What to Know About the Industrial Ergonomics 3D Motion Capture Enhancement
What is VelocityEHS motion capture technology?
This technology harnesses artificial intelligence (AI) and computer vision to help you quickly and accurately assess musculoskeletal disorder (MSD) risk for jobs/tasks using a single video camera, on any mobile device, without sensors or manual assessment.
What is artificial intelligence (AI)?
Artificial intelligence is the science of making machines perform like humans. AI can perform “smart” tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and language translation.
What is computer vision?
It is a field of AI that uses machine learning and neural networks to teach computers and systems to obtain meaningful information from digital images, videos, and other visual inputs.
What is the VelocityEHS 3D model for motion capture?
The model enables images from a single video camera (i.e., monocular vision) to be used to identify joint center locations and joint angles in the three-dimensions (3D) for the entire video.
Are the 3D joint center locations and joint angles used to conduct the VelocityEHS Advanced MSD risk assessment (i.e., Advanced Assessment)?
Yes. The 3D joint center locations and joint angles (kinematics) are the inputs to the MSD risk assessment.
Is the Advanced Assessment using the 3D model more comprehensive than an assessment using the 2D model?
Yes. The MSD risk assessment still assesses nine body areas: both wrists, both elbows, both shoulders, the neck, back, and legs. It also still assesses the four physical or biomechanical risk factor categories: force, posture, duration, and frequency. It prioritizes and weighs the scoring of the physical risk factor categories based on research and uses research to establish threshold values for the physical risk factor categories. However, the enhanced 3D Advanced Assessment includes the automatic assessment of additional movements and has improved accuracy.
Does the 3D model evaluate hand/wrist movements in the Advanced Assessment?
Yes. The 3D model captures hand/wrist movements and measures MSD risk for the hand/wrists.
Can the automatic risk scoring of hand/wrist movement be overridden?
Yes. The automatic scoring for the hand/wrist movement can be overridden in the Baseline and Projected phases at any time. To ensure comparability between the Baseline and Follow-Up phase, users may only override wrist-bending posture scores in the Follow-Up phase if they have already done so in the Baseline phase of a given job assessment.
Does the 3D model evaluate additional back movements in the Advanced Assessment?
Yes. The 3D model evaluates back twisting and side bending, resulting in more accurate measurement of MSD risk for the back.
Is the 3D model more accurate than the previous 2D model?
Yes. The 3D model is 50% more accurate. Based on our research, the current Mean Absolute Angle Error (MAE) is 3.5° for all body segment angles.
Does the 3D model handle occluded or hidden body segments better than the previous 2D model?
Yes. The 3D model acquires 100% of data and scoring for the elbows, shoulders, back, and neck, even when body segments are hidden from view.
Why doesn’t the 3D model acquire 100% of data and scoring for the hands and legs?
Capturing high quality and more accurate data for the hands and legs is challenging when using one camera and a single video. Acquiring 100% of data and scoring for the hands and legs is currently not possible, as these body areas can be hidden from view for extended periods of time. However, recording the video in a manner that minimizes occlusion of the hands and legs will significantly improve hand and leg data capture.
Will the Advanced Assessment overall priority scoring be the same using the 3D model and the previous 2D model?
No. Using the enhanced 3D model, it is to be anticipated that there will be slight differences in the overall priority scoring output.
How much will the Advanced Assessment overall priority scoring using the 3D model vary from that of the previous 2D model?
These differences are typically small. On average, overall risk priority scores will decrease by 5-10%. During the validation process, the 3D model increased scoring accuracy, and scoring aligned well with scoring conducted by experienced Certified Professional Ergonomists (CPEs).
Will the 3D model, with better accuracy, increase the Advanced Assessment overall priority scoring for a job/task that was previously assessed using the 2D model?
Due to the use of higher quality and more accurate data, the overall MSD risk assessment and overall risk priority scores will typically be slightly lower (5-10% on average) and will better align with current risk threshold research. In certain cases, risk priority scores may be slightly (5-10%) higher, especially if the original (2D) assessment did not include an accurate assessment (by manual observation) of risk for the hands and wrists; the 3D automated scoring for these body segments may yield a slightly higher, but more accurate risk priority score.
Is the 3D model faster than the previous 2D model?
Yes. The 3D model is a minimum of two times faster. Currently, the video processing time is 1:6, meaning that it takes 6 minutes to process 1 minute of video. However, additional enhancements are underway to further reduce the video processing time to 1:3 or even 1:2 in the near term.
Why does the 3D model take so long to process a video?
While processing time has been significantly reduced (compared with the 2D model), the 3D model takes longer than other motion capture models to process a video and to complete an MSD risk assessment. This is because we want to ensure that only high quality and more accurate data enters our MSD risk assessments. To do this, we deploy several post-processing techniques:
- Data filtering and smoothing techniques to capture human movement data and to eliminate noisy or jumpy data from the risk assessment
- Worker tracking that locks onto the worker of interest and prevents the color-coded “skeleton” from jumping to an adjacent worker
- Removal of bad data for hand and leg movements (although rare) from the risk assessment
What are the benefits of the 3D motion capture enhancement?
The new motion capture model has been purpose built for MSD risk assessments. Benefits include the following:
- No occlusion: Captures 100% of data and scoring for the elbows, shoulders, back, and neck, even when segments are hidden from view
- More accurate: 50% more accurate at identifying body segment angles
- Faster: Cuts video processing time in half
- More comprehensive: Automatic scoring of wrist bending, side bending, back twisting, and other postures, to deliver the most comprehensive MSD risk assessment in the market
- Data visualization: Displays a new, interactive posture risk timeline visualization, to identify the moments of highest postural risk to ensure more effective analysis of the direct causes of MSD risk
Will the user experience for uploading and processing a new video using the 3D model change significantly?
No. The process is the same. Users will not see any changes to the core workflow or process that they are familiar with when conducting a risk assessment using the 3D model.
Can a current (2D) Advanced Assessment be reevaluated using the 3D model?
Yes. For job/tasks assessments completed using the original 2D model, a banner near the top of the job/task assessment page provides notification that the risk assessment was conducted using the original 2D model and allows (if applicable) the current video to be reprocessed using the 3D model.
If privacy features (face blocking or background blurring) were applied to a video in a current (2D) Advanced Assessment, can the video be reprocessed using the 3D model?
Yes, but it requires a few additional steps. Videos originally processed using privacy features cannot be directly rerun using the new 3D model. This is because the raw, unprocessed versions of these videos are deleted at the time of their original processing, so they are no longer available in the system. However, if you have the original, unprocessed videos on file, you can follow these steps to reprocess them with 3D motion capture:
- On the Job Assessment page, next to the color-coded figure, select Edit > Delete Video.
- Re-upload the original, unprocessed video of the job that was assessed (or, if necessary, re-record the operator performing the task).
- Follow the usual steps to process the video, including adding face blocking and/or background blurring.
Can privacy features (face-blocking or background blurring) still be used to process new videos using the 3D model?
Yes. No changes have been made to these important privacy features.
Should we reprocess all of our current (2D) Advanced Assessments using the 3D model?
It is not necessary to re-run all previous assessments using the 3D model. The risk scoring and prioritization of older job assessments remains valid and should be considered useful for identifying and eliminating MSD risk in jobs across your organization. We recommend you re-run only the Baseline videos for jobs/tasks for which you are actively planning to implement controls (ergonomics interventions or solutions). This ensures that the Baseline and Follow-up assessments are based on the same (3D) model.
Has the Direct Causes phase of the job improvement process changed?
Yes. There is a new and interactive Posture Risk over Time graph that helps you identify the moments in time in the video of the highest cumulative posture risk.
How will reprocessing a current (2D) Advanced Assessment using the 3D model affect the Risk Reduction Index (RRI)?
Just like any other change to a job assessment that you make, any difference in score between the original priority score and the updated score generated by reprocessing the video will be credited to the current month in the RRI.
Will reporting now only reflect Advanced Assessment data and scoring from the new 3D model?
No. Reports will reflect all data and scoring resulting from assessments using both the 2D and 3D models.
What is the launch date for the 3D model enhancement?
June 28th, 2024.
What is the additional cost for the 3D model enhancement?
There is no additional cost for this enhancement. As a current customer, your annual subscription fee to the VelocityEHS Industrial Ergonomics software solution stays the same.
Who gets access to the 3D model enhancement?
It will be deployed to all VelocityEHS Industrial Ergonomics customers who use the Advanced Assessment.
