Fusing data to manage traffic
Tom Shelley reports on one of the most challenging sensor fusion problems tackled to date – and the ongoing efforts to solve it
Solving what might seem a simple problem of detecting traffic congestion and guiding motorists away turns out to be a highly complex one. As well as the technical problems of gathering and processing the required data, there are privacy, legal and organisational issues that have to be solved.
Yet these issues are being resolved and the UK’s first pilot scheme may soon be active, followed by a national rollout. Jean Bacon, professor of distributed systems in Cambridge University’s computer laboratory, offered her insights into the Time (Transport Information Monitoring Environment) project at the university’s recent Horizon seminar on ‘Energy in Cities’.
The core of the project is Time-EACM (Event Architecture and Context Management). The collaborators are the universities of Cambridge and Oxford, with BT as an industrial partner.
“Cambridge is being used as a test bed,” states Bacon. “It is being mandated by central government to expand, but it is already congested and polluted. Policymakers – city and county councils – need real data as a basis of decision.”
The goal, she says, is to take information from existing systems such as traffic light and additional sensors, and compose the data sets into a meaningful pattern. Ideally, people should have a desktop device that indicates a bus has been detected on their route, so they can go outside and wait for it. Alternatively, it should be possible to obtain quick answers to such questions as: “What is the best way to get to Stansted Airport?” taking into account possible traffic delays or rail problems.
After initial setbacks, the local Stagecoach bus company, monitoring system developer ACIS and the county council have agreed to cooperate on the project, with their data protected by a non-disclosure agreement. Stagecoach uses GPS location of its buses and so knows fairly accurately where they are at any given time.
Rather than use CCTV cameras, the team has been trialling infrared cameras – devised by Irisys for detecting queues of people indoors, but adapted to count vehicles outdoors. Tests have been carried out to count vehicles coming along a nearby road, with a 90% success rate relative to video. A similar success rate was seen when counting vehicles travelling on one of the city’s major arterial roads, using a mast-mounted camera on a vehicle.
Information about queues in front of traffic lights is being obtained from 1980s induction loop technology embedded at junctions to count queue lengths using a system called Scoot – Split Cycle Offset Optimisation Technique – originally invented by the Transport Research Laboratory (TRL) and currently maintained by Siemens in Cambridge. These were positioned to turn lights green when queues grew too long or buses were waiting.
* System can obtain real-time traffic congestion information, without infringing the privacy of individual motorists
* Sensing information is being obtained from traffic signal induction loops, bus tracking information from bus companies and experimental infrared vehicle counting cameras
End for motorway misery?
A pilot scheme to ease traffic congestion at peak times on a busy section of the M42 motorway could be rolled out nationwide. The 12-month trial was aimed at reducing congestion by part-time use of the hard shoulder and variable speed limits. It was also intended to increase safety by using video data, and offer enhanced driver information about traffic flow, journey times and accidents.
Journey times were 27% shorter, fuel consumption fell 4% and emissions were down 10%. The trial also saw the personal injury accident rate fall from 5.2 per month to 1.5 per month.
The traffic management system was implemented by professional services group Mouchel Parkman, which identified the need for an Internet Protocol (IP) over Ethernet based fibre network to connect the cameras to a manned control room. As a result, the group chose GarrettCom's field-hardened 6K16 managed industrial Ethernet switches with its S-Ring redundancy software.
This material is protected by MA Business copyright
see Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the