IDEAL - Intelligent drone-based operational support through the automation of situational awareness

Our technologies for authorities and organizations with security tasks (BOS)

Inital situation

Situation reconnaissance and risk assessment is the first measure in any operation by the fire department, police and rescue services. Drones are ideal for larger and complex situations. They are now an established operational tool for many authorities and organizations with security tasks (BOS). In addition to situation reconnaissance and updating, drones are also suitable for various other deployment scenarios: Monitoring critical infrastructure (e.g. dykes), detecting changes in the spread of fires or the spread of liquids on bodies of water, searching for pockets of embers, searching for people, searching for animals and documenting operations.

Problem definition

Images and videos from drones are examined by image analysts, then the information is processed and passed on to the incident commander. Both the image analysts and the incident commanders have to process a lot of information in a very short time, which creates a bottleneck during operations. Unfortunately, the potential of automated, process-supported live evaluation of the video stream and the associated metadata (such as camera position and orientation) is currently hardly utilized, so that the use of the drone is often reduced to simply viewing the video image, i.e. the drone is used as a camera hanging in the air. The fact that the information often has to be passed on to the incident command by radio adds to the challenge of accurately describing the situation.

Project team

At Fraunhofer IOSB and Fraunhofer FKIE, we have been researching various methods for automated sensor data evaluation, information processing and presentation for many years. Our expertise covers methods for image and video evaluation and sensor data fusion, position display and guidance systems as well as HMI and cloud-based data processing. The cooperation of several research groups from both institutes as part of the DRUM project created a synergy that enabled us to develop a solution to the above-mentioned problems.

Our solution

Localization of detected and classified objects in a location map.
© Fraunhofer IOSB
Localization of detected and classified objects in a location map. Classification: BOS (light and dark blue), civil (green) and lying persons (red).

The solution we have developed is based on the evaluation of image and audio data of drones in combination with the evaluation of sensor metadata (telemetry data, such as flight altitude and sensor orientation). In addition to the classic methods of signal processing (such as image enhancement, filtering out the drone's engine noise, etc.), methods from the field of artificial intelligence (AI) are used, such as the detection, classification and tracking of various objects such as people and vehicles. By evaluating the telemetry data, a real geocoordinate can be assigned to each image pixel and each audio source. This means that the objects and events detected in the sensor data can be geographically located and displayed live on a location map.

Project implementation and outlook

As part of the project, a series of BOS workshops were held to determine the current needs of the BOS. In particular, we identified two important scenarios that we focused on in the project. The first was locating missing persons (detecting people in video images and locating clapping and shouting noises with the help of a microphone array). The second scenario was a major emergency with a mass casualty incident (MANV). The focus here was on the visual differentiation between the emergency services and vehicles and other people and vehicles, as well as image-based pre-triage, i.e. differentiating between the condition of people based on their pose (lying, sitting or standing). Other focal points were live access to and correct interpretation of the telemetry data as well as the design of a user-friendly software interface.

In the preliminary phase and during the course of the project, cooperation was established with various BOS, so that we were able to train and test our procedures using data from the exercises and real BOS operations. In addition, further problems became apparent that we are tackling in the follow-up projects, e.g. the detection of dangerous situations at mass events in the DRUM project.

 

DRUM – Drone-based support for the BOS at mass events

  • Live estimation of the number and density of people in a drone image
  • Detection and geographical localization of dangerous densities and acoustic events (gunshot and screaming noises)
  • Representation in the site map
 

ABUL – Automated image analysis for unmanned aerial vehicles

  • Integration platform for the developed processes
  • Live video evaluation and offline post-evaluation
  • Video database with search functions based on video metadata
 

SIRIOS - Fraunhofer Center for the Security of Socio-Technical Systems

  • Digitization of security and protection of critical infrastructures
  • Situation awareness, communication and operational command
  • Virtual planning and support for major events
 

Other projects of the VID department

Would you like to find out more about projects and products in the field of Video evaluation systems?  

 

 

Interested in a cooperation?

Are you interested in cooperating with our department or do you have questions about our current research topics? Then get in touch with us!