Musculoskeletal mannequin and simulation
A full-body musculoskeletal mannequin created by Erica et al.18 was modified to supply a mannequin that’s appropriate for FFB harvesting exercise. The unique mannequin has 30 physique segments, 238 musculotendon actuators, 29 levels of freedom and eight higher extremity Hill-type muscle teams, together with the multifidus, rectus abdominis, psoas main, inside obliques, exterior obliques, quadratus lumborum, latissimus dorsi and erector spinae (longissimus and iliocostalis) muscle tissues. The thoracic and cervical components of the backbone, ribcage, head and scapula had been modelled as a inflexible physique, referred to as the torso, with spherical joints between six intervertebral joints (T12-L1, L1-L2, L2-L3, L3-L4, L4-L5, L5-S1). The vertebrae adopted a spinal rhythm that distributes the trunk movement over these six intervertebral joints by way of linear kinematic coordinate coupling constraints18. Three essential modifications had been made to the mannequin as listed under:
The unique Vary of Movement (ROM) of shoulder and elbow joints had been a lot higher than the Normative Vary of Movement (NROM); therefore they had been altered to the NROM19, as proven in Desk 1. This modification prevents a clumsy posture from being generated in OpenSim.
Hill-type biceps and triceps muscle tissues had been added. The muscle architectural parameters had been adopted from an current upper-extremity OpenSim mannequin by Daniel et al.20, utilized in pulling and pushing duties. The Optical Muscle Fibre Size (OMFL), Tendon Slack Size (TSL) and Pennation Angle (PA) of biceps and triceps had been sourced from a cadaver examine of21. The TSL of biceps was modified to protect its working ranges on the proximal joint it crosses22. The Peak Pressure (PF) was obtained based mostly on the measured muscle volumes of23 and isometric energy of24. General, these parameters (OMFL, PF, TSL, PA) had been based mostly on the information of younger male adults from completely different literatures21,23,24 and had been reported as follows:
OMFL: 11.6 cm (biceps lengthy), 13.2 cm (biceps quick), 13.4 cm (triceps lengthy), 11.4 cm (triceps lateral), 11.4 cm (triceps medial)
PF: 525.1 N (biceps lengthy), 316.8 N (biceps quick), 771.8 N (triceps lengthy), 717.5 N (triceps lateral), 717.5 N (triceps medial)
TSL: 27.8 cm (biceps lengthy), 20.0 cm (biceps quick), 14.3 cm (triceps lengthy), 9.8 cm (triceps lateral), 9.1 cm (triceps medial)
PA: 0° (biceps lengthy), 0° (biceps quick), 12° (triceps lengthy), 9° (triceps lateral), 9° (triceps medial)22
Reserve actuators had been added to the again joint (L5-S1), shoulder joints and elbow joints to supply additional actuation when the muscle tissues can’t generate adequate accelerations at a sure time. The optimum drive of the reserve actuator (FRA) have to be low sufficient to make sure that the muscle tissues are the principle contributor to the online joint moments (i.e., the joint second produced by the summed or internet impact of all of the constructions)1,25. Nonetheless, there is no such thing as a really useful FRA for harvesting exercise. Therefore, 5 completely different FRA values: 10 N, 20 N, 30 N, 40 N and 50 N had been investigated. They had been used to carry out the inverse dynamics and calculate the online joint moments. Static optimization was carried out to find out the reserve actuator moments. The height reserve actuator moments at completely different FRA had been normalized to the height internet joint moments. The outcomes are introduced in Desk 2. The height residual forces (FX, FY, FZ) and peak residual moments (MX, MY, MZ) had been added to the desk. Because the normalized peak reserve actuator moments must be lower than 10%26, solely the FRA of 30 N, 40 N and 50 N had been chosen to estimate the muscle activation.
Two assumptions had been made on this examine. The primary assumption is that the exerted harvesting drive of 303.5 N is the sum of the pulling drive produced by the harvesters (300 N27) and the burden of the harvesting instrument (3.5 N28). The second assumption is that the harvesting drive was equally distributed to the left and proper palms. Therefore, an exterior drive of 151.75 N was added to every hand. The exerted harvesting drive was present in27, which investigated the drive and power required to carry out the harvesting movement with a take a look at jig. The authors then carried out a purposeful take a look at on their prototype magnetic oil palm cutter at an oil palm plantation and the harvesting drive was discovered to be cheap to reap the fruit27. The burden of the harvesting instrument was obtained in28 as a result of it’s the harvesting instrument utilized by the harvesters on this examine.
Inverse kinematics was carried out to calculate the joint angle utilizing orientations measured by the IMU. This joint angle is denoted as θOS. Static optimization was then used to estimate the muscle activation. It resolves the online joint second into particular person muscle drive at every on the spot whereas minimizing the sum of squared muscle activation29. It’s the most typical method to calculate muscle activation in a dynamic exercise attributable to its robustness, greater effectivity and independence from the experimental EMG30. The simulated activations of all of the muscle fascicles for every muscle had been added collectively18. For instance, the estimated muscle activation of the longissimus is the full activations of the lumbar and thoracic parts of the longissimus thoracis. The identical applies to the multifidus, biceps and triceps.
The experiment was carried out at a Malaysia oil palm plantation. It concerned six skilled right-handed male oil palm FFB harvesters (33.5 ± 6.0 years previous, 168.83 ± 4.74 cm, 56.83 ± 4.26 kg). They offered written consent after they had been briefed on the experiment goals and procedures. Knowledgeable written consent for publication was obtained. This examine was reviewed and permitted by Monash College Human Analysis Ethics Committee (MUHREC).
Six IMU sensors (APDM OPAL, Portland, OR, USA) had been positioned on the harvesters‘ sternum, lumbar, higher arms and wrists, as proven in Fig. 1a. EMG electrodes (Biosignalplux, Lisbon, Portugal) had been positioned on the best facet of the higher extremities to measure the muscle activations of the biceps, triceps, multifidus and longissimus with reference electrode on the C7 area (Fig. 1b and c). The areas of IMU sensors and EMG electrodes had been based mostly on the suggestions of the sensor proprietary software program, Moveo Explorer31 and SENIAM conference32, respectively. Each sensors had been synchronized utilizing an exterior set off. The sampling frequency of IMU and EMG sensors had been 128 Hz and 1000 Hz, respectively. The harvesters had been requested to face upright and stay stationary for 3 seconds for sensor calibration. They then carried out the harvesting exercise for one minute. The harvesters solely harvested FFB from timber with a peak between 3 and 5 m for consistency functions. A video digital camera was used to report the harvesting exercise.
The IMU information had been filtered with a 4th order Butterworth low-pass filter with a cut-off frequency of 6 Hz. The EMG information had been processed utilizing codes written in MATLAB (Mathworks, Nantucket, MA, USA). The info had been filtered with a band-pass filter with 20 Hz and 450 Hz cut-off frequencies. They had been then full-wave rectified and smoothened utilizing a moving-average filter with a window measurement of 1000 ms33. Peak Dynamic Methodology was utilized to normalize the EMG information of every muscle to the utmost muscle activation of every harvester. This technique replaces the standard Most Voluntary Contraction (MVC) normalization technique34 due to the remoteness of the plantation websites, which made it difficult to amass the muscle MVC.
From the recorded video footage, it may be noticed that the harvesting course of is advanced. The harvester wants to maneuver across the tree, determine the ripe FFB and reduce the fronds, which may interrupt the FFB harvesting. The harvester then positions the sickle on the FFB stalk and pull the sickle downward to chop the stalk, as proven in Fig. 2. For a extra correct and consultant evaluation, the IMU and EMG information for every harvesting movement of various time lengths had been extracted, linearly interpolated and averaged to supply a harvesting movement with the identical timeframe35.
The joint angle θos was in contrast and validated with the joint angle obtained from the IMU proprietary software program, Moveo Explorer. The Pearson Correlation Coefficient (R) was calculated to find out the energy of affiliation between them. The nearer the R to zero, the weaker the energy of affiliation. Root Imply Sq. Error (RMSE) was additionally calculated to determine the error between them. A smaller RMSE worth signifies a smaller distinction between them. Nonetheless, there is no such thing as a clear guideline on the suitable R and RMSE values. Therefore, they had been in contrast with the latest work36, which covers the vary of R and RMSE that present good to wonderful validity.
For comparability functions, the muscle activations measured by EMG had been resampled to 128 Hz to match the IMU sampling fee. They had been then normalized to the utmost muscle activation of every harvester. The simulated normalized muscle activations had been then in contrast and validated with experimental information. The R and Imply Absolute Error (MAE) had been calculated and in contrast with research37,38, that categorize the vary of R and MAE into completely different qualities of correlation, proven in Desk 3.
High quality of harvesting posture
Apart from the joint angle and muscle activation, the harvesting mannequin was additionally used to compute the joint second, muscle drive and muscle activation of the opposite muscle tissues that weren’t measured through the experiment i.e., rectus abdominis, iliocostalis, exterior indirect, inside indirect and latissimus dorsi. The height joint moments had been normalized to every harvester’s physique weight, averaged and tabulated. Greater joint second signifies that the harvester faces a better threat of MSD when performing that joint movement, and vice versa. To determine the lively muscle used throughout harvesting, the normalized peak muscle activations and the normalized peak muscle forces at completely different FRA of 30 N, 40 N and 50 N had been calculated, expressed in share after which ranked accordingly.
The utmost ROM of every joint was recognized to find out the potential anxious joint movement. The Discomfort Worth (DV)17 was used as a measure to determine the standard of the harvesting posture. It makes use of the ROM of the joint to find out the DV after which classifies it into completely different ranges of perceived discomfort that correspond to the nice, so-so and poor postures. Completely different joint motions have completely different DV. The connection between NROM, DV and high quality of posture for the again flexion is proven in Desk 4. For instance, if the ROM of the harvester’s again flexion is 39° and the NROM of this movement is 80°19, the harvester’s again flexion can be equal to 48.75% of the NROM and may have a DV worth of twenty-two.08, which is related to a so-so posture. The next DV worth corresponds to a better threat of MSD and signifies that the joint experiences higher stress. These outcomes had been evaluated along with the simulated muscle activation, muscle drive and joint second to determine the muscle or joint that has a better threat of MSD.