Skip to main content
28 October 2021
Research & Development Group, Hitachi, Ltd.
One of the biggest challenges faced by society in developed countries is an aging society due to a declining birth rate. What this means is that there is or will be a serious labor shortage both in terms of overall numbers as well as those with expertise or experience. To overcome this shortage, older people or foreign workers are being increasingly employed to fill the gap. At such sites, accidents or injuries due to unfamiliar work may occur. Thus, it is important to understand how the physical workload is being handled in order to ensure a healthy and safe work environment. In this blog, I’d like to talk about the work that we are doing to quantitatively visualize strain and enable the most appropriate response based on the physical load.
In collaboration with Xenoma Inc., the German Research Center for Artificial Intelligence (DFKI) and its spin-off sci-track GmbH, Hitachi is pursuing the research and development of wearable AI technology that monitors workers ' physical load at all times in an effort to achieve solutions that improve worker safety and health in industrial fields. 
In this post, we’d like to introduce a brief overview of our technology.
To measure the physical load, we need to accurately understand the bodily movement of the worker. To do so however, the worker needs to wear sensory devices but these may restrict movement or result in discomfort. In our system, we used a smart apparel (e-skin), developed by Xenoma Inc. With their technology, more than 15 IMU (Inertial Measurement Unit) sensors can be embedded in a work clothes normally worn in the field. Since it can be worn comfortably without any stress, and it enabled us to capture the worker's movement at all times. Figure 1 shows our jacket-typed sensor.
Figure 1. Our jacket-typed sensor
One feature of clothing such as a jacket, however, is that it “deforms” as the body moves. This makes it difficult to accurately estimate the worker’s posture. The sensor may become misaligned with the body and body movement may not match sensor movement. To address this problem, Hitachi in collaboration with DFKI is developing AI technology which compensates errors caused by clothing misalignment and other factors, making it possible to correctly recognize the posture of the worker. The error correction and estimation of skeletal information are shown in Figures 2 and 3, respectively.
Figure 2. Errors caused by clothing deformation compensated by our AI technology
Figure 3. Estimated skeletal information from the wearable sensor
Based on the estimated worker’s posture, our system recognizes the workload on the worker’s body. The system visualizes workloads of various tasks by using RULA (Rapid Upper Limb Assessment), a standard ergonomic tool that assesses the magnitude of the workload according to the angle of each joint of the worker and the amount of force applied to the worker. (Figure 4.) Using these technologies, we can collect and visualize the worker’s physical workload in real time.
Figure 4. Assessing the magnitude of the physical load from the worker’s posture using our system
In this blog, I briefly outlined how we are using technology to recognize the posture of a worker using a jacket-typed sensor and estimate the workload according to the posture to implement improvements. (Figure 5.) Furthermore, by measuring the accumulated workload, it will be possible to prevent work-related injuries such as lower back pain. Our goal is to develop technology to support healthy and safe work environments in an active and inclusive society. To find our more, watch “Solutions to visualize workers' loads” (in English, in Japanese).
Figure 5. Example of solution with our technology