoulder running has two principal applications: to provide extra capacity at times of need; and to provide a priority lane(s) for public transport.@@@Although this new market has increasing potential as the number of such projects increases, the focus for Traficon remains on tunnel safety - not least because it is mandated that tunnels over 1km in length should have an incident detection system fitted, says the company's Stijn Vanderbuerie: "The biggest barrier to wider adoption for hard shoulder running is user resistance. Systems manufacturers can't claim 100 per cent detection of an obstruction or an incident. At Traficon, for example, we state that we can detect in excess of 95 per cent of stopped vehicles." That leads to a certain amount of timidity on the part of potential customers but, notes Vanderbuerie, should be viewed in a different light.

"The decision to open lanes for hard shoulder running should always rest with the traffic operator, not an automated system. It underlines more than anything else that technology should only be there to aid, and not replace, well-trained operatives."