While 3D motion capture (MoCap) has enabled the high-resolution analysis of human movement, major limiting factors of these systems are their high cost and restriction to laboratory settings. This limits the generalizability of data captured in a controlled environment to movements performed during sport. The development of algorithms to support portable markerless MoCap systems provide a low-cost functional alternative to measure kinematics during sport-specific movement tasks in a real-world setting, with minimal invasiveness. PURPOSE: Assess the accuracy of a custom, portable, and markerless MoCap system in capturing lower limb joint kinematics during functional motor tasks compared to a gold-standard 3D MoCap system. METHODS: 15 subjects (12 females; age: 20±1 year) performed three motor tasks (8 trials each): walking, running, and cutting in a laboratory. The portable MoCap system comprised of 5 high-definition 2D cameras and proprietary pose-estimation software (wrnchAI ver 1.2.2) to capture lower limb landmark points and establish joint positions. Subjects were outfitted with 29 retroreflective markers for concurrent capture via a 10-camera Vicon 3D MoCap system. We extracted bilateral summary variables of ankle, knee, and hip angles in the sagittal and frontal planes. To assess the agreement between systems, coefficient of multiple determination (R 2 ), root mean squared error (RMSE), and mean joint angle estimation error (MEE) were calculated per trial. RESULTS: R 2 values revealed a strong relationship (range: 0.64-0.94). RMSE and MEE results indicate a strong to moderate accuracy (RMSE range: 0.83°-8.15°, MEE range: 0.86° -9.40°). See Table 1 for summary of results. CONCLUSION:The findings indicate the portable MoCap system with wrnchAI software effectively captured lower limb kinematics during functional motor tasks. Overall, portable MoCap technology can provide a cost-effective tool to assess athletes in a natural sport environment.Summary of R^2, RMSE, MEE results across conditions reported as mean ± standard deviation.

While 3D motion capture (MoCap) has enabled the high-resolution analysis of human movement, major limiting factors of these systems are their high cost and restriction to laboratory settings. This limits the generalizability of data captured in a controlled environment to movements performed during sport. The development of algorithms to support portable markerless MoCap systems provide a low-cost functional alternative to measure kinematics during sport-specific movement tasks in a real-world setting, with minimal invasiveness. PURPOSE: Assess the accuracy of a custom, portable, and markerless MoCap system in capturing lower limb joint kinematics during functional motor tasks compared to a gold-standard 3D MoCap system. METHODS: 15 subjects (12 females; age: 20±1 year) performed three motor tasks (8 trials each): walking, running, and cutting in a laboratory. The portable MoCap system comprised of 5 high-definition 2D cameras and proprietary pose-estimation software (wrnchAI ver 1.2.2) to capture lower limb landmark points and establish joint positions. Subjects were outfitted with 29 retroreflective markers for concurrent capture via a 10-camera Vicon 3D MoCap system. We extracted bilateral summary variables of ankle, knee, and hip angles in the sagittal and frontal planes. To assess the agreement between systems, coefficient of multiple determination (R 2 ), root mean squared error (RMSE), and mean joint angle estimation error (MEE) were calculated per trial. RESULTS: R 2 values revealed a strong relationship (range: 0.64-0.94). RMSE and MEE results indicate a strong to moderate accuracy (RMSE range: 0.83°-8.15°, MEE range: 0.86° -9.40°). See Table 1 for summary of results. CONCLUSION:The findings indicate the portable MoCap system with wrnchAI software effectively captured lower limb kinematics during functional motor tasks. Overall, portable MoCap technology can provide a cost-effective tool to assess athletes in a natural sport environment.