Högskolan i Skövde

Digital manikins give a powerful aid in evaluating assembly station ergonomics. A proper verification of an assembly station can avoid costly changes that might occur later on due to injuries and discomfort among the workers.

However, how to select relevant test manikins for an evaluation is a non-trivial task, since the work of identifying which anthropometric variables that are critical is a tedious and time consuming process.

Usually only a few variables are selected as a base for building the test manikins. Even if there exist Digital Human Modeling (DHM) software which allow the user to evaluate batches of manikins, the designer still have to select the anthropometric variables of those batches. When several dimensions are considered, the designer have to either use a set of predefined manikins, or determine which anthropometric variables to test and generate manikins based on the confidence intervals of these variables.

When considering more complex assembly tasks, is it then true that these predetermined test manikins cover all the cases, or does there exist manikins that suffer from bad ergonomics even though all the test manikins turned out well?

In this paper, we propose a new algorithm for automatically building a set of test manikins. The set is iteratively constructed from the ergonomics results obtained by simulating the assembly operation. Different manikins perform the assembly operation and the ergonomics is evaluated. The anthropometric variables which affect the ergonomics are identified and used to iteratively build up the next manikin. In this way the test manikins are always selected throughout the whole set instead of only considering the boundary manikins, or assuming that the same set of predetermined manikins represents the entire set in every assembly operation.

The algorithm has been compared with a boundary method, and the results shows that the algorithm can find manikins with worse ergonomics than those tested by the boundary method.