The purpose of this research was to determine if the commercially available Perception Neuron motion capture system was valid and reliable in clinically relevant lower limb functional tasks. Twenty healthy participants performed two sessions on different days: gait, squat, single-leg squat, side lunge, forward lunge, and counter-movement jump. Seven IMUs and an OptiTrack system were used to record the three-dimensional joint kinematics of the lower extremity. To evaluate the performance, the multiple correlation coefficient (CMC) and the root mean square error (RMSE) of the waveforms as well as the difference and intraclass correlation coefficient (ICC) of discrete parameters were calculated. In all tasks, the CMC revealed fair to excellent waveform similarity (0.47–0.99) and the RMSE was between 3.57° and 13.14°. The difference between discrete parameters was lower than 14.54°. The repeatability analysis of waveforms showed that the CMC was between 0.54 and 0.95 and the RMSE was less than 5° in the frontal and transverse planes. The ICC of all joint angles in the IMU was general to excellent (0.57–1). Our findings showed that the IMU system might be utilized to evaluate lower extremity 3D joint kinematics in functional motions.

Understanding the fine-grained temporal structure of human actions and its semantic interpretation is beneficial to many real-world tasks, such as sports movements, rehabilitation exercises, and daily-life activities analysis. Current action segmentation methods mainly rely on deep neural networks to derive feature embedding of actions from motion data, while works on analyzing human actions in fine-granularity are still lacking due to the lack of clear and generic definitions of subactions and related datasets. On the other hand, the motion representations obtained in current action segmentation methods lack semantic or mathematical interpretability that can be used to evaluate action/subaction similarity in quantitative motion analysis. Toward the goal of fine-grained, interpretable, scalable, and efficient action segmentation, we propose a novel unsupervised action segmentation and distributed representation framework based on intuitive motion primitives defined on pose data. Metrics for comprehensive evaluation of the unsupervised fine-grained action segmentation task performance are proposed, and both public and self-constructed datasets are adopted in the experiments. The results show that the proposed method has good performance and generality across different subjects, datasets, and application scenarios.

Although commercial motion-capture systems have been widely used in various applications, the complex setup limits their application scenarios for ordinary consumers. To overcome the drawbacks of wearability, human posture reconstruction based on a few wearable sensors have been actively studied in recent years. In this paper, we propose a deep-learning-based sparse inertial sensor human posture reconstruction method. This method uses bidirectional recurrent neural network (Bi-RNN) to build an a priori model from a large motion dataset to build human motion, thereby the low-dimensional motion measurements are mapped to whole-body posture. To improve the motion reconstruction performance for specific application scenarios, two fundamental problems in the model construction are investigated: training data selection and sparse sensor placement. The problem of deep-learning training data selection is to select independent and identically distributed (IID) data for a certain scenario from the accumulated imbalanced motion dataset with sufficient information. We formulate the data selection into an optimization problem to obtain continuous and IID data segments, which comply with a small reference dataset collected from the target scenario. A two-step heuristic algorithm is proposed to solve the data selection problem. On the other hand, the optimal sensor placement problem is studied to exploit most information from partial observation of human movement. A method for evaluating the motion information amount of any group of wearable inertial sensors based on mutual information is proposed, and a greedy searching method is adopted to obtain the approximate optimal sensor placement of a given sensor number, so that the maximum motion information and minimum redundancy is achieved. Finally, the human posture reconstruction performance is evaluated with different training data and sensor placement selection methods, and experimental results show that the proposed method takes advantages in both posture reconstruction accuracy and model training time. In the 6 sensors configuration, the posture reconstruction errors of our model for walking, running, and playing basketball are 7.25°, 8.84°, and 14.13°, respectively.

Detecting foot-ground contact is important for inertial motion capture systems because it can provide kinematic constraints to improve human motion capture accuracy. The popular contact detection methods based on zero-velocity detection are only applicable to applications with regular movement patterns, such as pedestrian dead reckoning and gait analysis. As for arbitrary locomotion in motion capture, reliable foot-ground contact detection for universal inertial motion capture is challenging in the presence of motion kinematics and dynamics estimation errors. This paper proposes a novel foot-ground contact detection method for inertial motion capture by the fusion of inaccurate estimations of body kinematics and dynamics with a Naive Bayes probabilistic contact model. Based on physical analysis of contact, a series of kinematic and dynamic motion features calculated from motion capture are considered as observations of contact status classification. The proposed framework consists of offline model training based on a constructed labeled multi-person foot contact dataset, real-time contact detection, and an online model adaptation mechanism for enhancing detection performance of different users and setups. Quantitative evaluations of the proposed contact detection method are presented, and the resulting average foot-ground contact detection accuracy among various locomotion is 95.8%.

Objective: This study examined the effects of 12-week complex training (CT) programs on professional firefighters’ occupational activities, strength, and power.Methods: Thirty men professional firefighters were randomly assigned to the CT group (n = 15) and control group (n = 15). The CT group performed complex training and the control group completed resistance training (RT) twice a week over 12 weeks. The occupational activities, strength, and power were assessed at baseline and immediately after the intervention by measuring the performance of 100 m load-bearing run (100 m LR), 60 m shoulder ladder run (60 m SLR), 5 m × 20 m shuttle run (5 m × 20 m SR), 4th-floor climbing rope (4th-floor CR), countermovement jump with arm swing (CMJas), seated medicine-ball throw (SMT), one-repetition maximum bench press (1RM BP), and one-repetition maximum back squat (1RM BS).Results: The results showed that compared to RT, CT induced significantly greater improvements in 60 m SLR (p = 0.007), 4th-floor CR (p = 0.020), CMJas (p = 0.001), and SMT (p < 0.001).Conclusion: These findings suggest that CT is a novel intervention with great promise of improving professional firefighters’ occupational activities, strength, and power.

The study analysed the set-to-set variation in performance using match statistics of 146 completed main-draw matches in Australian Open and US Open 2016-2017 men’s singles. Comparisons of technical-tactical and physical performance variables were done between different sets; and the within-match coefficients of variation (CV) of these variables were contrasted between match winning and losing players. All comparisons were realized via standardized (Cohen’s d) mean differences and uncertainty in the true differences was assessed using non-clinical magnitude-based inferences. Results showed that there was possibly to very likely decreases in the serve, net and running related variables (mean difference, ±90%CL: -0.16, ±0.14 to -0.45, ±0.24, small) and an increase in the return and winner related variables (0.17, ±0.24 to 0.24, ±0.14, small) in the last sets when compared to the initial sets, indicating the influence of match fatigue and the player’s choice of match tactics and pacing strategy. Besides, winning players were revealed to have lower CV values in most of performance variables (-0.16, ±0.24 to -0.82, ±0.23, small to moderate) except for the second serve, winner, and physical performance variables (0.25, ±0.26 to 1.6, ±0.25, small to large), indicating that they would sacrifice the consistency to gain more aggressiveness and to dominate the match.