Realization And Validation Of Using Virtual Reality To Evaluate Physical Ergonomic Attributes Of Physical Work

Musculoskeletal disorder (MSD) has been one of the most serious occupationalinjuries in industrialized countries. Studies indicate that MSD is closely related toinappropriate design of products and works with poor ergonomics considerations.Traditional ergonomics evaluation methods usually require building physical mock-upsto assess potential ergonomic problems of a design, and then revise the design. The"design-evaluation" iteration based on physical mock-ups is both time consuming andexpensive. The emerging virtual mock-ups based on virtual reality has advantages ofquickly and effectively evaluating and revising a design. However, validity of such anew method is a major concern.This dissertation proposed and realized a new method to conduct physicalergonomics evaluation for physical work based on motion tracking and virtual realitysimulation. Then two sets of experiments were designed and completed to study (1) thedifferences and correlations between evaluation results measured in virtual and realconditions, and (2) the impact of virtual reality feedback channels on the differences.The experimental results have demonstrated the feasibility of using virtual realitysimulation to conduct physical ergonomics evaluation, and have indicated the directionof how to improve the predicting accuracy of virtual reality based evaluation.Experiment one was conducted to compare ergonomics measurements for the samegroup of participants performing the same manual task under virtual environment (VE)and real environment (RE), respectively; then to analyze the differences and correlationsfor measurements under the two conditions. The experiment recruited30male workersfrom a mechanical manufacturing factory as the subjects. None of the subject has MSDas reported. The experiment has three within-group factors: working posture, number oftask repetition, and target size. Five ergonomics indices were measured, including armposture (elbow angle), task completion time (TCT), fatigue (reduction rate of maximumforce capacity, MFCR), body part discomfort (BPD), and ratings of perceived exertion(RPE). The results show:(1) the difference of elbow angle in VE and RE is statisticallysignificant, but the relative difference is small and has no effect on traditional postureanalysis methods;(2) ergonomics indices measured in VE is significantly worse than indices measured in RE, with longer task completion time, more accumulated fatigue,stronger BPD and RPE;(3) BPD and accumulated fatigue measured in VE is fairlycorrelated with the corresponding index measured in RE. Some literatures show thatpeople need extra cognitive workload to perform the same task in VE. This underlyingphenomenon is consistent with our experimental findings and can explain thedifferences and correlations of ergonomics indices measured in VE and RE.Experiment two compared and analyzed the impact of various feedback settings ofvirtual reality simulation on the evaluation results, in order to provide guidelines tominimize the errors of using measurements in VE to predict measurements in RE. Theexperiment recruited60workers from a mechanical manufacturing factory, none ofthem has MSD as reported, adverse loss of bare or corrected vision and hearing asmeasured on site. Four different feedback settings of VE were treated as between-groupfactor levels, including visual feedback only (V), visual and auditory feedback (VA),visual and tactile feedback (VT), visual and auditory and tactile feedback (VAT). Thesefour settings were compared with condition of RE, which was used as the base group.Each of the five groups had12subjects performing the same manual task under theircorresponding condition setting. In addition to the ergonomics indices measured inexperiment one, this experiment also measured the presence of VE and rated taskdifficulty (RTD). The results show:(1) Compared to RE group, all VE groups (V, VA,VT, and VAT) had worse ergonomics indices, indicated by longer TCT, more fatigue(higher MCFR), stronger BPD and RPE, higher RTD, and lower presence;(2) Amongthe four VE groups, V group had significantly worse ergonomics indices than the otherthree VE groups (VA, VT, and VAT);(3) No significance among VA, VT, and VATgroups. The conclusion is that extra auditory or tactile feedback is positive forsimulation accuracy, but auditory plus tactile is not needed.The ergonomics evaluation method proposed, realized and validated in thisdissertation is helpful for, and has the potential to be applied to, ergonomics evaluationin early design stage. The results and conclusions in experiment two is helpful to designand select more appropriate feedbacks for virtual reality system under economic andtechnical constraint. The system, method, and results also provide reference informationfor evaluation of other physical works.