No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Automatic identification and counting of repetitive actions related to an industrial worker
... Some authors filter the raw data with band-pass [33] and low-pass configurations before computing features [9,19,20,26,32,34], but they do not use another algorithm to clean the data. Usually, the authors place greater emphasis on feature engineering and model building and even achieve accuracies of up to 99.98% [30]. ...
By observing the actions taken by operators, it is possible to determine the risk level of a work task. One method for achieving this is the recognition of human activity using biosignals and inertial measurements provided to a machine learning algorithm performing such recognition. The aim of this research is to propose a method to automatically recognize physical exertion and reduce noise as much as possible towards the automation of the Job Strain Index (JSI) assessment by using a motion capture wearable device (MindRove armband) and training a quadratic support vector machine (QSVM) model, which is responsible for predicting the exertion depending on the patterns identified. The highest accuracy of the QSVM model was 95.7%, which was achieved by filtering the data, removing outliers and offsets, and performing zero calibration; in addition, EMG signals were normalized. It was determined that, given the job strain index’s purpose, physical exertion detection is crucial to computing its intensity in future work.
... In their work, Taborri et al. [27] implemented the following algorithms, one for recognizing activities based on SVMs and one for counting actions related to workers in the industry. Twenty-three body-worn sensors collected data from the participants, which were divided into windows of 0.6 s and had features such as mean, standard deviation, maximum, and minimum, were computed for each activity. ...
This paper presents a novel approach for counting hand-performed activities using deep learning and inertial measurement units (IMUs). The particular challenge in this task is finding the correct window size for capturing activities with different durations. Traditionally, fixed window sizes have been used, which occasionally result in incorrectly represented activities. To address this limitation, we propose segmenting the time series data into variable-length sequences using ragged tensors to store and process the data. Additionally, our approach utilizes weakly labeled data to simplify the annotation process and reduce the time to prepare annotated data for machine learning algorithms. Thus, the model receives only partial information about the performed activity. Therefore, we propose an LSTM-based architecture, which takes into account both the ragged tensors and the weak labels. To the best of our knowledge, no prior studies attempted counting utilizing variable-size IMU acceleration data with relatively low computational requirements using the number of completed repetitions of hand-performed activities as a label. Hence, we present the data segmentation method we employed and the model architecture that we implemented to show the effectiveness of our approach. Our results are evaluated using the Skoda public dataset for Human activity recognition (HAR) and demonstrate a repetition error of ± 1 even in the most challenging cases. The findings of this study have applications and can be beneficial for various fields, including healthcare, sports and fitness, human–computer interaction, robotics, and the manufacturing industry.
... Barkokebas et al. (2022) proposed a virtual reality (VR)-motion capture (MOCAP)-based ergonomic assessment system based on REBA and RULA to support workstation development in offsite construction. In addition to using existing ergonomic rules that systematically evaluate postural behavior, OCRA (Occupational Repetitive Action) (Colombini et al. 2013;Occhipinti 1998) Repetitive movements of the upper limbs • Inertial sensors and an algorithm based on Support Vector (Taborri et al. 2019). ...
Work-related musculoskeletal disorders (WMSDs) are the main causes of physical diseases of workers in the construction industry. The occurrence of WMSDs can result in project delays, cost overruns, workers' worsened health status, and even unanticipated safety risks. Ergonomic rules and instrument-based methods have enabled automatic postural ergonomic evaluation and intervention, while a research gap lies in how to consider more comprehensive personalized factors and how to integrate individual risks for project-level risk control. This study developed a two-hierarchy method to assess the ergonomic risk for both individual workers and the entire project. To alleviate the problems of data insufficiency and imprecision in real-life construction projects, experts' evaluations were incorporated with existing instrument-based methods. A methodology combining fuzzy theory, D-S evidence theory, and Bayesian network was employed to deal with fuzziness, uncertainty, and conflicting judgments, and to provide a probabilistic assessment. To measure the potential impact on projects from the perspectives both of cost and productivity, this study also developed a practical framework for project integration. A real-life case study was conducted to validate the approach and demonstrated its applicability. The results of sensitivity analysis indicate that previous WMSD records, age, working posture, and working intensity are the top critical risk factors that have the most significant impact on individuals' WMSD risk status. This study provides deeper insights into WMSD risk assessment, which can facilitate personalized instructions and intervention for individual WMSD risk control, and support managerial initiatives and decisions to mitigate the negative effects on construction projects.
... Thus, to the best of the authors' knowledges, the validation of an automatic procedure for the recognition and counting of the TA needed for the computation of the frequency factor, mandatory for the OCRA value, is still required. From this perspective, this paper extends the preliminary work presented in [29] and aims at understanding if it is possible to find an innovative experimental procedure able to automatically count dynamic technical actions needed for the computation of the frequency factor. To achieve this aim, we proposed a novel procedure and tested it in industrial scenarios, by discussing four different case studies. ...
OCRA (OCcupational Repetitive Action) is currently one of the most widespread procedures for assessing biomechanical risks related to upper limb repetitive movements. Frequency factor of the technical actions represents one of the OCRA elements. Actually, the frequency factor computation is based on workcycle video analysis, which is time-consuming and may lead to up to 30% of intra-operator variability. This paper aims at proposing an innovative procedure for the automatic counting of dynamic technical actions on the basis of inertial data. More specifically, a threshold-based algorithm was tested in four industrial case studies, involving a cohort of 20 workers. Nine combinations of the algorithm were tested by varying threshold values related to time and amplitude. The computation of frequency factor showed an average relative error lower than 5.7% in all industrial-based case studies after the appropriate selection of the time and amplitude threshold values. These findings open the possibility to use the threshold-based algorithm proposed here for the automatic computation of OCRA frequency factor, avoiding the time efforts in video analysis.
Modern manufacturing faces significant challenges, including efficiency bottlenecks and high error rates in manual assembly operations. To address these challenges, we implement artificial intelligence (AI) and propose a gaze-driven assembly assistant system that leverages artificial intelligence for human-centered smart manufacturing. Our system processes video inputs of assembly activities using a Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) network for assembly step recognition, a Transformer network for repetitive action counting, and a gaze tracker for eye gaze estimation. The application of AI integrates the outputs of these tasks to deliver real-time visual assistance through a software interface that displays relevant tools, parts, and procedural instructions based on recognized steps and gaze data. Experimental results demonstrate the system's high performance, achieving 98.36% accuracy in assembly step recognition, a mean absolute error (MAE) of 4.37%, and an off-by-one accuracy (OBOA) of 95.88% in action counting. Compared to existing solutions, our gaze-driven assistant offers superior precision and efficiency, providing a scalable and adaptable framework suitable for complex and large-scale manufacturing environments.
We live in a manufactured world and the manufacturing and production industry is one of the primary sources of economic prosperity for most countries. Industries strive to produce the right goods for the right customer at the right time, location, quantity, quality, and cost. In addition to this overarching objective, they are faced with a set of challenges such as meeting increased customer requirements, increasing complexity of products and supply chains, ensuring machine and workforce productivity, and empowering the workforce by creating opportunities for training and lifelong learning. To meet these challenges organizations have leveraged technologies in production systems since the beginning of industrialization. In the 1960s, two approaches to improve production emerged: automation and augmentation. Until recently, physical and mechanical automation dominated industrial research. However, with the emergence of the Fourth Industrial Revolution, the potential of Human-Computer Interaction (HCI) serving a transformative role in manufacturing has gained significant interest and relevance in industry. Where full automation is ineffective or infeasible, Operator Assistance Systems (OAS) can augment workers’ cognitive or physical capabilities. In this monograph, the authors frame OAS as a subset of HCI systems designed for the purpose of workforce augmentation in production systems. However, while OAS are anticipated to address key needs in industry, a challenge for both OAS researchers and industrial practitioners is to identify the most promising applications of OAS and justify them from a value-added perspective. This monograph addresses the challenges in OAS by presenting a systematic literature review revealing 11 application areas for OAS and 12 approaches for assessing the value-added of OAS. The authors also discuss implications for OAS, with a particular focus on integration in industry.
In industry, the Fourth Industrial Revolution is transforming the roles of people, technology and work on the shop floor. Despite ongoing strides towards automation, people are anticipated to remain integral contributors in future manufacturing. Where full automation is ineffective or in-feasible, Operator Assistance Systems (OAS) can augment workers' cognitive or physical capabilities. We frame OAS as a subset of Human-Computer Interaction (HCI) systems designed for the purpose of workforce augmentation in production systems. However, while OAS are anticipated to address key needs in industry, a challenge for both OAS researchers and industrial practitioners is to identify the most promising applications of OAS and justify them from a value-added perspective. This paper addresses this challenge by presenting a systematic literature review of 2,928 papers, revealing (a) 11 application areas for OAS; and (b) 12 approaches for assessing the value-added of OAS. Moreover, we discuss implications for OAS, with a particular focus on integrating OAS in industry.
A high estimate of not only workplace fatal injuries but also nonfatal injuries and illnesses via overexertion, and contact with objects, equipment, and machinery workplace has been reported over the years. To address this issue, the Occupational Health and Safety (OHS) program has put an emphasis on policy and regulation for prevention, protection, and improvement of the individuals’ health related to the working conditions and industrial environment. In the last 10 years, the Internet of Things (IoT) has matured to seamlessly enable real-time communications and cooperation among machines, environments, and humans using data analytics. Thus, IoT offers potential technical solutions for the prevention and protection of workplace injuries and illnesses. Moreover, IoT invites opportunities for collaboration with OHS in various industries. This article presents a systematic mapping study of the literature to address the impact of IoT on occupational well-being, analyzing the progress of Industry 4.0 during the last decade. This study systematizes the literature providing a taxonomy of the area through the results of four general, four focused, and two statistical research questions. These questions outline industrial environments and aspects of health concerning workers’ well-being, concentrating on a human-centered approach leveraged by physiological measurements and psychological health. In addition, this paper explores questions regarding IoT’s technical components, such as sensors, devices, and communication technologies, investigating methods of data processing supported by the employment of classification algorithms and data fusion strategies. As a result, the systematic mapping process initially found 7515 articles from six academic databases in the period from 2009 to 2019. After the execution of filtering methods, a complete read of 67 articles allowed to answer quantitatively and qualitatively the research questions. The classification of the answers contributed to systematize the literature through the taxonomy and the relationships among the topics covered by the articles. Accordingly, this research produced theoretical benefits, mainly, a broad view of the state-of-the-art, a taxonomy to guide related researches, and guidelines for future works. Furthermore, this research would benefit management and corporations by shedding light on technologies explored in the literature and elucidating their feasibility in support of the workforce’s safety, psychological, and physical health.
Monitoring gait quality in daily activities through wearable sensors has the potential to improve medical assessment in Parkinson’s Disease (PD). In this study, four gait partitioning methods, two based on thresholds and two based on a machine learning approach, considering the four-phase model, were compared. The methods were tested on 26 PD patients, both in OFF and ON levodopa conditions, and 11 healthy subjects, during walking tasks. All subjects were equipped with inertial sensors placed on feet. Force resistive sensors were used to assess reference time sequence of gait phases. Goodness Index (G) was evaluated to assess accuracy in gait phases estimation. A novel synthetic index called Gait Phase Quality Index (GPQI) was proposed for gait quality assessment. Results revealed optimum performance (G < 0.25) for three tested methods and good performance (0.25 < G < 0.70) for one threshold method. The GPQI resulted significantly higher in PD patients than in healthy subjects, showing a moderate correlation with clinical scales score. Furthermore, in patients with severe gait impairment, GPQI was found higher in OFF than in ON state. Our results unveil the possibility of monitoring gait quality in PD through real-time gait partitioning based on wearable sensors.
Objective:
The degree in which practitioners use the observational methods for musculoskeletal disorder risks assessment correctly was evaluated.
Background:
Ergonomics assessment is a key issue for the prevention and reduction of work-related musculoskeletal disorders in workplaces. Observational assessment methods appear to be better matched to the needs of practitioners than direct measurement methods, and for this reason, they are the most widely used techniques in real work situations. Despite the simplicity of observational methods, those responsible for assessing risks using these techniques should have some experience and know-how in order to be able to use them correctly.
Method:
We analyzed 442 risk assessments of actual jobs carried out by 290 professionals from 20 countries to determine their reliability.
Results:
The results show that approximately 30% of the assessments performed by practitioners had errors. In 13% of the assessments, the errors were severe and completely invalidated the results of the evaluation.
Conclusion:
Despite the simplicity of observational method, approximately 1 out of 3 assessments conducted by practitioners in actual work situations do not adequately evaluate the level of potential musculoskeletal disorder risks.
Application:
This study reveals a problem that suggests greater effort is needed to ensure that practitioners possess better knowledge of the techniques used to assess work-related musculoskeletal disorder risks and that laws and regulations should be stricter as regards qualifications and skills required by professionals.
This study aims to develop an innovative approach based on a wearable inertial system, which enables objective evaluations on the of loss of ground contact in race-walking, in order to assist coaching and judging. The architecture of the system, its positioning on the human body and functional requirements were defined through a Kansei Engineering approach by using a significant sample of athletes, coaches and judges within the race-walking environment. The analysis of variance supports decisions concerning the optimal system architecture consisting of an inertial sensor positioned on the centre-of-mass of the subject and a control unit. The selected device was then validated in laboratory conditions by means of an integrated system, including dynamic (680 Hz) and kinematic (340 Hz) devices, which are more accurate than the inertial system (200 Hz). The experiment was carried out at the Fraunhofer JL IDEAS-MISEF at CESMA, Laboratory of Advanced Measures on Ergonomics and Shapes of the University of Naples Federico II where four elite race walkers performed 60 test-runs according to a well-defined experimental protocol. Results proved that the inertial system could improve the accuracy in detecting illegal steps. Through statistical classification, it was found that the proposed approach has achieved encouraging results in comparison with state of the art approaches and could be a good architecture to develop a valuable tool to assist experts.
In the last years, gait phase partitioning has come to be a challenging research topic due to its impact on several applications related to gait technologies. A variety of sensors can be used to feed algorithms for gait phase partitioning, mainly classifiable as wearable or non-wearable. Among wearable sensors, footswitches or foot pressure insoles are generally considered as the gold standard; however, to overcome some inherent limitations of the former, inertial measurement units have become popular in recent decades. Valuable results have been achieved also though electromyography, electroneurography, and ultrasonic sensors. Non-wearable sensors, such as opto-electronic systems along with force platforms, remain the most accurate system to perform gait analysis in an indoor environment. In the present paper we identify, select, and categorize the available methodologies for gait phase detection, analyzing advantages and disadvantages of each solution. Finally, we comparatively examine the obtainable gait phase granularities, the usable computational methodologies and the optimal sensor placements on the targeted body segments.
The aim of this work is the evaluation of Distributed Classifier for the detection of gait phases that can be implemented in an active knee orthosis for the recovery of locomotion of pediatric subjects with neurological diseases, such as Cerebral Palsy (CP). The classifier is based on a Hierarchical Weighted Decision applied to the outputs of two or more scalar Hidden Markov Models (HMMs) trained by linear accelerations and angular velocities measured at shank and thigh. The kinematics of the dominant lower limb of ten healthy subjects were acquired by means of linear accelerometers and gyroscopes embedded in two inertial sensors. The actual sequence of gait phases was captured by means of foot switches. The experimental procedure consisted in one walking task, repeated for three times, on a treadmill at the preferred velocity of each subject. We compared the performance, in terms of sensitivity and specificity, of both Scalar Classifiers (SCs) and Distributed Classifiers (DCs) based on all the combinations of sagittal acceleration and sagittal angular velocity of the two body segments. The DC based on the angular velocities showed the highest values of sensitivity and specificity. The SC based on the angular velocity of shank was the better among others SCs, but the values of sensitivity and specificity are lower than 0.95. When we use only one sensor, placed on shank or thigh, the DC based on kinematic variables of shank showed better results, but not higher than 0.95. Consequently, the additional information provided by linear acceleration did not improve the performance and then, the gait-phase detection algorithm, which can be implemented in an active knee orthosis, has to be based on the output of two gyroscopes placed on shank and thigh.
Gait-phase recognition is a necessary functionality to drive robotic rehabilitation devices for lower limbs. Hidden Markov Models (HMMs) represent a viable solution, but they need subject-specific training, making data processing very time-consuming. Here, we validated an inter-subject procedure to avoid the intra-subject one in two, four and six gait-phase models in pediatric subjects. The inter-subject procedure consists in the identification of a standardized parameter set to adapt the model to measurements. We tested the inter-subject procedure both on scalar and distributed classifiers. Ten healthy children and ten hemiplegic children, each equipped with two Inertial Measurement Units placed on shank and foot, were recruited. The sagittal component of angular velocity was recorded by gyroscopes while subjects performed four walking trials on a treadmill. The goodness of classifiers was evaluated with the Receiver Operating Characteristic. The results provided a goodness from good to optimum for all examined classifiers (0 < G < 0.6), with the best performance for the distributed classifier in two-phase recognition (G = 0.02). Differences were found among gait partitioning models, while no differences were found between training procedures with the exception of the shank classifier. Our results raise the possibility of avoiding subject-specific training in HMM for gait-phase recognition and its implementation to control exoskeletons for the pediatric population.
The use of on-body wearable sensors is widespread in several academic and industrial domains. Of great interest are their applications in ambulatory monitoring and pervasive computing systems; here, some quantitative analysis of human motion and its automatic classification are the main computational tasks to be pursued. In this paper, we discuss how human physical activity can be classified using on-body accelerometers, with a major emphasis devoted to the computational algorithms employed for this purpose. In particular, we motivate our current interest for classifiers based on Hidden Markov Models (HMMs). An example is illustrated and discussed by analysing a dataset of accelerometer time series.
To study the influence of work related physical and psychosocial factors and individual characteristics on the occurrence of low back pain among young and pain free workers.
The Belgian Cohort Back Study was designed as a prospective cohort study. The study population of this paper consisted of 716 young healthcare or distribution workers without low back pain lasting seven or more consecutive days during the year before inclusion. The median age was 26 years with an interquartile range between 24 and 29 years. At baseline, these workers filled in a questionnaire with physical exposures, work related psychosocial factors and individual characteristics. One year later, the occurrence of low back pain lasting seven or more consecutive days and some of its characteristics were registered by means of a questionnaire. To assess the respective role of predictors at baseline on the occurrence of low back pain in the following year, Cox regression with a constant risk period for all subjects was applied.
After one year of follow up, 12.6% (95% CI 10.1 to 15.0) of the 716 workers had developed low back pain lasting seven or more consecutive days. An increased risk was observed for working with the trunk in a bent and twisted position for more than two hours a day (RR 2.2, 95% CI 1.2 to 4.1), inability to change posture regularly (RR 2.1, 95% CI 1.3 to 3.5), back complaints in the year before inclusion (RR 1.7, 95% CI 1.1 to 2.8), and high scores of pain related fear (RR 1.8, 95% CI 1.0 to 3.1). Work related psychosocial factors and physical factors during leisure time were not predictive.
This study highlighted the importance of physical work factors and revealed the importance of high scores of pain related fear in the development of low back pain among young workers.
My first exposure to Support Vector Machines came this spring when
heard Sue Dumais present impressive results on text categorization using
this analysis technique. This issue's collection of essays should help
familiarize our readers with this interesting new racehorse in the
Machine Learning stable. Bernhard Scholkopf, in an introductory
overview, points out that a particular advantage of SVMs over other
learning algorithms is that it can be analyzed theoretically using
concepts from computational learning theory, and at the same time can
achieve good performance when applied to real problems. Examples of
these real-world applications are provided by Sue Dumais, who describes
the aforementioned text-categorization problem, yielding the best
results to date on the Reuters collection, and Edgar Osuna, who presents
strong results on application to face detection. Our fourth author, John
Platt, gives us a practical guide and a new technique for implementing
the algorithm efficiently
Market competition searches continuous improvement to reduce production costs in order to profit increase. In this sense, there are different approaches to improve the production process. Specifically in shop floor, an expertise methodology must evaluates work stations considering every operation tasks and determines the difficult levels [1].
Thesupport-vector network is a new learning machine for two-group classification problems. The machine conceptually implements the following idea: input vectors are non-linearly mapped to a very high-dimension feature space. In this feature space a linear decision surface is constructed. Special properties of the decision surface ensures high generalization ability of the learning machine. The idea behind the support-vector network was previously implemented for the restricted case where the training data can be separated without errors. We here extend this result to non-separable training data.High generalization ability of support-vector networks utilizing polynomial input transformations is demonstrated. We also compare the performance of the support-vector network to various classical learning algorithms that all took part in a benchmark study of Optical Character Recognition.
In today’s race walking competition, the determination of whether an athlete fouls is mainly affected by a referee’s subjective judgment, leading to a high possibility of misjudgment. The purpose of this work is to determine whether race walking can be automatically recognized by accelerometers embedded in smartphones. In this work, acceleration data are collected by a smartphone app developed by ourselves. Nineteen features are extracted from the raw sensor data, and are used by an unsupervised classification method for activity recognition, named MCODE. We evaluate various data sampling rates and window lengths during feature extraction in the experiments. We also compare our method with other well-known methods on the metrics such as sensitivity, specificity and adjusted rank index. The results show that our method is viable to recognize race walking using smartphone accelerometers.
This article presents a comparative analysis of easy-to-use methods for assessing musculoskeletal load and the risk for developing musculoskeletal disorders. In all such methods, assessment of load consists in defining input data, the procedure and the system of assessment. This article shows what assessment steps the methods have in common; it also shows how those methods differ in each step. In addition, the methods are grouped according to their characteristic features. The conclusion is that the concepts of assessing risk in different methods can be used to develop solutions leading to a comprehensive method appropriate for all work tasks and all parts of the body. However, studies are necessary to verify the accepted premises and to introduce some standardization that would make consolidation possible.
In the light of data and speculation contained in the literature, and based on procedures illustrated in a previous research project in which the author described and evaluated occupational risk factors associated with work-related musculoskeletal disorders of the upper limbs (WMSDs), this paper proposes a method for calculating a concise index of exposure to repetitive movements of the upper limbs. The proposal, which still has to be substantiated and validated by further studies and applications, is conceptually based on the procedure recommended by the NIOSH for calculating the Lifting Index in manual load handling activities. The concise exposure index (OCRA index) in this case is based on the relationship between the daily number of actions actually performed by the upper limbs in repetitive tasks, and the corresponding number of recommended actions. The latter are calculated on the basis of a constant (30 actions per minute), which represents the action frequency factor; it is valid--hypothetically--under so-called optimal conditions; the constant is diminished case by case (using appropriate factors) as a function of the presence and characteristics of the other risk factors (force, posture, additional elements, recovery periods). Although still experimental, the exposure index can be used to obtain an integrated and concise assessment of the various risk factors analysed and to classify occupational scenarios featuring significant and diversified exposure to such risk factors.
The objective of this study was to analyze associations of three indicators of perceived work stress (physical job demand, low control at work, and an imbalance between effort and reward), and of overcommitment, a personal pattern of coping with work demands, with musculoskeletal pain. A standardized questionnaire measuring these conditions in addition to self-reported musculoskeletal pain at different locations was administered to a group of 316 male and female employees of a public transport enterprise. After we adjusted for confounding effects of age, sex, socioeconomic status, shift work, and negative affectivity, we observed elevated prevalence odds ratios in employees who scored high on overcommitment, who were exposed to physical job demand, and, to a lesser extent, who reported psychosocial work stress. Results have implications for a more comprehensive approach to primary and secondary prevention of musculoskeletal pain.
Support vector machines
- M A Hearst
- S T Dumais
- E Osuna
- J Platt
- B Scholkopf
M. A. Hearst, S. T. Dumais, E. Osuna, J. Platt, and B.
Scholkopf, "Support vector machines," IEEE Intell.
Syst. their Appl., vol. 13, no. 4, pp. 18-28, Jul. 1998.