Patient Safety System for Psychiatric Hospitals
The project partners Intenta GmbH, Eckstein Design and Technical University Chemnitz (Faculty for Informatics) develop - together with application partner Klinikum Chemnitz (Klinik für Psychiatrie, Verhaltensmedizin und Psychosomatik) - an assistance system that provides higher safety for patients and staff members in psychiatric hospitals through the analysis of complex behavioural patterns.
Benefits of the system:
- Independent sensor system combining image data and acoustic components
- 24/7 surveillance while protecting intimate personal data
- Protection from self-endangerment and/or being attacked
- Discreet application from emergency psychiatry to high risk areas
The system uses a multi-sensor-network (acoustic and image data) for the baviour analysis.
Using multi sensor data fusion the image and acoustic data is processed and evaluated without the transmission of video material or sounds (to protect personal rights) on a newly developed sensor platform in real time.
The system recognizes automatically dangerous situations and sends an alarm signal to the nursing staff (via mobile app or stationary).
As part of this project a protoypical system was successfully tested in real environment thanks to the Klinikum Chemnitz. The next steps are further developments to make the components ready for the market.
Data processing and software development
- Development of multi-sensor platform
- Localisation of person on room and pose detection
- Extraction of critical movement patterns
- Extraction of acoustic data
- Scene interpretation with multi sensor data fusion
Klinikum Chemnitz compiled and classified data concerning critical patient behaviour for this project. TU Chemnitz supported the development of human models for the detection of critical movement patterns.
- Faster reaction through early detection of critical situations
- Encreased safety for patients and staff
- Integrated data processing (embedded system) enables high level of data security
- Descreet sensor design for unintrusive integration from Eckstein Design
Intenta develops in cooperation with the Fraunhofer IWU Institut (Chemnitz) a new technology for safe human robot interaction. Additional sensors and expanded decision and regular routines enables the robotic system to interact intuitively with human workers within its reach. Collisions between human and robot are therefore eliminated without restrictions in productivity.
Through a new hybrid control strategy the robot can react to the specific anthropometry of the worker (e.g. body height) while successfully finishing the assembly job.
To analyse the woker's behaviour in the whole work area a network of Intenta S2000 3D vision sensors is used. An additional RFID-Sensor enables the identification of the worker combining the RFID information with the sensor network.
Last but not least, the robot is equipped with force moment sensors to measure the direkt interaction with the worker.
Data processing and software development:
- Survaillance of work space with Intenta S2000 3D vision sensors for specific scene interpretation (e.g. presence, position, ID, movement pattern of worker)
- RFID worker identification
- Plausibility check of robot position
The background of this project is the growing demand of automation of complex work processes in the assembly hall of automobile manufacturers. Our development provides:
- A possibility to encrease automation of complex processes through intuitive cooperation between humans an robotivc systems
- Decrease of physical strain for workers resulting in the preservation of work force and productivity
For these reasons Fraunhofer develops a security concept which guarantees a safe human-robot cooperation based on the processing of various sensor information.
Flock Early Warning System for the Airport in Frankfurt
Accurate prediction of potential collisions between aircraft and flocks of birds
Analysis of the hazard potential of a flock of birds by identifying
- Flight direction
The information is then forwarded to air traffic controllers
The software features:
- Stereo image processing
- Dirt detection
- Automatic recalibration of the cameras
- Software for monitoring and management of the system components
- Networking/communication across multiple sites
- Software for remote access
Zeiss® independently tested the quality and functionality of the software.
The system consists of the following components:
- 13 networked computers
- Multiple cameras, including 12 thermal imaging cameras from Cassidian and 6 light cameras
- Air chambers with compressed air
- Wash water
- Optical bench
- Control electronics
- Independent power supply (IPS)
Since commissioning in 2011
- We ensure the highest possible level of availability for technical systems support of the airport operator in Frankfurt AG
- We are the first point of contact and responsible for the proper functioning of the entire system (24/7)
Software for Crack Detection in Solar Cells
Developing a software for crack detection in solar cells
Our software includes:
- A module for the extraction of attributes
- An optimized procedure for the classification of objects
The result is an ability to distiguish between damaged and undamaged solar cells reliably.
- Elimination of the very complex process of manual inspection of solar cells
- An overall time savings of approximately 60%
We will adapt the software in conjunction with other updates:
- New image resolutions and formats
- New types of solar cells
- New error types can be added through classifier adaptation/optimization