Human insight is indispensable in the traffic control room. Operators spot abnormalities being shown on a video wall immediately. However, the number of image sources is increasing exponentially. It's impossible to keep an eye on all those images at once. In traffic management, we see the need arising for technology to manage image sources in an intelligent manner. A look into the future of video management in traffic control rooms.
An increasing amount of tasks require cameras. In order to monitor public space, law enforcement agencies, for instance, have access to thousands of cameras throughout the entire nation. The enormous amount of footage that these cameras generate, holds an invaluable amount of information. Cameras in areas with an active nightlife register disturbances. Along the highway, traffic is being filmed and in the larger cities, dozens of cameras help monitor demonstrations and events. It is a huge challenge to make sure that the right images are shown at the right time on the correct display. With so many image sources, it is impractical to go over all images manually. The risk of incidents going unnoticed by the employees in the control room rises.
Stream of images in traffic control rooms
The police is not the only one facing this stream of images. Traffic control centers are also facing an increasing amount of images that support the execution of their tasks. In some environments, like large traffic tunnels, there is so much footage that the majority of the images is not being checked by human eyes. When it is required, in case of an incident for example, the images are retrieved and reviewed in hindsight. The visual information is there, but it is only viewed after events took place. The strength of human operators – quickly recognizing abnormalities in the images – loses its effectiveness when image are no longer viewed as a live stream.
Many systems for video management exist. Often they are electronic solutions, like SigmaXG, but there are also systems that are based on software. For instance for centralized traffic management and the control of infrastructure. These applications combine the management of video signals with process managers, task control and sensors. Flexible controls over camera images and applications offer the user a firmer grip on the tasks at hand. Video walls with monitors will be replaced by large displays with high resolution that are easier on the eyes and show many sources at once. The controls are simpler and more versatile. Despite these options and technologies, control rooms are still at risk of being overwhelmed with visual data.
The challenge for video management systems is therefore primarily located in the selection of the correct images. The system needs to predetermine through relevant information which images should be shown in the traffic control room. In addition, there is a need for video managers with more autonomy that actively support operators during the execution of their tasks.
Sensors as triggers for traffic management
How does a video manager select the correct images from the available source? A relatively simple way is to use (existing) sensors. Based on the sensor data, the system determines which images are relevant for the operator in the control room. Video management through sensors and other triggers is applicable in many situations.
For example: in some tunnels it is crucial that traffic always maintains a minimal speed. When the speed of traffic decreases, this is detected by sensors in the surface of the road. An alarm will be generated in the control room to get the operator's attention. Upon confirmation of the alarm, the right camera images will automatically be shown at the video wall.
In another application we developed based on triggers, the control interfaces of multiple bridges are combined. When a request (trigger) comes in to open the bridge, the operator will immediately get to see the control elements and the camera images of that particular bridge.
Aiming for autonomy
The next step is taken when the video management system can determine by itself which images and applications require the attention of an operator. This requires more than triggers. The system will have to be capable of analyzing and filtering images autonomously through pattern recognition and deep learning. Video management is therefore developing from image registration and distribution to a more active support of the operator's control tasks.
Possible applications for intelligent, (semi-)autonomous video management are endless. The speed of traffic in tunnels can be determined through analysis of the camera images, which makes sensors obsolete. Observation systems at events monitor flows. When unexpected movement is detected, these images are directly shown in the control room.
With intelligent video management, tasks in traffic control rooms will change. When the system is reliable enough to independently select the relevant images, humans no longer are required to monitor all the images. The role of the traffic manager will shift from monitoring and analysis to incident management.
The intelligent video manager
The increase of the number of video sources and the demand for advanced types of video management will only continue in the foreseeable future. It is important to take the future into account while designing video management systems. A clear segmentation in software functionalities, hardware elements and infrastructure results in a modular and controllable architecture. Currently major progress is being made in the field of intelligent algorithms and controls. It is crucial that these parts are separated from the hardware so that the implementation of new algorithms can be realized quickly.
The intelligent video manager analyzes all sources and only redirects the relevant images to the operator. The system provides suggestions for the deployment of instruments, or actually deploys them. The operator becomes a director that coordinates the handling of incidents. When, for instance, a video management system detects a traffic accident on the highway, it can automatically warn emergency services through other systems and block traffic lanes. When emergency services give the all-clear for the road, the system will check whether this is correct and activate the proper scenarios for re-opening the traffic lanes. The traffic manager thus gets maximum support of his work flow and only needs to step in when human intervention is actually required. Does it end there? Probably not. Autonomous systems have way too much potential in the field of monitoring and security for that. Intelligent video management plays an essential role.