Motion-based prediction of external forces and moments and back loading during manual material handling tasks.

This paper evaluates a method for motion-based prediction of external forces and moments on manual material handling (MMH) tasks. From a set of hypothesized contact points between the subject and the environment (ground and load), external forces were calculated as the minimal forces at each contact point while ensuring the dynamics equilibrium. Ground reaction forces and moments (GRF&M) and load contact forces and moments (LCF&M) were computed from motion data alone. With an inverse dynamics method, the predicted data were then used to compute kinetic variables such as back loading. On a cohort of 65 subjects performing MMH tasks, the mean correlation coefficients between predicted and experimentally measured GRF for the vertical, antero-posterior and medio-lateral components were 0.91 (0.08), 0.95 (0.03) and 0.94 (0.08), respectively. The associated RMSE were 0.51 N/kg, 0.22 N/kg and 0.19 N/kg. The correlation coefficient between L5/S1 joint moments computed from predicted and measured data was 0.95 with a RMSE of 14 Nm for the flexion/extension component. In conclusion, this method allows the assessment of MMH tasks without force platforms, which increases the ecological aspect of the tasks studied and enables performance of dynamic analyses in real settings outside the laboratory.

Validation of a low-cost inertial motion capture system for whole-body motion analysis.

While some low-cost inertial motion capture (IMC) systems are now commercially available, generally, they have not been evaluated against gold standard optical motion capture (OMC). The objective was to validate the low-cost Neuron IMC system with OMC. Whole-body kinematics were recorded on five healthy subjects during manual handling of boxes for about 32 min while wearing 17 magnetic and inertial measurement units with Optotrak clusters serving as a reference. The kinematical model was calibrated anatomically for OMC and with poses for IMC. Local coordinate systems were aligned with angular velocities to dissociate differences due to technology or kinematical model. Descriptive statistics including the root mean square error (RMSE), coefficient of multiple correlation (CMC) and limits of agreement (LoA) were applied to the joint angle curves. The average technological error yielded 5.8° and 4.9° for RMSE, 0.87 and 0.96 for CMC and 0.4 ± 8.6° and -0.3 ± 6.0° for LoA about the frontal and transverse axes respectively, whereas the longitudinal axis yielded 10.5° for RMSE, 0.78 for CMC and 3.3 ± 13.1° for LoA. Differences due to technology and to the model contributed similarly to the total difference between IMC and OMC. For many joints and axes, RMSE stayed under 5°, CMC over 0.9 and LoA under 10°, especially for the transverse axis and lower limb. The Neuron low-cost IMC system showed potential for tracking complex human movements of long duration in a normal laboratory environment with a certain error level that may be suitable for many applications involving large IMC distribution.

A machine-learning method for classifying and analyzing foot placement: Application to manual material handling.

Foot placement strategy is an essential aspect in the study of movement involving full body displacement. To get beyond a qualitative analysis, this paper provides a foot placement classification and analysis method that can be used in sports, rehabilitation or ergonomics. The method is based on machine learning using a weighted k-nearest neighbors algorithm. The learning phase is performed by an observer who classifies a set of trials. The algorithm then automatically reproduces this classification on subsequent sets. The method also provides detailed analysis of foot placement strategy, such as estimating the average foot placements for each class or visualizing the variability of strategies. An example of applying the method to a manual material handling task demonstrates its usefulness. During the lifting phase, the foot placements were classified into four groups: front, contralateral foot behind, ipsilateral foot behind, and parallel. The accuracy of the classification, assessed with a holdout method, is about 97%. In this example, the classification method makes it possible to observe and analyze the handler's foot placement strategies with regards to the performed task.