Hannover Messe 2019

AI technology detects physical strain at the workplace

Hitachi, Ltd. and DFKI have jointly developed an AI-based technology to measure the physical strain of workers with a sensor suit. The system detects and quantifies physical stress in real time and informs users of incorrect posture and unergonomic movement patterns. DFKI and Hitachi will present the technology at the Hannover Messe in Hall 2, Stand C59.

In a society in which skilled labor is in shortage and in which recruiting new workers is becoming increasingly difficult, the preservation of the workforce is an urgent concern. Issues such as preventive risk avoidance and health protection, especially for workers in manufacturing and maintenance, are moving more and more into the spotlight.

An analysis of the extent of their physical strain helps to increase the safety of employees and protect their health. Traditional approaches use stationary cameras to detect and record worker activity. However, this approach does not ensure a quantitative and stable assessment of physical stress: Cameras are limited in their measuring range, and when taking pictures in production facilities or outdoor facilities, there are often blind spots in which the worker's movement profiles are not recorded or only partially recorded.

DFKI and Hitachi are developing a new AI-based technology that records a worker's movement data directly on the body using wearables, recognizes the physical load and evaluates it. The combination of Hitachi's activity detection technology and DFKI's AI technology is a unique feature:

Direct measurement of body movements during various activities and quantification of the physical load
Sensors in wearables measure the movements of more than 30 body parts directly on the body. The AI has already learned status recognition models for every part of the human body in advance. Now it analyzes the collected data and recognizes the characteristics of certain action steps by combining the body part positions. This allows a quantification of the physical load using a time series analysis based on Deep Learning.

Real-time estimation of physical stress and representation of differences between correct and incorrect working posture
By automatically comparing the movement data of an actual worker with that of a model worker, specific differences in the movements are identified. The operator is then shown an evaluation of his movement sequences, which indicates body parts with increased load.

Hitachi and DFKI will use the newly developed AI solution to support production employees in operations and training. In addition, the partners want to create a more efficient and safe working environment by improving productivity through the use of already developed AI for activity detection.

Other applications of AI technology developed for the manufacturing and logistics industries include sports, fitness, and games.

Contact

Prof. Dr. Andreas Dengel
Head of DFKI Research Department Smart Data & Knowledge Services

Phone: +49 631 20575 1010

Dr. Sheraz Ahmed
DFKI Research Department Smart Data & Knowledge Services

Phone: +49 631 20575 4818

Press Contact Hitachi
Mizuki Matsumara
Hitachi, Ltd.

German Research Center for Artificial Intelligence
Deutsches Forschungszentrum für Künstliche Intelligenz