Smart Solutions for Safety and Health

What is Smart Safety?

During the revolution of Industry 4.0, current technological advancements have been leading to better understand the working conditions in workplaces and help optimize productivity and efficiency, while enhancing the quality of life of millions of workers. We make workplaces smarter and safer by utilizing sensor fusion and IoT (Internet of Things) concepts together with the recent machine learning and data mining methods.

We develop automated tools and systems equipped with sensors and actuators to prevent accidents, injuries, and diseases.

Real-time monitoring of the workplace with different sensors and cameras helps us understand the workplace conditions. We remotely detect and identify humans with their basic health conditions, detect other objects, and machinery with their physical properties. Utilizing machine learning, we can remotely detect team behavior (normal working, emergency-running people, and no motion) and harmful motions. Yet, we can even predict incidents before they occur.

Basic Statistics

 
  • According to the data analyzed by Eurostat, in the EU-27 during 2018:
  • Non-fatal accidents: ~3.1 million
  • Fatal accidents: 3332 deaths
  • At least for calendar days of absence from work
  • In Germany, the number of accidents has been reduced by more than half from 1992 (~1.9 million) to 2019 (~872000).
  • The number of fatal accidents dropped from 1443 to 497 during this period of time.
  • For the first six months of 2020, the number of accidents: ~367000.

Custom Solutions

  • We describe the problem and the situation together in a kick-off workshop and decide how to start.
  • Our researchers inspect on-site, analyze existing data, and make preliminary evaluation by collecting various data:
    • Displacement, velocity, and vital signs (e.g. breath and heart rates, body motions) by radar technologies
    • RGB, thermal, multispectral, and depth images
    • Temperature, humidity, sound, gas level, acceleration, and angular velocity of objects
  • After evaluating the preliminary inspection in an intermittent workshop we focus on one of the two actions or both:
    • Behavioral safety: to prevent accidents and injuries to influence employee actions
    • Predictive analysis: to predict present and future using historical big data
  • Implementation: We implement state-of-the-art numerical analysis, sensor fusion, and machine learning methods or develop new ones if needed. 

FeuerWERR

In the fireplace, thermal imaging cameras are used to find people who need help. These cameras can penetrate thick smoke but can only map a two-dimensional temperature. Therefore, the firefighters have no knowledge of three-dimensional information and they can hardly estimate local conditions.

Fusion of sensory data from radar and inertial measurement unit sensors provides distance, position, and orientation information which is later superposed on the infrared image. The implemented sensor fusion filter for navigation run with a position accuracy of 5 cm sufficiently accurate trajectories so that a combination of 2D radar and 3D infrared data is possible.

FeuerWERR project enabled a novel, portable, firefighter-friendly add-on to thermal imaging cameras by which the depth information is linked to heat images. This novel device helps the firefighters to minimize the cognitive burden of the firefighters.

Access Checker

The infections in the respiratory system cause high fever and change the breathing rate and pattern. In Covid-19, infection causes dry coughing and difficulty in breathing. Remote measurement of these vital signals is crucial for protecting the healthcare workers in the hospitals and security officers in airports, public offices, etc. There are many commercially available products to measure fever remotely with thermal cameras and non-contact infrared thermometers. The breath rate is an underestimated vital signal which can be measured manually or by chest straps. The breath rate can be measured remotely by detecting and analyzing the chest micro-motions caused by breathing. We combined a thermal camera and radar sensor within a single measurement box to simultaneously measure the fever and breath rate.

Click here for the press release.

Downloads

Real-Time Capable μ-Doppler Decomposition of Walking Human Limbs

Micro-Doppler Based Human-Robot Classification Using Ensemble and Deep Learning Approaches

Person Identification and Body Mass Index: A Deep Learning-Based Study on Micro-Dopplers