Development and evaluation of a metrological concept for collaborative robots

Status:

completed 03/2022

Aims:

When using collaborative robots, the biomechanical limit values from the ISO/TS 15066 standard or from DGUV informative publication 080 published by the expert committee for woodworking and metalworking (FBHM) of the German Social Accident Insurance (DGUV) must be complied with in the event of a collision. Compliance with these values needs to be verified through measurements before a workstation with a collaborative robot is commissioned, and recurring tests must be performed in accordance with §14 BetrSichV (national Ordinance on Industrial Safety and Health).

In earlier projects, researchers were able to derive simplified tissue models by researching the literature. As part of this process, 29 defined locations of the body were assigned to 12 defined body regions. This meant that complex physical characteristics could be transferred to a simple system consisting of two compression elements, enabling the deformation behaviour of body regions that occur in the event of a collision to be simulated in a practical setting using a measurement device. For the measurement, the biomechanical properties of the various body regions are replicated using springs and rubber materials, or a combination of the two. The problem here is the high number of variants. This method requires either measurement devices with exchangeable springs, which are difficult to handle, or a separate measurement device for each spring constant. For this reason, measurements in companies are often performed using only two springs with the hardest spring constant. This results in incorrect assessment of the actual hazard situation at the workstation with a collaborative robot, which, in turn, results in unfavourable programming of the safety parameters on the plant, potentially resulting in a higher mechanical stress than that which is permissible.

A sample evaluation of recent studies from the German Social Accident Insurance Institution for the Woodworking and Metalworking Industries (BGHM) and the German Social Accident Insurance (DGUV) on the topic of collaborative robots (FP411 and FP430) found that the method of replicating biomechanical characteristics that is currently used in the measurement device – which involves using a combination of springs and rubber material – is not optimal for some body regions, and that the tissue properties were not sufficiently replicated. An adjustment to the measurement concept and an evaluation of the compression characteristics is required. In this context, the high number of variations should also be analysed and concepts to improve the method by simplifying it should be developed.

Activities/Methods:

On the one hand, in cooperation with Fraunhofer IFF, the measurement data collected from human subjects studies conducted by the BGHM and the DGUV on the topic of collaborative robots was analysed with regard to the characteristic values . For this purpose, a biomechanical corridor was defined for each body location, and typical stiffness parameters were derived based on these biomechanical corridors. These stiffness parameters are to be used in measurement instruments to correctly assess mechanical hazards. As a result, it is now possible to map out the empirically ascertained limit values with the corresponding biomechanical properties of a person on the measurement instrument, and therefore to carry out correct assessments.

On the other hand, the total number of body regions was reduced by clustering the 29 body locations so that a practical number of variants can be used for the measurement. This makes the measurement easier to implement in companies. The five determined variants were implemented and tested using pressure and force measurement devices at the Institute for Occupational Safety and Health of the German Social Accident Insurance (IFA). During this process, measurement uncertainties were taken into account. For validation purposes, measurements were carried out using a pendulum impact tester and on three robot models. In addition, a conversion table was created at the Institute for Occupational Safety and Health of the German Social Accident Insurance (IFA). This table enables optimised and simplified evaluation by providing energy-based figures.

Results:

A total of 24 biomechanical corridors were determined for the empirically ascertained limit values from projects FP411 and FP430, whereby the dominant and non-dominant sides of the body were grouped together. The force deformation data originates from investigations in which an F-Q10 contact body was used to determine pressure limit values for semi-sharp contact. The force limit values were determined using an elastic contact body (FZ30), which meant that a scaling process for blunt contact was necessary. Stiffness parameters that should be used in measurement instruments were derived from the biomechanical corridors. For example, for the measurements on the forearm muscles, a force of 170 N with a medium stiffness of 21 N/mm was determined for the 75th percentile of the onset of pain in the event of blunt contact.

To enable practical implementation, the body locations were grouped into five different clusters. The curves toe of the biomechanical response corridor was not taken into account. The clusters are almost logarithmically distributed from stiff (150 N/mm) to very soft (10 N/mm), whereby a 7-mm-thick rubber material with a Shore A hardness of 70 was used as a top layer, resulting in the real spring rate being reduced slightly due to the series connection of the two compression elements. The top layer and the springs used are already available in measurement devices on the market. As an example, a measurement device with a 25-N/mm spring and the top layer provides a good replication of the stiffness of the forearm muscle. Overall, a maximum error of 25% was permitted for the clusters. In addition, an energy-based conversion table was developed for improved comparability and optimisation. This table allows for potential deviations to be offset, resulting in fewer losses.

In addition to using one of the two hardest springs, it is also recommended to use one of the two softest variants, as these better replicates soft tissue. Unnecessary stress on humans caused by incorrect control can be avoided through the correct configuration of the safety parameters of collaborative robots.

The findings are now being transferred to a simplified measurement specification with a reduced number of measurements and will serve as a basis for international standardisation. Example measurements performed on various robot models demonstrated that, for lightweight models and blunt contact in the hand and arm region, speeds of up to 400 mm/s can be reached. However, the permissible speeds decrease for larger effective masses or semi-sharp contours.

Last Update: