3D Motion Capture Research Articles

Practical tips for teaching medicine in the metaverse

The metaverse is based on immersive technologies such as virtual and augmented reality, body tracking, tactile sensation, etc. A growing number of studies are demonstrating the potential of the metaverse as an attractive resource for learning medicine. However, in practice, medical teachers and students often encounter significant challenges when utilizing the underlying technologies, potentially leading to frustrating learning experiences. A significant part of the teaching time is often devoted to troubleshooting technical issues with the metaverse, and the medical content itself taking a backseat until students become proficient in navigating the metaverse environment. Therefore, it is essential to fit the metaverse's underlying technologies specifically for medical education, minimizing technical hurdles for both teachers and students. In this paper, we deal with this challenge and we present a collection of practical tips that serves as a guide for medical educators making decisions in this emerging field, where they may lack prior experience. Drawing on our observation with a cohort of 776 medical students, we conclude how to effectively identify, design, or implement educational applications tailored for efficient medical learning through the metaverse. Our work may support teachers considering metaverse learning platforms for their classrooms and it is a beneficial reference for the medical education community during the initial stages of implementing the metaverse for teaching.

Calibration to Differentiate Power Output by the Manual Wheelchair User from the Pushrim-Activated Power-Assisted Wheel on a Force-Instrumented Computer-Controlled Wheelchair Ergometer

To examine the biomechanical demands of manual wheelchair propulsion, it is crucial to determine the wheelchair user’s (WCU) force for propulsion technique parameter calculation. When using a pushrim-activated power-assisted wheelchair (PAPAW) on a wheelchair ergometer, a combined propulsion force from the WCU and PAPAW is exerted. To understand PAPAW’s assistance and distinguish the WCU’s force application from the force exerted by the PAPAW, both propulsion components must be assessed separately. In this study, a calibration of the PAPAW on an ergometer was developed to achieve this separation. The calibration consists of five steps: (I) Collecting data on force and velocity measured from the ergometer, along with electrical current and velocity from the PAPAW. (II) Synchronizing the velocity signals of the wheelchair ergometer and PAPAW using cross-correlation. (III) Calibrating the PAPAW’s electromotors to convert electrical current (mA) to force (N). A product-specific motor constant of 0.30, provided an average ICC of 0.563, indicating a moderate agreement between the raw ergometer data (N) and the motor constant-converted drive-rim (PAPAW) data (from mA to N). (IV) Subtracting the PAPAW’s force signal from the ergometer’s measured force to isolate forces generated by the WCU. (V) Using markerless motion capture to determine and validate the phase of hand contact with the handrim. This technical note provides an example of PAPAW calibration for researchers and clinicians. It emphasizes the importance of integrating this calibration into the development of PAPAW devices to reveal the complex interaction between PAPAW and WCU during wheelchair propulsion.

An Efficient Immersive Self-Training System for Hip-Hop Dance Performance with Automatic Evaluation Features

As a significant form of physical expression, dance demands ongoing training for skill enhancement, particularly in expressiveness. However, such training often faces restrictions related to location and time. Moreover, the evaluation of dance performance tends to be subjective, which necessitates the development of effective training methods and objective evaluation techniques. In this research, we introduce a self-training system for dance that employs VR technology to create an immersive training environment that facilitates a comprehensive understanding of three-dimensional dance movements. Furthermore, the system incorporates markerless motion capture technology to accurately record dancers’ movements in real time and translate them into the VR avatar. Additionally, the use of deep learning enables multi-perspective dance performance assessment, providing feedback to users to aid their repetitive practice. To enable deep learning-based dance evaluations, we established a dataset that incorporates data from beginner-level dances along with expert evaluations of those dances. This dataset was specifically curated for practitioners in a dance studio setting by using a total of four cameras to record dances. Expert annotations were obtained from various perspectives to provide a comprehensive evaluation. This study also proposes three unique automatic evaluation models. A comparative analysis of the models, particularly contrastive learning (and autoencoder)-based expression learning and a reference-guided model (where a model dancer’s performance serves as a reference), revealed that the reference-guided model achieved superior accuracy. The proposed method was able to predict dance performance ratings with an accuracy of approximately ±1 point on a 10-point scale, compared to ratings by professional coaches. Our findings open up novel possibilities for future dance training and evaluation systems.

Protocol to record and analyze primate leaping in three dimensions in the wild.

Several studies comparing primate locomotion under lab versus field conditions have shown the importance of implementing both types of studies, as each has their advantages and disadvantages. However, three-dimensional (3D) motion capture of primates has been challenging under natural conditions. In this study, we provide a detailed protocol on how to collect 3D biomechanical data on primate leaping in their natural habitat that can be widely implemented. To record primate locomotion in the dense forest we use modified GoPro Hero Black cameras with zoom lenses that can easily be carried around and set up on tripods. We outline details on how to obtain camera calibrations at greater heights and how to process the collected data using the MATLAB camera calibration app and the motion tracking software DLTdv8a. We further developed a new MATLAB application "WildLeap3D" to generate biomechanical performance metrics from the derived x, y, z coordinates of the leaps. We provide details on how to collect data on support diameter, compliance, and orientation, and combine these with the jumps to study locomotor performance in an ecological context. We successfully reconstructed leaps of wild primates in the 3D space under natural conditions and provided data on four representative leaps. We provide exemplar data on primate velocity and acceleration during a leap and show how our protocol can be used to analyze segmental kinematics. This study will help to make motion capture of freely moving animals more accessible and help further our knowledge about animal locomotion and movement.

Biomechanical Motion Changes in Adjacent and Noncontiguous Segments Following Single-Level Anterior Cervical Discectomy and Fusion: A Computed Tomography-Based 3D Motion Capture Study.

Anterior cervical discectomy and fusion (ACDF) is known to elicit adverse biomechanical effects on immediately adjacent segments; however, its impact on the kinematics of the remaining nonadjacent cervical levels has not been understood. This study aimed to explore the biomechanical impact of ACDF on kinematics beyond the immediate fusion site. We hypothesized that compensatory motion following single-level ACDF is not predictably distributed to adjacent segments due to compensation from noncontiguous levels. Six fresh-frozen cervical spines (C2-T1) underwent fluoroscopic screening and sagittal and coronal reformats from computed tomography scans and were utilized to grade segmental degeneration. Each specimen was tested to 30° of flexion and extension intact and following single-level ACDF at the C5-C6 level. The motions of each vertebral body were tracked using 3-dimensional (3D) motion capture into an inverse kinematics model, facilitating correlations between the 3D reconstruction from computed tomography images and the 3D motion capture data. This model was used to calculate each level's flexion/extension range of motion (ROM). Single-level fusion at the C5-C6 level across all specimens resulted in a significant motion reduction of -6.8° (P = 0.002). No significant change in ROM occurred in the immediate adjacent segments C4-C5 (P = 0.07) or C6-C7 (P = 0.15). Hypermobility was observed in 2 specimens (33%) exclusively in adjacent segments. In contrast, the other 4 spines (66%) displayed hypermobility at noncontiguous segments. Hypermobility occurred in 42% (5/12) of the adjacent segments, 28% (5/18) of the noncontiguous segments, and 50% (3/6) of the cervicothoracic segments. Single-level ACDF impacts ROM beyond adjacent segments, extending to noncontiguous levels. Compensatory motion, not limited to adjacent levels, may be influenced by degenerative changes in noncontiguous segments. Surprisingly, hypermobility may not occur in adjacent segments after ACDF. Overall, the multifaceted biomechanical effects of ACDF underscore the need for a comprehensive understanding of cervical spine dynamics beyond immediate adjacency, and it needs to be taken into consideration when planning single-level ACDF.

Running biomechanical and isokinetic strength differences in masters runners with and without Achilles tendinopathy

ABSTRACT Masters runners are an increasing proportion of the running community. The most significant musculoskeletal changes in runners occur after the age of 50 in addition to changes in injury rates and types, the most common being Achilles tendinopathy (AT). Previous evidence has suggested similarities between risk factors for AT and age-related changes that are focused at the hip and the ankle during the propulsive stage of running. The purpose of this study was to investigate biomechanical and peak torque association to AT in masters runners. Thirty-two masters runners over age 50 with AT (60.31 ± 8.37, n = 16) and without (59.94 ± 4.95 n = 16) were included. 3D motion capture and force plates were used to assess running biomechanics. A motor-driven dynamometer was used to assess isokinetic peak torque production. No significant differences in running biomechanics were found between masters runners with and without AT. Hip peak isokinetic torque production was found to be significantly less in masters runners with AT, but no significant differences in ankle plantarflexion peak isokinetic torque production were found. Masters runners with AT may be able to adapt their running biomechanics and muscular torque production during submaximal running efforts.

Automating Video-Based Two-Dimensional Motion Analysis in Sport? Implications for Gait Event Detection, Pose Estimation, and Performance Parameter Analysis.

Two-dimensional (2D) video is a common tool used during sports training and competition to analyze movement. In these videos, biomechanists determine key events, annotate joint centers, and calculate spatial, temporal, and kinematic parameters to provide performance reports to coaches and athletes. Automatic tools relying on computer vision and artificial intelligence methods hold promise to reduce the need for time-consuming manual methods. This study systematically analyzed the steps required to automate the video analysis workflow by investigating the applicability of a threshold-based event detection algorithm developed for 3D marker trajectories to 2D video data at four sampling rates; the agreement of 2D keypoints estimated by an off-the-shelf pose estimation model compared with gold-standard 3D marker trajectories projected to camera's field of view; and the influence of an offset in event detection on contact time and the sagittal knee joint angle at the key critical events of touch down and foot flat. Repeated measures limits of agreement were used to compare parameters determined by markerless and marker-based motion capture. Results highlighted that a minimum video sampling rate of 100 Hz is required to detect key events, and the limited applicability of 3D marker trajectory-based event detection algorithms when using 2D video. Although detected keypoints showed good agreement with the gold-standard, misidentification of key events-such as touch down by 20 ms resulted in knee compression angle differences of up to 20°. These findings emphasize the need for de novo accurate key event detection algorithms to automate 2D video analysis pipelines.

Robust Methods for Assessing the Characteristics of Locomotor Activity Based on Markerless Video Capture Data

Introduction. Analysis of locomotor activity is essential in a number of biomedical and pharmacological research designs, as well as environmental monitoring. The movement trajectories of biological objects can be represented by time series exhibiting a complex multicomponent structure and non-stationary dynamics, thus limiting the effectiveness of conventional correlation and spectral time series analysis methods. Recordings obtained using markerless technologies are typically characterized by enhanced noise levels, including both instrumental noise and anomalous errors associated with false estimates of the location of the points of interest, as well as gaps in the trajectories, promoting an urgent need in the development of robust methods to assess the characteristics of locomotor activity.Aim. Development of robust methods for assessing the characteristics of locomotor activity capable of efficient processing of noisy recordings obtained by markerless video-based motion capture systems.Materials and methods. In order to assess the characteristics of locomotor activity, the relative movements of body parts of laboratory animals were analyzed using the stability metrics of the mutual dynamics of their trajectories, their relative delays, as well as the relative duration of the recording fragments when relatively stable mutual dynamics could be observed. The local maxima of the cross-correlation function of two body fragments, the minima of the standard deviation of the difference between their Hilbert phases, as well as their relative delays, were used as the metrics of mutual dynamics.Results. The considered phase metrics were shown to explicitly reflect changes in locomotor activity, while the assessment of time delays using phase metric was shown to be prone to periodic error. The above limitation could be largely overcome using the correlation metrics, assuming that phase and correlation metrics could be combined.Conclusion. The proposed robust methods provide stable estimates of the characteristics of locomotor activity based on markerless video capture recordings, altogether increasing the efficiency of diagnostic procedures and assessment of the therapeutic effect during rehabilitation.

Validity of Inertial Measurement Unit (IMU Sensor) for Measurement of Cervical Spine Motion, Compared with Eight Optoelectronic 3D Cameras Under Spinal Immobilization Devices.

The assessment of cervical spine motion is critical for out-of-hospital patients who suffer traumatic spinal cord injuries, given the profound implications such injuries have on individual well-being and broader public health concerns. 3D Optoelectronic systems (BTS SmartDX) are standard devices for motion measurement, but their price, complexity, and size prevent them from being used outside of designated laboratories. This study was designed to evaluate the accuracy and reliability of an inertial measurement unit (IMU) in gauging cervical spine motion among healthy volunteers, using a 3D optoelectronic motion capture system as a reference. Twelve healthy volunteers participated in the study. They underwent lifting, transferring, and tilting simulations using a long spinal board, a Sked stretcher, and a vacuum mattress. During these simulations, cervical spine angular movements-including flexion-extension, axial rotation, and lateral flexion-were concurrently measured using the IMU and an optoelectronic device. We employed the Wilcoxon signed-rank test and the Bland-Altman plot to assess reliability and validity. A single statistically significant difference was observed between the two devices in the flexion-extension plane. The mean differences across all angular planes ranged from -1.129° to 1.053°, with the most pronounced difference noted in the lateral flexion plane. Ninety-five percent of the angular motion disparities ascertained by the SmartDX and IMU were less than 7.873° for the lateral flexion plane, 11.143° for the flexion-extension plane, and 25.382° for the axial rotation plane. The IMU device exhibited robust validity when assessing the angular motion of the cervical spine in the axial rotation plane and demonstrated commendable validity in both the lateral flexion and flexion-extension planes.

Four-quadrant propeller hydrodynamic performance mapping for improving ship motion predictions

On the path toward fully autonomous sea vessels, forecasting a ship’s exact velocity and position during its route plays a crucial role in dynamic positioning, target tracking, and autopilot operations of the unmanned body navigating toward predetermined locations. This paper addresses the prediction of the operational performance of a free-running submarine advancing in a straight route (in surge motion). Along with the forward advancing vessel (straight-ahead motion) the study covers all possible scenarios of ship’s surge, including crash-ahead, crash-back, and astern motions. Conventional maneuvering models cannot handle motions other than forward advancement due to the absence of propeller data in all four quadrants of hydrodynamic performance map. This study proposes an approach for predicting submarine performance in all these surge conditions by utilizing four-quadrant propeller performance and resistance test data. We developed an in-house code, SMot4QP, to simulate ship speed and position in the time domain. We obtained satisfying results for the straight-ahead and crash-ahead motions, while the crash-back and astern maneuvers require further refinement due to propeller wake interaction with the hull. The proposed method is capable of predicting the motions of all types of vessels using the ship’s resistance and four-quadrant propeller test results. Thus, SMot4QP offers a fast and robust alternative to computationally expensive free-running self-propulsion simulations for operational performance prediction in broader naval applications.

Poster 171: Comparative Analysis of Kinetic Variables Between Different Pitch Types in Professional Baseball Pitchers

Objectives: The persistent high rates of elbow injuries among professional baseball pitchers indicates a critical gap in our understanding of how a pitch influences load on the elbow. Traditionally, research has examined the maximum value of various kinetics experienced at the throwing arm during a pitch, especially elbow varus torque. The maximum elbow varus torque has been used as a surrogate for potential elbow injury; however, the elbow injury epidemic has still persisted. It is possible that analysis focused on only the maximum value misses the total load experienced on the throwing arm. Therefore, the purpose of this study was to investigate different metrics to evaluate elbow stress, including maximum elbow varus torque, elbow loading rate, and cumulative varus torque between different pitch types. By evaluating these aspects of elbow stress, our study aims to uncover previously overlooked kinetic variables that might be potential contributors to elbow injuries, bridging the gap between existing kinetic analyses and the persistent high injury rates. Methods: Data from a cohort of 41 professional baseball pitchers were utilized for this analysis (188 ± 6 cm; 91.9 ± 8.2 kg). Pitchers threw a series of fastball, changeup, and curveball pitches (fastball: 10 ± 3 pitches, changeup: 3 ± 1 pitches, curveball: 4 ± 1 pitches) captured using a 3D motion capture system (480 Hz). Kinetic variables were calculated for each pitch, including maximum elbow varus torque, loading torque rate (measured as the derivative of the varus torque curve), and cumulative elbow varus torque (measured as the area under the elbow varus torque curve). A linear mixed effects model was employed to examine within player associations between pitch type and ball velocity, maximum elbow varus torque, loading torque rate, and cumulative elbow varus torque. Statistical significance was set at p<0.01. Results: Our findings reveal that the fastball pitch created greater loading on the elbow compared to the change up and curveball pitches (Table 1). Pitchers threw their fastball with increased ball velocity (40.3 ± 0.6 m/s) compared to throwing their changeup (36.6 ± 0.5 m/s) and curveball (33.7±0.5 m/s) (p<0.001). Maximum elbow varus torque was also greater in the fastball pitch (93.7 ± 3.3 N∙m) compared to the changeup pitch (84.1 ± 2.4 N∙m) and curveball (91.0 ± 2.3 N∙m) (p<0.001); the curveball pitch had greater maximum torque compared to changeup pitches (p<0.001). The loading rate was 15% lower for the changeup pitch (690.0 ± 49.4 N∙m/s) (p<0.001) and 6% lower for the curveball (753.6 ± 60.3 N∙m/s) compared to the fastball (820.1 ± 72.2 N∙m/s) (p<0.001). The cumulative elbow varus torque for the fastball was 4% greater than the changeup pitch (2996.9 ± 92.9 vs 2894.1 ± 84.1 N∙m) and 7% greater than the curveball pitch (2996.9 ± 92.9 vs 2797.2 ± 85.8 N∙m) (p<0.001). Conclusions: This study reveals the importance of considering pitch type when evaluating kinetic variables in professional baseball pitchers. The observed differences in elbow varus torque, loading rate, and other kinematic metrics between fastball, changeup, and curveball pitches highlights the need for a comprehensive understanding of biomechanical variations associated with different pitch types. These findings have implications for optimizing pitcher training programs, injury prevention strategies, and player management. By unraveling the intricate connections between kinetic variables and pitch types, this research contributes to advancing the biomechanical knowledge within the field of orthopedics and sports science.

Disturbance observer based adaptive heading control for unmanned surface vehicle with event‐triggered and signal quantization

SummaryThis article delves into the adaptive heading tracking control of unmanned surface vehicle (USV) by incorporating an event‐triggered mechanism and signal quantization. The primary objective is to save communication resources while alleviating the burden of signal transmission. To address time‐varying external perturbations inherent in the control system, a disturbance observer is employed for precise estimation. Additionally, a linear model is introduced to delineate the procedure of quantization. By furnishing the controller with purpose‐designed quantized control input, the adaptive tracking control system can effectively track desired input without requiring any prior knowledge of the quantized parameters. The article substantiates its claims by demonstrating the system's stability in the absence of quantization considerations and the bounded nature of quantization errors through a series of presented lemmas. Further, the stability of the USV heading control system, integrated with an event‐triggered mechanism and signal quantization, is proofed in accordance with Lyapunov stability theory. Finally, the proposed strategy's efficacy and practical applicability are validated through experimental simulations.

Egocentric 3D Skeleton Learning in a Deep Neural Network Encodes Obese-like Motion Representations.

Obesity is a growing health concern, mainly caused by poor dietary habits. Yet, accurately tracking the diet and food intake of individuals with obesity is challenging. Although 3D motion capture technology is becoming increasingly important in healthcare, its potential for detecting early signs of obesity has not been fully explored. In this research, we used a deep LSTM network trained with individual identity (identity-trained deep LSTM network) to analyze 3D time-series skeleton data from mouse models with diet-induced obesity. First, we analyzed the data from two different viewpoints: allocentric and egocentric. Second, we trained various deep recurrent networks (e.g., RNN, GRU, LSTM) to predict the identity. Lastly, we tested whether these models effectively encode obese-like motion representations by training a support vector classifier with the latent features from the last layer. Our experimental results indicate that the optimal performance is achieved when utilizing an identity-trained deep LSTM network in conjunction with an egocentric viewpoint. This approach suggests a new way to use deep learning to spot health risks in mouse models of obesity and should be useful for detecting early signs of obesity in humans.

Evaluating Gesture Generation in a Large-scale Open Challenge: The GENEA Challenge 2022

This article reports on the second GENEA Challenge to benchmark data-driven automatic co-speech gesture generation. Participating teams used the same speech and motion dataset to build gesture-generation systems. Motion generated by all these systems was rendered to video using a standardised visualisation pipeline and evaluated in several large, crowdsourced user studies. Unlike when comparing different research articles, differences in results are here only due to differences between methods, enabling direct comparison between systems. The dataset was based on 18 hours of full-body motion capture, including fingers, of different persons engaging in a dyadic conversation. Ten teams participated in the challenge across two tiers: full-body and upper-body gesticulation. For each tier, we evaluated both the human-likeness of the gesture motion and its appropriateness for the specific speech signal. Our evaluations decouple human-likeness from gesture appropriateness, which has been a difficult problem in the field. The evaluation results show some synthetic gesture conditions being rated as significantly more human-like than 3D human motion capture. To the best of our knowledge, this has not been demonstrated before. On the other hand, all synthetic motion is found to be vastly less appropriate for the speech than the original motion-capture recordings. We also find that conventional objective metrics do not correlate well with subjective human-likeness ratings in this large evaluation. The one exception is the Fréchet gesture distance (FGD), which achieves a Kendall’s tau rank correlation of around -0.5. Based on the challenge results we formulate numerous recommendations for system building and evaluation.

1D1-2 Validation of a Joint Moment Analysis Method measured by Markerless Motion Capture

1D1-2 Validation of a Joint Moment Analysis Method measured by Markerless Motion Capture

The effect of age and bracing on vehicle occupant responses during sled-simulated evasive swerving

Objective Bracing can reduce adult occupants’ out-of-position postures, but it is unclear if this finding can be extended to child occupants. We investigated the effect of bracing, and age on the motion of vehicle occupants of different ages during sled-simulated pre-crash maneuvers. Methods Forty seatbelt restrained subjects (9–40 y.o.) experienced sled simulated evasive swerving maneuvers during a brace condition, where subjects actively brace before the maneuver onset, and an unbraced condition. A 3D motion capture system, electromyography (EMG), and seatbelt load cells captured head and trunk kinematics, muscle activation, and seat belt reaction forces, respectively. The effects of age and bracing on peak lateral head and trunk displacement were examined with Mixed Effects Model (p ≤ 0.05). Results In the braced condition, all subjects had reduced normalized peak head (0.09 ± 0.05) and trunk (0.11 ± 0.05) displacements compared to the unbraced condition (Head: 0.14 ± 0.06, Trunk: 0.17 ± 0.07) (p < 0.05). For the into-the-belt direction in the unbraced condition, children showed the smallest peak head and trunk displacement of all age groups (p < 0.01) while teens had the greatest peak head and trunk displacement compared to all other age groups (p < 0.02). Similar trends were found in the out-of-the-belt direction in the unbraced condition (p < 0.01). In the unbraced condition, all occupants showed greater shoulder seat belt reaction loads than the braced condition (unbraced 1.6–2.7 N/kg, braced 1.0–1.4 N/kg). Children also showed greater neck muscle activation (50–64% MVIC) compared to other muscles and older occupants regardless of the bracing condition or direction of motion. Conclusion Bracing reduced out-of-position postures across all ages. In the unbraced condition, children move less than teens and adults due to greater activation in the neck muscles. In the unbraced condition, teens showed the greatest motion, potentially due to their developing neuromotor control transitioning to a more mature stage. Across all age groups in the unbraced condition, occupants relied more heavily on the seat belt to maintain their position rather than their muscles to brace, suggesting the importance of good seatbelt-torso interaction in unbraced occupants.

Mapping the Behavioral Signatures of Shank3b Mice in Both Sexes

Autism spectrum disorders (ASD) are characterized by social and repetitive abnormalities. Although the ASD mouse model with Shank3b mutations is widely used in ASD research, the behavioral phenotype of this model has not been fully elucidated. Here, a 3D-motion capture system and linear discriminant analysis were used to comprehensively record and analyze the behavioral patterns of male and female Shank3b mutant mice. It was found that both sexes replicated the core and accompanied symptoms of ASD, with significant sex differences. Further, Shank3b heterozygous knockout mice exhibited distinct autistic behaviors, that were significantly different from those those observed in the wild type and homozygous knockout groups. Our findings provide evidence for the inclusion of both sexes and experimental approaches to efficiently characterize heterozygous transgenic models, which are more clinically relevant in autistic studies.

The effect of reclined seat-back angles on the LODC with and without a belt-positioning booster during far-side lateral-oblique impacts

Objective In frontal crashes belt-positioning boosters (BPB) may prevent submarining when the seatback is reclined. It is unclear if the BPB can also mitigate injuries in far-side lateral-oblique crashes in reclined conditions, where current restraints are less effective in reducing lateral excursion. This study aimed to understand reclined child injury risk during lateral-oblique impacts, with and without a booster seat, by using the Large Omni-Directional Child (LODC) test device. Methods The LODC was tested in nine lateral-oblique impact (80° from frontal) sled tests (target 31.3 km/h, duration 58 ms maximum, peak acceleration 21 g). Three seatback angles (25°, 45°, and 60°) with and without the BPB were compared on a production passenger seat with an integrated seatbelt. The LODC moved toward the buckle side during the far-side lateral-oblique impact. Abdominal pressure (left and right), seatbelt loads, anterior superioriliac spine (ASIS) forces, and pelvis lateral rotation were examined. The LODC head and knee excursions were extracted from a 3D-motion capture system. Results In the reclined noBPB condition, peak abdominal pressures on the buckle side reached up to 186 kPa and the ASIS forces demonstrated an early (∼ 50 ms) peak and a subsequent drop in the reclined noBPB conditions, suggesting that the belt slid into the abdomen. With the BPB, peak pelvis lateral rotation was greater than in the noBPB conditions but decreased with increasing reclined seatback angles (BPB: 31° to 36.5° vs no-BPB: −5° to −4.6°). Peak lap belt forces were greater in the reclined noBPB conditions (4.8-4.9 kN) compared to all conditions with the BPB (2.9-3.2 kN). The greatest lateral head excursion was observed in the BPB 25° condition (895 mm) and the lowest in the noBPB 45° condition (748 mm). Conclusions The BPB may prevent the lap belt intrusion into the abdomen in reclined configurations in far side lateral-oblique impacts. The greater pelvis lateral rotation and head displacement with the BPB may lead to contact with other rear-seated occupants and/or vehicle structures, although motion decreased with the BPB as the reclined seatback angle increased. This suggests that reclined seats may help reduce BPB-seated children’s lateral excursion in far-side lateral-oblique impacts.

Lower Limb Muscle Activation in Young Adults Walking in Water and on Land

Previous research has shown that exercise interventions requiring increased activation of the tibialis anterior (TA), the primary ankle dorsiflexor, can improve walking performance in individuals with foot drop. Correspondingly, heightened drag forces experienced during walking performed in water may augment TA activation during the swing phase of gait, potentially leading to improved walking gait on land. Therefore, this study aimed to compare surface electromyographic (sEMG) activation in the TA and medial gastrocnemius (GM) during gait performed in water versus on land. Thirty-eight healthy, recreationally active young adults, comprising 18 females and 20 males, participated in the study. Each participant completed 2 min walking trials under five conditions: land 2.5 mph, land 3.5 mph, water 2.5 mph, water 3.5 mph, and water 3.5 mph with added jet resistance. Stride kinematics were collected using 2-dimensional underwater motion capture. TA and GM, muscle activation magnitudes, were quantified using sEMG root-mean-square (RMS) amplitudes for both the swing and stance phases of walking. Additionally, TA and GM co-activation (Co-A) indices were estimated. Two-way within-subjects repeated measures analyses of variance were used to evaluate the main effects of and interactions between the environment and walking speed. Additionally, paired sample t-tests were conducted as a secondary analysis to investigate differences between walking in water at 3.5 mph with and without added jet resistance. Main effects and interactions were observed across various stride kinematics and sEMG measures. Notably, TA sEMG RMS during the swing phase of walking gait performed at 2.5 mph was 15% greater in water than on land (p &lt; 0.001). This effect increased when walking gait was performed at 3.5 mph (94%; p &lt; 0.001) and when jet resistance was added to the 3.5 mph condition (52%; p &lt; 0.001). Furthermore, TA Co-A was increased during the stance phase of gait in water compared to on land (p &lt; 0.001), while GM Co-A was reduced during the swing phase (p &lt; 0.001). The findings of this study offer compelling evidence supporting the efficacy of aquatic treadmill walking as a potential treatment for individuals suffering from foot drop. However, further research is needed to evaluate whether a causal relationship exists between heightened TA activation observed during aquatic treadmill walking and improvements in voluntary dorsiflexion during gait.

Skeleton Cluster Tracking for robust multi-view multi-person 3D human pose estimation

The multi-view 3D human pose estimation task relies on 2D human pose estimation for each view; however, severe occlusion, truncation, and human interaction lead to incorrect 2D human pose estimation for some views. The traditional “Matching-Lifting-Tracking” paradigm amplifies the incorrect 2D human pose into an incorrect 3D human pose, which significantly challenges the robustness of multi-view 3D human pose estimation. In this paper, we propose a novel method that tackles the inherent difficulties of the traditional paradigm. This method is rooted in the newly devised “Skeleton Pooling-Clustering-Tracking (SPCT)” paradigm. It initiates a 2D human pose estimation for each perspective. Then a symmetrical dilated network is created for skeleton pool estimation. Upon clustering the skeleton pool, we introduce and implement an innovative tracking method that is explicitly designed for the SPCT paradigm. The tracking method refines and filters the skeleton clusters, thereby enhancing the robustness of the multi-person 3D human pose estimation results. By coupling the skeleton pool with the tracking refinement process, our method obtains high-quality multi-person 3D human pose estimation results despite severe occlusions that produce erroneous 2D and 3D estimates. By employing the proposed SPCT paradigm and a computationally efficient network architecture, our method outperformed existing approaches regarding robustness on the Shelf, 4D Association, and CMU Panoptic datasets, and could be applied in practical scenarios such as markerless motion capture and animation production.