Intelligent Surveillance System To Detect Aberrant Behavior By Drivers And Pedestrians

"We have developed an intelligence surveillance software and related theoretical model in order to define 'normality' in any setting one wishes to monitor, such as a traffic scenario", David Vallejo, from the ORETO Applied Intelligent Systems research group of the UCLM and co-author of a study published in the latest issue of Expert Systems with Applications, tells SINC.

The study focused on a pedestrian crossing in a two-way street, regulated by a traffic Light. The authors defined 'normal' behaviour of cars and pedestrians in this setting, in which they can move when the lights are green, but must stop and not cross the safety lines when the lights are red.

The system, working in a similar way to a human monitor, can detect whether the vehicles and pedestrians are moving "normally". If at any point any of the movements related to these "objects" is not 'normal' (driving through a red light, for example), the programme recognizes that the behaviour differs from the normal framework established.

The supporting architecture underlying the model is a multi-agent artificial intelligence system (made up of software agents that carry out the various tasks involved in monitoring the environment. It has been designed according to standards recommended by the FIPA (Foundation for Intelligent Physical Agents), an international committee working to promote the adoption and diffusion of this kind of technology.

In order to prove the effectiveness of the model, its creators have developed a monitoring tool (OCULUS), which analyses images taken from a real setting. In order to do this, the team members placed a video camera close to their place of work, the Higher School of Information Technology in Ciudad Real.

"In this way we are able to identify any drivers and pedestrians behaving abnormally, meaning the programme could be used in order to penalise such behaviours", says David Vallejo.

The researchers are continuing their work to fine tune the system, and believe it will be possible to use it in future in other situations, for example in analysing behaviour within indoor environments (museums, for example), or in detecting overcrowding.