Multi-task Optimization to Evaluate Workstation Suitability over a Population of Virtual Humans
Abstract
In industry, workstations need to be optimized in order to reduce work-related musculoskeletal disorders (WMSD) [1]. To avoid iterations over costly physical mock-ups, digital human modelling (DHM) tools can be used to assess the ergonomic risks on virtual humans performing the industrial tasks. In this context, the physical abilities and characteristics of each worker must be considered. A given activity can be performed in many different ways depending on the body dimensions or the physical strength of the worker [2], which leads to different levels of ergonomic risk and activity feasibility. Therefore, there is a need for tools to predict how well a workstation is suited to a population of workers. This would help in designing workstations better suited to the considered population.
In this work, we combine whole-body control and multi-task optimization to assess the suitability of workstation activities over a population of workers. Activities are simulated on a variety of virtual humans with a torque controller based on quadratic programming [3] [4]. The physics simulation guarantees the dynamic consistency of the virtual human postures
and torques, which helps with estimating biomechanical-based ergonomics indicators [5]. On top of that, the multi-task algorithm
[6] explores different ways of performing the activity, by searching for controller parameters and initial postures that optimizes the ergonomics and activity completion. As a result, the proposed approach generates suitability maps which enable an intuitive visualization of the workstation suitability over the considered population, as well as the optimized behaviors. On an example screwdriving activity, we show how our method helps to identify unsuitable workstation designs as well as high-risk behaviors in terms of ergonomics.
[1] S. Bevan, “Economic impact of musculoskeletal disorders (MSDs) on
work in Europe,” Best Practice and Research: Clinical Rheumatology,
vol. 29, no. 3, pp. 356–373, 2015.
[2] D. B. Chaffin, J. J. Faraway, X. Zhang, and C. Woolley, “Stature, Age,
and Gender Effects on Reach Motion Postures,” Human Factors: The
Journal of the Human Factors and Ergonomics Society, vol. 42, no. 3,
pp. 408–420, Sep. 2000.
[3] J. Salini, “Dynamic control for the task/posture coordination of humanoids
: toward synthesis of complex activities,” Ph.D. dissertation,
Université Pierre et Marie Curie, 2013.
[4] A. Del Prete, N. Mansard, O. E. Ramos, O. Stasse, and F. Nori, “Implementing
Torque Control with High-Ratio Gear Boxes and Without Joint-
Torque Sensors,” International Journal of Humanoid Robotics, vol. 13,
no. 1, 2016.
[5] P. Maurice, “Virtual ergonomics for the design of collaborative robots,”
Ph.D. dissertation, Universit´e Pierre et Marie Curie, 2015.
[6] J.-B. Mouret and G. Maguire, “Quality Diversity for Multi-task
Optimization,” in Proceedings of the 2020 Genetic and Evolutionary
Computation Conference, Jun. 2020, pp. 121–129.
Origin | Files produced by the author(s) |
---|