Digital methods in the field of ergonomics

Digital models and methods can be useful for the ergonomic design of products and work processes. They include digital human models and the capture, evaluation and presentation of biomechanical data. Numerous solutions are already available on the market. Standardized, mutually compatible data formats and structures are however still lacking.

Digital human models are software systems or extensions with which users can simulate and study certain anthropometric, biomechanical and physiological properties of human beings in virtual development environments. The focus lies on analysis of ergonomic issues such as visibility (e.g. for construction machinery in accordance with ISO 5006), accessibility and usability (EN ISO 14738) and application of force (DIN 33411, EN 1005-3, ISO 11228) during the operation of machinery. Postures adopted during the performance of work, for example at control stations and in offices and production areas, are also considered.

Standardized ergonomic methods (for example in accordance with DIN 1005-4, OWAS body posture analysis¹ or key indicator methods²) are usually implemented in ergonomic digital human models by means of software. They enable health risks to be assessed and, based on the results, prospective or corrective measures to be determined by which a work system can be optimized (for example in accordance with EN ISO 6385).

Application of digital ergonomics methods requires the relevant information on the work activity to be imported into the software. Posture and body movement are particularly relevant here. Although digital human models do generally enable different body dimensions and workflows to be created manually, the process is very time-consuming. Digital motion capture technologies constitute a more efficient approach.

The first capture systems, which were mechanical in nature, now date back several decades. The systems have however evolved significantly over the past decade in their usability and accuracy. Inertial and optical capture technologies are now widely used in industry and research. Inertial systems process the data stream from multiple sensors (accelerometers and gyroscopes). These are fitted to the body and detect acceleration and changes in joint angles. Optical systems employ cameras that detect markers (reference points) applied to the body, or calculate the progress of motion from a series of still images (synchronized RGB or depth image data), without the need for markers.

Advantages and disadvantages of the different technologies

Markerless single-camera systems (e.g. Microsoft Kinect) are inexpensive and suitable for mobile use. Conversely, calibrated camera systems employing markers on the person for motion detection (e.g. OptiTrack, Vicon) may attain very high capture accuracies in laboratory environments. Inertial motion capture systems (e.g. XSens MVN) represent a compromise: although based on sensor systems that usually also require calibration, they do not require fixed installation in the room. The accuracy of inertial systems is comparatively high, but decreases with increasing recording duration.

Finally, the wide range of technical options for capture is accompanied by a large number of data formats differing in their structure and content. The content differs for example in the accuracy, number and type of geometric representations of the body segments (position, absolute rotation, relative rotation), the hierarchical structure of the digital skeleton and the resolution on the time axis. Structural differences can be found in the presentation of the data (tables or hierarchies), the readability and the terms of the licence for use. Some formats constitute a de-facto standard (e.g. Biovision Hierarchy/BVH) but are not suitable for universal use, since they are not fully standardized. For this reason, publicly available research results often employ specially defined data formats, in most cases in the form of plain-text tables (CSV, comma-separated values).

Harmonized formats and interfaces are required

ISO/IEC 19774 proposes a standardized data structure for representation of a human figure. It consists of two parts: the architecture and animation of the motion data. Part 1 also specifies different levels of detail, Part 2 the animation of the captured motion. This specification is based on the research field of computer graphics. To date, computer graphics have only rarely been implemented in digital ergonomics, not least because they have not yet adequately been able to model the particular characteristics of ergonomics.

Digital methods can be used for products or workflows – even at the development stage – for estimation of the anticipated strain on human beings and for assessment of the ergonomic quality. Resource-intensive changes to a finished product/during subsequent operation can thus be reduced or avoided altogether. Car manufacturers have already developed dedicated solutions for assessing, at an early stage of development, the ergonomic quality of the passenger compartment with respect to visibility and accessibility. Workplaces, too, can already be planned and assessed digitally. To date, however, only stand-alone solutions for specific applications have been implemented. If they are to meet with widespread adoption, the individual methods must lend themselves to combination. The use of defined data formats to standardize interfaces is both beneficial and necessary.

Professor Martin Schmauder, Dresden Technical University
martin.schmauder@tu-dresden.de