Stochastic-Biomechanic Modeling and Recognition of Human Movement Primitives, in Industry, Using Wearables

Abstract

In industry, ergonomists apply heuristic methods to determine workers’ exposure to ergonomic risks; however, current methods are limited to evaluating postures or measuring the duration and frequency of professional tasks. The work described here aims to deepen ergonomic analysis by using joint angles computed from inertial sensors to model the dynamics of professional movements and the collaboration between joints. This work is based on the hypothesis that with these models, it is possible to forecast workers’ posture and identify the joints contributing to the motion, which can later be used for ergonomic risk prevention. The modeling was based on the Gesture Operational Model, which uses autoregressive models to learn the dynamics of the joints by assuming associations between them. Euler angles were used for training to avoid forecasting errors such as bone stretching and invalid skeleton configurations, which commonly occur with models trained with joint positions. The statistical significance of the assumptions of each model was computed to determine the joints most involved in the movements. The forecasting performance of the models was evaluated, and the selection of joints was validated, by achieving a high gesture recognition performance. Finally, a sensitivity analysis was conducted to investigate the response of the system to disturbances and their effect on the posture.

Computational ergonomics for task delegation in Human-Robot Collaboration: spatiotemporal adaptation of the robot to the human through contactless gesture recognition

“…To summarize, highly adaptable and rapidly reconfigurable frameworks with real-time data processing capabilities are necessary to handle the industry's ergonomics challenges while simultaneously ensuring productivity. This paper expands on the authors' prior work on adaptive HRC [24], and automatic ergonomic monitoring [25], [26], [27], with the goal of creating a unified ergonomic and reconfigurable HRC framework.…”

“…Thus, HMMs were trained to recognize only 14 motion primitives with different ergonomic risk levels. Additionally, in a second prior study, where these motion primitives were analyzed by modeling their spatiotemporal dynamics through a Gestural Operational Model (GOM) [27], it was discovered that with only five inertial sensors, high recognition accuracy can be achieved. These sensors were placed on the lumbar spine, left upper arm, right shoulder, right upper leg, and left forearm.…”

Towards On-Device Dehydration Monitoring Using Machine Learning from Wearable Device’s Data

“…Though this calculation is considered an approximation of the total change in velocity, in this application, the exact value for the change at each time step was not important as we did not calculate the distance or time from it. In those cases, calculating Euler angles to predict the motion performed by a person would be more accurate [24]. It can reflect the effort and state of activity of the subject together with the accumulative change of the magnetometer and gyroscope magnitudes.…”

Fast Recognition Algorithm for Human Motion Posture Using Multimodal Bioinformation Fusion

“…At the same time, the category of unknown sample vector x is determined through the vector two norm, that is, the distance between sample vectors is analyzed, and the recognition process of human motion postures is realized through the measurement of distance. e proposed algorithm is compared with the multimodal fusion driver stress detection algorithm based on attention CNN LSTM in literature [6], the fall detection algorithm based on the spatiotemporal evolution of human posture in literature [7], the two-stage fall recognition algorithm based on human posture features in literature [8], the online adaptive prediction algorithm of human motion intention based on surface EMG signal in literature [9], and the stochastic biomechanical modeling and recognition algorithm of wearable devices for human motion in literature [10] and analyze the recognition performance of different algorithms.…”

“…However, the algorithm cannot accurately extract human motion features, leading to errors in recognition. Olivas-Padilla et al [10] studied wearable devices for stochastic biomechanical modeling and recognition of human motion, constructing a gesture-based manipulation model, which uses an autoregressive model to learn the dynamics of joints by assuming associations between them. e statistical significance of each model hypothesis is calculated to identify the joints most involved in the motion, and this result is combined to achieve human motion posture recognition.…”