Digital ergonomics and AI-based evaluation in agricultural work systems

Furthermore, dependence on observer judgment and worker selfreporting often results in underreporting of ergonomic hazards and limits the early identification of injury risks, thereby reducing the effectiveness of preventive interventions in real agricultural environments (Singh et al., 2024; Yousefi et al., 2026) [50,56] . The emergence of Agriculture 4.0 has accelerated the adoption of digital technologies that address many of these limitations. Digital ergonomics combines wearable sensing technologies, computer vision, artificial intelligence (AI), Internet of Things (IoT) platforms, and Digital Human Modeling (DHM) to enable continuous monitoring of workers, automated posture recognition, and predictive ergonomic risk assessment. Wearable devices such as inertial measurement units (IMUs), biometric sensors, and smart glasses provide objective biomechanical and physiological measurements, while AI-based computer vision systems automate posture analysis using skeletal tracking and deep learning algorithms. Simultaneously, Digital Human Modeling and Computer-Aided Design (CAD) facilitate virtual simulation of human-machine interactions during equipment design, allowing ergonomic deficiencies to be identified before physical prototypes are developed (Apicella & Tarabella, 2025; Patel et al., 2016) [5,38] . Against this background, this review synthesizes recent advances in digital ergonomics and AI-based ergonomic evaluation within agricultural work systems. Emphasis is placed on digital sensing technologies, wearable devices, computer vision, machine learning, digital twins, and human-centered design approaches that support continuous ergonomic monitoring across both manual agricultural activities, such as harvesting and pruning, and mechanized operations including tractor driving and machinery operations. The review further examines emerging applications, current challenges, and future research directions that support the transition from conventional reactive safety management toward predictive, data-driven occupational health systems.

Digital Ergonomics

Digital ergonomics, frequently referred to as "cybergonomics", represents the integration of classical ergonomic principles with emerging technologies associated with Industry 4.0 and Agriculture 4.0, including artificial intelligence, robotics, IoT, wearable sensing, and digital simulation. Rather than relying solely on observational assessments performed after ergonomic problems occur, digital ergonomics continuously acquires, transmits, and analyzes physiological, biomechanical, and environmental data to support real-time monitoring and predictive decision-making. In this framework, agricultural workers become active components of smart farming ecosystems through continuous interaction with digital technologies (Apicella & Tarabella, 2025; Benos et al., 2020) [5,8] . A central component of digital ergonomics is Ergonomic Posture Risk Assessment (EPRA), which aims to identify potentially hazardous working postures before musculoskeletal injuries develop. Compared with traditional observation-based EPRA methods, computer vision-based approaches provide automated posture evaluation by continuously extracting biomechanical information from workers during actual field operations, thereby reducing observer bias and improving assessment consistency (Figure 1). By integrating these complementary sources of information, digital ergonomics enables a transition from retrospective injury evaluation toward proactive identification of ergonomic risks, allowing preventive interventions before occupational injuries occur.

Digital Tools in Ergonomic Assessment

Recent advances in digital technologies have fundamentally transformed ergonomic assessment from subjective observational methods into objective, data-driven monitoring systems capable of continuous evaluation. Wearable sensing technologies constitute one of the most widely adopted approaches for measuring workers' biomechanical and physiological responses during agricultural activities. Inertial Measurement Units (IMUs), surface electromyography (sEMG), accelerometers, and gyroscopes provide quantitative measurements of joint kinematics, body orientation, movement dynamics, and muscular activation, thereby enabling detailed assessment of physical workload (Kapse et [27,30,53] .

Surface electromyography is particularly valuable for evaluating muscular effort during agricultural operations. Sensors positioned on muscles including the brachioradialis (BR), extensor digitorum (ED), flexor carpi radialis (FCR), and middle deltoid (MD) provide continuous measurements of muscle activation patterns associated with repetitive work tasks (Fig. 2) (Chouriya, 2025) [15] . Complementary physiological monitoring devices, such as Polar sensors, record heart rate and heart rate variability, enabling objective assessment of cardiovascular workload, fatigue, and worker stress responses throughout task performance (Fig. 3) (Chouriya, 2025) [15] . Computer vision and artificial intelligence have further expanded the capabilities of ergonomic assessment by automating conventional observational methods. Deep learning algorithms extract skeletal key points from RGB cameras or depth sensors such as Kinect V2, enabling automatic calculation of ergonomic indices including RULA and REBA without direct human observation (Singh et al [50,53,55,56] . [5,30,46,50] .

Role of Artificial Intelligence in Ergonomic Evaluation

Artificial Intelligence (AI) has become a core enabling technology within digital ergonomics by transforming conventional ergonomic assessment from a reactive, observation-based process into a predictive, data-driven system. By processing large volumes of physiological, biomechanical, and visual data collected through wearable sensors and computer vision systems, AI supports continuous monitoring of workers, automated posture recognition, and early identification of ergonomic risks. Within Agriculture 4.0, AI extends digitalization to the human component of farming systems, integrating workers into intelligent production environments where occupational health can be continuously assessed and optimized (Apicella & Tarabella, 2025; Donisi et al., 2022) [5,19] . [37,55] .

Machine Learning for Posture Analysis

Machine learning (ML) forms the computational foundation of modern AI-based ergonomic evaluation by automating posture classification and ergonomic risk assessment. Traditional observational techniques require trained assessors to evaluate worker postures manually, making the process labor-intensive and susceptible to subjective variation. Machine learning algorithms overcome these limitations by learning relationships between biomechanical features and ergonomic risk categories from sensor and image datasets, enabling rapid and objective posture evaluation (Kapse et al., 2024; Yang et al., 2024) [30,55] . Several conventional machine learning algorithms have been successfully applied for ergonomic posture classification, including Support Vector Machines (SVM), Artificial Neural Networks (ANN), Random Forest (RF), and K-Nearest Neighbors (KNN). These algorithms classify worker postures into predefined ergonomic risk levels using features extracted from images, wearable sensors, or motion capture systems. By evaluating biomechanical parameters such as joint angles and body orientation, these models distinguish safe postures from those associated with elevated musculoskeletal risk.

The increasing availability of large visual datasets has accelerated the adoption of deep learning approaches that eliminate the need for manual feature engineering. Lightweight pose estimation models such as MoveNet automatically detect anatomical key points and classify trunk postures into categories including neutral posture and forward bending. Applications in agricultural environments, particularly nursery operations, have demonstrated posture classification accuracies approaching 87.8%, highlighting the potential of deep learning for reliable ergonomic monitoring under practical working conditions (Kapse et al., 2024) [30] . Beyond static posture recognition, temporal machine learning models have expanded ergonomic assessment to include dynamic movement analysis. Recurrent Neural Networks (RNNs), particularly Long Short-Term Memory (LSTM) architectures, analyze sequential motion and wearable sensor data to recognize repetitive agricultural activities such as manual pruning and crate transportation. By incorporating temporal information, these models capture cumulative movement patterns and repetitive workload characteristics that cannot be identified through isolated posture observations, thereby enabling predictive assessment of ergonomic risks throughout the working day (Apicella & Tarabella, 2025) [5] .

The general architecture of a computer vision-and machine learning-based Ergonomic Posture Risk Assessment (EPRA) system framework consists of three sequential stages: data preparation, pose estimation, and ergonomic risk assessment (Fig 4 [34,55] . An important advantage of computer vision over wearable technologies is its capability to monitor multiple workers simultaneously. AI-based image analysis systems perform unsupervised posture assessment across several individuals within the same working environment while maintaining continuous observation under varying illumination conditions and partial visual occlusions. Such capability is particularly valuable for large-scale agricultural operations where simultaneous monitoring of multiple workers is required (Yang et al., 2024; MassirisFernández et al., 2020) [34,55] . Computer vision has also been incorporated into mechanized agricultural operations to evaluate human-machine interactions.

Combined with motion capture technologies, these systems analyze the influence of tractor dashboard configurations, control locations, and display positioning on operator posture and muscular activity. Such analyses provide valuable information for optimizing equipment design while minimizing operator discomfort and long-term musculoskeletal loading (Kapse et al., 2024) [30] .

AI-Based Ergonomic Risk Prediction

Beyond posture recognition, artificial intelligence enables predictive ergonomic assessment by identifying complex relationships among movement patterns, workload characteristics, and physiological responses that precede the development of work-related musculoskeletal disorders. Rather than evaluating isolated postures, AI analyzes continuous streams of biomechanical and physiological data to estimate injury risk before clinical symptoms become evident (Naranjo et al., 2025) [37] . AI-driven ergonomic assessment systems automate the calculation of established ergonomic indices, including Rapid Upper Limb Assessment (RULA) and Rapid Entire Body Assessment (REBA), by converting skeletal information extracted from images or motion sensors into standardized ergonomic risk scores. This automation substantially reduces observer bias, increases assessment consistency, and shortens the time required for ergonomic evaluation compared with conventional manual methods (Yang et al., 2024) [55] . Artificial intelligence further improves biomechanical analysis through sensor fusion approaches that combine measurements obtained from inertial measurement units (IMUs), wearable motion sensors, and pressure insoles. Integration of multiple sensing modalities enables accurate estimation of spinal loading and lower back injury risk during manual material handling activities by capturing dynamic forces and postural loads that are difficult to quantify using observational techniques alone (Apicella & Tarabella, 2025; Benos et al., 2020; Hilmi et al., 2024) [5,8,27] . Recent developments increasingly employ hybrid deep learning architectures, particularly CNN-LSTM models, to analyze temporal changes in posture throughout complete work cycles. By incorporating both spatial posture information and temporal movement sequences, these models account for repetitive motions, cumulative fatigue, and changing workload conditions over time. Consequently, AI-based ergonomic systems progress beyond static posture evaluation toward comprehensive predictive risk management capable of supporting timely interventions and reducing the incidence of occupational musculoskeletal disorders (Yang et al., 2024) [55] .

Agricultural Work Systems

Agricultural work systems encompass a broad range of manual and mechanized operations that expose workers to substantial biomechanical, physiological, and environmental stresses. Although mechanization has reduced the physical demands of certain farming activities, many agricultural tasks continue to require repetitive movements, awkward body postures, heavy manual material handling, and prolonged exposure to adverse environmental conditions. These occupational demands contribute significantly to the high prevalence of work-related musculoskeletal disorders (WMSDs), making agriculture one of the most ergonomically challenging occupational sectors (Benos et al., 2020; Das, 2023) [8,18] .

Physical Workloads

Manual agricultural activities frequently require prolonged trunk flexion, kneeling, squatting, and repetitive bending, particularly during the cultivation, harvesting, and maintenance of lowgrowing crops such as vegetables and strawberries. These sustained awkward postures substantially increase mechanical loading on the lumbar spine, knees, and lower extremities, thereby contributing to chronic low back pain, knee osteoarthritis, and other musculoskeletal disorders (Benos et al., 2020; Das, 2023) [8,18] . The repetitive nature of agricultural work further accelerates muscle fatigue and cumulative tissue damage. Continuous repetition of similar movements without sufficient recovery increases biomechanical stress on joints and soft tissues, leading to progressive deterioration of musculoskeletal health. These risks are particularly pronounced among aging agricultural workers, whose reduced physical capacity makes them more susceptible to fatigue and injury under demanding workload conditions (Benos et al., 2020) [8] . Figure 5 illustrates representative agricultural operations characterized by repetitive manual handling, sustained bending, lifting, and prolonged awkward working postures. These activities collectively demonstrate the considerable physical demands associated with agricultural labor and highlight the importance of ergonomic interventions for reducing occupational injury risk.

Manual Material Handling and Tool Use

Manual material handling (MMH) remains one of the principal contributors to musculoskeletal injury in agriculture. Agricultural workers routinely lift, carry, transport, and manipulate loads that frequently exceed 22.5 kg (50 lb) during harvesting, transportation, and storage operations. Repeated lifting of heavy loads imposes substantial mechanical stress on the spine, shoulders, wrists, and upper limbs, particularly when tasks are performed under awkward or asymmetrical body postures (Benos et al., 2020; Fathallah, 2010; Sharma et al., 2025; Kee, 2022) [8,21,31,49] . The ergonomic burden associated with manual material handling is further intensified by the widespread use of traditional agricultural hand tools. Conventional implements such as sickles, pruning shears, and cutting tools often require excessive gripping force, repetitive wrist movements, and repeated elbow extension, thereby increasing cumulative loading on the upper extremities. Additionally, poor compatibility between tool dimensions and workers' anthropometric characteristics results in inefficient working postures, greater physical exertion, and elevated injury risk (Benos et al., 2020; Das, 2023) [8,18] These findings emphasize the need for ergonomically designed hand tools and improved manual handling strategies that minimize excessive physical loading while enhancing worker comfort and operational efficiency.

Tractor and Machinery Operation

Although agricultural mechanization has substantially improved farm productivity, it has also introduced new categories of ergonomic hazards. Tractor operation, in particular, has been consistently identified as a major contributor to musculoskeletal disorders because operators are frequently required to maintain constrained sitting postures while simultaneously performing multiple visual and manual tasks (Benos et al., 2020; Fathallah, 2010) [8,21] . During routine field operations, tractor operators repeatedly rotate the neck and trunk to monitor rear-mounted implements while simultaneously manipulating steering controls, transmission systems, and dashboard interfaces. These sustained twisting postures increase loading on the cervical spine, shoulders, and lower back, particularly during prolonged working periods (Benos et al., 2020; Grandi et al., 2022) [8,23] . Ergonomic deficiencies in cabin design further contribute to operator discomfort. Poorly positioned control panels, limited adjustability of seats and dashboards, and intrusive cabin layouts restrict natural body movement and promote awkward working postures. Similarly, digital displays incorporated into semiautomated agricultural machinery may adversely affect posture when viewing angles require excessive head rotation or prolonged neck extension, leading to increased muscular activation and long-term musculoskeletal strain (Grandi et al., 2022; Vitale et al., 2024) [23,53] . These observations demonstrate the importance of incorporating ergonomic principles into the design of agricultural machinery to improve operator comfort, reduce physical workload, and minimize long-term occupational health risks.

Environmental Exposures

Agricultural workers are routinely exposed to multiple environmental hazards that adversely affect both physical health and ergonomic performance. Among these, prolonged exposure to elevated temperatures and solar radiation is particularly significant, frequently resulting in dehydration, heat exhaustion, heat stroke, and cumulative physiological stress during intensive field operations, especially under summer conditions (Kee, 2022; Petrosyan et al., 2020; Ross et al., 2016) [31,40,44] . Mechanized agricultural operations introduce additional physical stressors through continuous exposure to whole-body vibration (WBV) generated by tractor engines and uneven terrain. Longterm WBV exposure has been strongly associated with chronic lower back pain, spinal degeneration, and reduced operator comfort. Likewise, prolonged use of handheld powered equipment exposes workers to hand-arm transmitted vibration (HATV), increasing the likelihood of vibration-induced white finger syndrome and neuromuscular impairments affecting the upper extremities (Benos et al., 2020) [8] . The agricultural working environment may also present additional occupational hazards associated with inadequate workplace conditions. Poor ventilation and insufficient illumination within agricultural buildings, including curing barns and enclosed farm structures, increase the likelihood of slips, falls, and traumatic injuries. Furthermore, tractor-mounted implements and agricultural machinery frequently generate noise levels exceeding 85 dB(A), exposing workers to hearing impairment, fatigue, irritability, and increased cardiovascular stress during prolonged operation (Schuman, 2002; Martins et al., 2014; Mushtaq et al., 2025) [33,36,48] . Collectively, these physical, mechanical, and environmental stressors illustrate the multifactorial nature of ergonomic risks within agricultural work systems. Effective occupational health management therefore requires integrated assessment approaches that simultaneously consider biomechanical loading, environmental exposures, and worker physiology to support comprehensive ergonomic risk reduction.

Digital Technologies Used in Agricultural Ergonomics

The advancement of Agriculture 4.0 has accelerated the integration of digital technologies into ergonomic assessment, enabling continuous monitoring of workers and proactive management of occupational health risks. Unlike conventional observational methods, digital ergonomic technologies collect and analyze biomechanical, physiological, and environmental information in real time, allowing ergonomic interventions to be implemented before injuries occur. The principal technologies include wearable sensors, computer vision systems, motion capture technologies, digital twins, virtual prototyping, robotics, and human-robot collaboration, all of which contribute to datadriven ergonomic management within modern agricultural systems (Apicella & Tarabella, 2025) [5] .

Wearable Sensors and IoT-Based Monitoring

Wearable sensing technologies constitute one of the most mature and widely implemented components of digital ergonomics. By integrating workers into Internet of Things (IoT)-based monitoring systems, wearable devices continuously acquire physiological and biomechanical information, enabling real-time assessment of physical workload, posture, fatigue, and environmental exposure. This approach transforms agricultural workers into active data sources within smart farming ecosystems, thereby facilitating predictive occupational health management (Apicella & Tarabella, 2025; Donisi et al., 2022) [5,19] . A wide range of wearable technologies has been adopted for agricultural ergonomic assessment. Inertial Measurement Units (IMUs), accelerometers, gyroscopes, and biopotential sensors such as surface electromyography (sEMG) quantify body movements, joint orientations, angular velocity, and muscular activity during agricultural operations. These devices provide objective measurements of worker posture and physical workload while supporting continuous evaluation of musculoskeletal stress under field conditions. Wearable systems are also employed for physiological monitoring. Biometric bands and sensor-integrated garments continuously record heart rate, body temperature, heat stress, and other physiological responses associated with physical exertion.

Environmental sensors complement these measurements by monitoring workplace conditions, including exposure to hazardous gases such as ammonia, thereby enabling comprehensive evaluation of occupational health risks. The information acquired by wearable devices is transmitted through IoT-enabled communication networks to cloud-based monitoring platforms, where machine learning algorithms analyze sensor streams to classify worker activities, detect abnormal movements, and identify potential safety risks. For example, SHIMMER3-based wearable systems have demonstrated the ability to distinguish accidental falls from routine agricultural activities, while integrated monitoring dashboards provide immediate visual or auditory alerts whenever hazardous conditions are detected (Apicella & Tarabella, 2025; Donisi et al., 2022) [5,19] . Table 1 summarizes representative wearable technologies used in agricultural ergonomics, together with their monitored parameters, technological components, application domains, and functional outputs. These examples demonstrate the diversity of wearable solutions currently available for improving worker safety, health monitoring, and ergonomic evaluation.

Computer Vision and Image-Based Posture Assessment

Computer vision (CV) has become an increasingly important technology for ergonomic assessment because it enables objective posture evaluation without requiring workers to wear body-mounted sensors. This noninvasive approach is particularly advantageous in agricultural environments, where wearable devices may interfere with manual operations or become uncomfortable during prolonged outdoor work (MassirisFernández et al., 2020) [34] . Modern computer vision systems employ deep learning algorithms, particularly Convolutional Neural Networks (CNNs), to estimate human body posture directly from digital images or video sequences. By automatically identifying anatomical landmarks and constructing digital skeletal models, these systems provide quantitative measurements of body posture that support objective ergonomic evaluation. Among the available pose estimation models, MoveNet has demonstrated considerable effectiveness in agricultural applications. The model extracts anatomical key points and classifies trunk posture into neutral, slight forward bending, and full forward bending categories, achieving posture classification accuracies of approximately 87.8% when combined with Artificial Neural Networks (ANNs). Such performance illustrates the capability of AI-assisted computer vision to identify potentially harmful working postures during actual agricultural operations (Kapse et al., 2024) [30] . [12] Lumbar support exoskeleton -(no direct measurement) IMU, EMG Agriculture -Manual material handling

Lower-back load reduction and ergonomic support Chittar and Barve, 2022 [14] Printable sensor Ammonia concentration in the air 2020 [41] Sensorized jacket for the chest Height, trunk inclination, acceleration, magnetic field SHIMMER3, MLP Logistics -Lifting and handling

Lifting activity recognition and biomechanical risk assessment

Pistolesi and Lazzerini, 2020 [42] Computer vision systems also automate traditional ergonomic assessment methods by calculating joint angles from skeletal models and converting them into standardized ergonomic indices such as Rapid Upper Limb Assessment (RULA) and Rapid Entire Body Assessment (REBA). Unlike manual observational techniques, AI-based image analysis enables continuous monitoring of multiple workers simultaneously under practical field conditions, including environments characterized by variable illumination, partial obstruction, and unstable camera viewpoints. Consequently, computer vision provides an efficient, scalable, and objective alternative for ergonomic risk assessment within agricultural work systems. Figure 6 presents representative trunk posture classifications generated using the MoveNet-based posture estimation framework (Kapse et al., 2024; Yang et al., 2024) [30,55] .

Motion Capture Systems

Motion capture (MoCap) technologies provide highly accurate digital representations of human movement and are regarded as one of the most objective methods for ergonomic assessment. Unlike observational approaches, motion capture systems quantify body kinematics directly, enabling comprehensive biomechanical analysis of complex agricultural tasks (Rybnikár et al., 2023) [45] . Motion capture technologies are broadly categorized into marker-based and marker-less systems. Marker-based systems employ reflective markers attached to the worker's body, whereas marker-less approaches utilize intelligent camera systems to estimate body posture without physical markers. Both approaches provide detailed measurements of body segment orientation, joint motion, and movement frequency during agricultural operations (Salisu et al., 2023; Yang et al., 2024) [47,55] . Wearable inertial motion capture systems, including Xsens, integrate multiple inertial sensors distributed across different body segments to monitor orientation, acceleration, and movement at sampling frequencies approaching 120 Hz. Such systems have been applied to agricultural activities including banana harvesting and grapevine pruning, allowing researchers to quantify repetitive movements, joint amplitudes, and extreme working postures associated with elevated musculoskeletal risk (Merino et al., 2019) [35] . By converting human movement into digital datasets that can be stored, replayed, and analyzed quantitatively, motion capture systems facilitate identification of critical joint angles and repetitive movement patterns associated with occupational injury. This capability provides valuable support for ergonomic intervention, equipment redesign, and worker training (Rybnikár et al., 2023) [45] .

Digital Twins and Virtual Prototypes for Farm Work Design

Digital Twin (DT) technology enables the creation of virtual representations of physical systems that continuously simulate real-world behavior. Within agricultural ergonomics, Digital Twins provide a powerful platform for evaluating equipment design, work processes, and human-machine interactions before physical implementation, thereby reducing development costs while improving worker safety (Slob et al., 2023; Peladarinos et al., 2023) [39,51] . Digital Human Modeling (DHM) (Fig. 7) integrated with Computer-Aided Design (CAD) allows ergonomic principles to be incorporated during the earliest stages of agricultural equipment development. Virtual human manikins, including the Jack manikin, simulate manual agricultural operations within three-dimensional environments, enabling designers to evaluate visibility, reachability, working posture, and ergonomic indices such as EAWS before constructing physical prototypes (Caputo et al., 2019; Naranjo et al., 2025) [10,37] . These technologies have proven particularly valuable for optimizing tractor cabins and operator workstations. Virtual simulations evaluate the accessibility of controls, operator visibility, and overall driver comfort across workers with different anthropometric characteristics, thereby facilitating inclusive and human-centered equipment design (Patel et al., 2016; Salisu et al., 2023) [38,47] . Beyond machinery design, Digital Twin technology has been extended to greenhouse horticulture, where virtual environments enable farmers to visualize production systems, monitor crop development remotely, and evaluate operational strategies through Virtual Reality (VR). These digital environments provide opportunities to identify ergonomic and operational inefficiencies within risk-free simulation platforms before implementation under actual field conditions (Slob et al., 2023; Peladarinos et al., 2023) [39,51] .

Robotics and Automation as Ergonomic Interventions

Robotics and automation represent an important strategy for reducing worker's exposure to physically demanding agricultural tasks. Rather than replacing human labor entirely, modern robotic systems increasingly function as collaborative technologies that complement workers during activities associated with repetitive movements, heavy lifting, and awkward postures (Adamides & Edan, 2023) [1] . Collaborative robots (cobots) operate alongside agricultural workers within shared workspaces, assuming responsibility for physically intensive activities such as transporting harvested produce and carrying heavy crates. By reducing repetitive lifting and prolonged bending, cobots substantially decrease physical workload while lowering the risk of work-related musculoskeletal disorders (Adamides & Edan, 2023) [1] . Exoskeletons have emerged as another promising ergonomic intervention. Both passive and active exoskeleton systems provide mechanical assistance during lifting, overhead work, and manual material handling, thereby reducing muscle activation and perceived physical exertion in the lower back and shoulders. These wearable assistive technologies have demonstrated considerable potential for minimizing fatigue and preventing musculoskeletal injury during physically demanding agricultural operations (Naranjo et al., 2025; Hilmi et al., 2024) [27, 37] . Automation has also reduced ergonomic risks associated with repetitive manual operations. GPS-guided automated weeding systems eliminate prolonged stooping during weed removal, while advanced human-robot collaboration systems employ intelligent control algorithms capable of predicting hazardous situations and adapting robot behavior accordingly. Complementary wearable haptic interfaces further improve worker safety by providing immediate vibration-based feedback whenever unfavorable postures are detected (Fathallah, 2010; Hilmi et al., 2024) [21,27] .

AI-Based Ergonomic Evaluation Methods

Artificial intelligence has fundamentally transformed ergonomic assessment by enabling automated, objective, and predictive evaluation of occupational risks. Unlike conventional observation-based methods, AI-driven ergonomic systems integrate machine learning, deep learning, wearable sensing, computer vision, and cloud computing to continuously analyze worker posture, movement, and physiological responses. These approaches improve assessment accuracy, reduce observer bias, and facilitate timely ergonomic interventions that minimize the occurrence of work-related musculoskeletal disorders (WMSDs) (Naranjo et al., 2025; Yang et al., 2024) [37,55] .

Machine Learning Models for Ergonomic Risk Classification

Machine learning (ML) algorithms have become central to the automatic classification of ergonomic risks by replacing subjective observational assessments with objective, data-driven decision-making. These algorithms learn relationships between biomechanical variables and ergonomic risk levels, allowing continuous classification of worker posture with high precision across diverse agricultural tasks (Kapse et al., 2024) [30] . Several supervised learning algorithms have demonstrated strong performance in ergonomic risk assessment, including Artificial Neural Networks (ANN), Support Vector Machines (SVM), Random Forests (RF), and K-Nearest Neighbors (KNN). These models classify working postures into standardized ergonomic risk categories using features extracted from wearable sensors, computer vision systems, or motion capture technologies. Their ability to process large datasets enables rapid identification of hazardous working conditions while substantially reducing observer variability associated with conventional ergonomic assessment methods. Among the reported applications, a posture assessment framework integrating MoveNet with Artificial Neural Networks achieved a posture classification accuracy of approximately 87.8% during outdoor nursery operations. Similarly, ANN-based monitoring systems used in industrial assembly processes reported approximately 95% accuracy for detecting defective operational conditions through vibration and force analysis. Random Forest algorithms have also demonstrated superior performance in classifying biomechanical risks according to the Revised NIOSH Lifting Equation, frequently achieving classification accuracies exceeding 90% (Donisi et al., 2022; Naranjo et al., 2025) [19,37] . Machine learning applications have further expanded to portable ergonomic assessment systems. Smartphone based solutions mounted on the upper arm utilize embedded sensors and machine learning algorithms to classify ergonomic risks during occupational activities with reported accuracy of approximately 90.2%. In addition, Probabilistic Neural Networks (PNNs) have been employed to evaluate dynamic changes in biomechanical workload throughout complete work cycles, thereby extending ergonomic assessment beyond static posture evaluation toward continuous monitoring of occupational risk (Yang et al., 2024; Naranjo et al., 2025) [37,55] .

Deep Learning for Human Pose Estimation

Deep learning has substantially advanced ergonomic posture analysis by enabling automatic extraction of complex spatial features from images and video sequences without manual feature engineering. Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) have become the dominant architectures for human pose estimation because of their ability to identify anatomical landmarks accurately while accommodating the variability associated with agricultural work environments (Yang et al., 2024) [55] . State-of-the-art pose estimation frameworks, including OpenPose and MoveNet, identify two-dimensional or threedimensional skeletal key points directly from digital images or video streams. These algorithms generate digital skeletal models from which joint angles, body orientation, and movement characteristics can be computed automatically, thereby facilitating objective ergonomic assessment without requiring specialized motion capture equipment (Kapse et al., 2024; MassirisFernández et al., 2020) [34] . Among these approaches, MoveNet has attracted particular attention because of its lightweight architecture and computational efficiency, making it suitable for deployment on resource constrained devices commonly used in agricultural environments. As illustrated in Figure 8, the model identifies anatomical landmarks represented by key points and connects them to generate a complete skeletal representation of worker posture. Visible body segments are represented by yellow lines, whereas predicted but non-visible body segments are indicated by purple lines based on learned spatial relationships among anatomical landmarks (Kapse et al., 2024) [30] . [34,55] .

AI-Enabled Decision Support Systems

Artificial intelligence has expanded the scope of ergonomic assessment beyond posture recognition toward intelligent decision support capable of assisting both equipment designers and farm managers. AI-enabled Decision Support Systems (DSS) integrate biomechanical analysis with predictive analytics to recommend ergonomic improvements before hazardous conditions develop, thereby supporting proactive occupational health management (Grandi et al., 2022) [23] . One important application involves fuzzy logic-based decision support systems, which convert three-dimensional joint angle measurements into continuous ergonomic risk levels using rulebased inference rather than discrete scoring systems. Compared with conventional integer-based ergonomic indices, fuzzy logic provides smoother transitions between risk categories and more realistic representation of gradual posture changes encountered during agricultural work (Yang et al., 2024; Hilmi et al., 2024) [27, 55] . Artificial intelligence also strengthens Digital Human Modeling (DHM) by integrating predictive analysis into virtual design environments. Within these systems, designers simulate agricultural tasks using virtual prototypes and evaluate the physical demands associated with different workstation layouts, machinery configurations, and equipment dimensions before physical implementation. Such capabilities facilitate early identification and correction of ergonomic deficiencies, reducing development costs while improving worker safety (Naranjo et al., 2025; Caputo et al., 2019) [10,37] . Beyond equipment design, AI-driven decision support systems contribute to predictive occupational health management by identifying movement patterns associated with elevated injury risk. Within collaborative agricultural environments, AI algorithms optimize the movement of collaborative robots (cobots) to reduce worker exposure to hazardous tasks while simultaneously providing workers with immediate visual or haptic feedback whenever unsafe postures are detected. These capabilities facilitate continuous behavioral correction and proactive ergonomic intervention during routine agricultural operations (Apicella & Tarabella, 2025; Naranjo et al., 2025) [5,37] .

Integration of Artificial Intelligence with Wearables and IoT Platforms

The integration of artificial intelligence with wearable sensing technologies and Internet of Things (IoT) platforms has created comprehensive digital ecosystems capable of continuous occupational health monitoring. Within these systems, wearable devices collect biomechanical and physiological measurements that are transmitted to cloud-based analytical platforms, where AI algorithms perform predictive ergonomic analysis in real time (Naranjo et al., 2025) [37] . Wearable devices such as Inertial Measurement Units (IMUs), pressure insoles, and physiological sensors continuously monitor worker posture, movement, and physical workload. Industrial IoT gateways provide communication between field devices and cloud-based analytical services, enabling centralized storage, processing, and visualization of ergonomic data. AI models operating within these cloud environments generate individualized ergonomic assessments while identifying patterns associated with developing musculoskeletal disorders (Apicella & Tarabella, 2025; Hilmi et al., 2024; Yang et al., 2024) [5,27,55] . The combination of AI, IoT, and Digital Twin technology further enhances ergonomic management by integrating realtime wearable sensor data into virtual representations of workers and agricultural systems. These Digital Twins provide personalized recommendations and early warning notifications whenever physiological or biomechanical indicators suggest increasing occupational risk (Peladarinos et al., 2023) [39] . Recent developments have also incorporated smartphone-based ergonomic assessment systems that combine high-definition cameras with cloud-based AI models to provide immediate posture evaluation and corrective feedback. The adoption of high-speed communication technologies, particularly 5G, supports low-latency transmission of large sensor datasets, ensuring stable real-time operation of integrated ergonomic monitoring systems across modern agricultural environments (Yang et al., 2024) [55] .

Applications of Digital Ergonomics in Agricultural Tasks

The practical implementation of digital ergonomics has demonstrated considerable potential for reducing occupational risks across a wide range of agricultural operations. By integrating wearable sensing technologies, artificial intelligence (AI), computer vision, Digital Human Modeling (DHM), robotics, and Human-Robot Collaboration (HRC), digital ergonomic systems enable continuous monitoring of workers while reducing physical workload during labor-intensive agricultural tasks. Current applications are particularly prominent in harvesting, pruning, planting, and weeding operations, where repetitive movements, awkward postures, and manual material handling represent major ergonomic challenges (Apicella & Tarabella, 2025; Benos et al., 2020) [5,8] .

Harvesting and Picking Tasks

Harvesting is widely recognized as one of the most physically demanding agricultural operations because it requires repetitive reaching, prolonged bending, overhead work, heavy manual handling, and sustained awkward body postures. These occupational demands substantially increase the risk of workrelated musculoskeletal disorders, particularly affecting the lower back, shoulders, neck, and upper limbs (Benos et al., 2020) [8] .

Wearable Motion Capture Systems

Wearable motion capture technologies have been extensively employed to quantify biomechanical loading during harvesting activities. Inertial motion capture systems such as Xsens continuously record body posture, joint movements, and movement frequencies while workers perform harvesting operations. These systems provide objective biomechanical measurements that facilitate identification of repetitive movements and awkward working postures associated with elevated musculoskeletal risk (Merino et al., 2019) [35] . Applications in banana harvesting have demonstrated the capability of wearable motion capture systems to identify excessive trunk flexion, shoulder abduction, and repetitive upper-limb movements resulting from handling large and heavy banana bunches. Such biomechanical analyses provide quantitative evidence that supports ergonomic redesign of harvesting methods and equipment (Merino et al., 2019) [35] .

Human-Robot Collaboration

Human-Robot Collaboration (HRC) has emerged as an effective strategy for reducing the physical demands associated with harvesting operations. Rather than replacing workers entirely, collaborative robots (cobots) assist agricultural workers by transporting harvested produce and carrying heavy loads throughout the harvesting process. This approach substantially reduces repetitive lifting, prolonged carrying, and frequent bending, thereby decreasing physical workload while simultaneously improving harvesting efficiency (Adamides & Edan, 2023) [1] . Within collaborative harvesting systems, workers continue to perform tasks requiring human judgment and ability, whereas cobots assume responsibility for repetitive material transportation. This complementary distribution of labor minimizes musculoskeletal loading without compromising operational productivity and represents an important step toward human-centered automation in Agriculture 4.0 (Adamides & Edan, 2023) [1] .

Ergonomic Harvesting Platforms

Mechanical harvesting aids have also been developed to reduce ergonomic risks during fruit and vegetable harvesting. Motorized picking platforms eliminate the need for ladders and substantially reduce prolonged overhead reaching, repetitive climbing, and carrying of heavy collection bags. Consequently, these platforms decrease the likelihood of falls while minimizing asymmetric spinal loading associated with conventional harvesting practices (Fathallah, 2010) [21] . Compared with traditional harvesting methods, ergonomic harvesting platforms improve worker safety while simultaneously increasing harvesting efficiency by reducing unnecessary physical movements and shortening transportation distances during produce collection.

Digital Ergonomic Assessment During Harvesting

The integration of wearable sensing technologies, motion capture systems, and artificial intelligence enables continuous ergonomic monitoring throughout harvesting operations. Realtime posture analysis allows early identification of hazardous movements and cumulative physical loading, facilitating timely ergonomic interventions before musculoskeletal injuries occur. Such objective biomechanical assessment provides valuable guidance for optimizing harvesting techniques, equipment design, and work organization while promoting long-term occupational health.

Weeding and Planting Tasks

Weeding and planting operations are characterized by prolonged stooping, repetitive upper-limb movements, kneeling, squatting, and sustained trunk flexion, making them among the most ergonomically demanding agricultural activities. Continuous exposure to awkward postures contributes substantially to lower back pain, knee disorders, and muscular fatigue, particularly when work is performed under unfavorable ground conditions (Benos et al., 2020) [8] .

Automated Weeding Systems

Artificial intelligence has facilitated the development of automated weeding technologies that substantially reduce the need for manual weed removal. GPS-guided computer vision systems accurately identify crop rows and distinguish cultivated plants from weeds using digital image analysis. Automatic weed knives subsequently remove unwanted vegetation with minimal human intervention, thereby reducing repetitive manual weeding and limiting worker exposure to herbicides (Fathallah, 2010) [21] . Automation not only decreases physical workload but also improves operational precision by enabling targeted weed removal while minimizing crop damage. Consequently, AIassisted weeding systems contribute simultaneously to improved ergonomic conditions and enhanced agricultural productivity.

Digital Human Modeling for Tool Design

Digital Human Modeling (DHM) combined with Computer-Aided Design (CAD) has become an effective approach for optimizing agricultural hand tools used during planting and weeding. Virtual ergonomic simulations enable designers to evaluate worker posture while manipulating different tool geometries, facilitating the development of equipment that better matches workers' anthropometric characteristics (Patel et al., 2016) [38] . For example, DHM-based simulations have demonstrated that appropriately designed long-handled hoes substantially reduce trunk flexion during manual weeding operations. By improving tool dimensions before physical manufacturing, digital ergonomic design minimizes physical strain while simultaneously enhancing worker productivity and comfort (Benos et al., 2020; Patel et al., 2016) [8,38] .

Real-Time Monitoring Under Challenging Field Conditions

Agricultural planting frequently occurs under difficult environmental conditions that influence worker biomechanics.

Investigations of rice transplantation have shown that muddy terrain significantly increases lower-extremity muscular activity and physical discomfort because workers must continuously stabilize themselves while maintaining awkward working postures (Apicella & Tarabella, 2025) [5] . Digital ergonomic monitoring systems incorporating wearable sensors and computer vision enable continuous assessment of worker posture under these challenging conditions. Real-time biomechanical feedback facilitates early identification of excessive physical loading while supporting the implementation of protective equipment and ergonomic interventions that reduce environmental risk factors during planting operations (Benos et al., 2020; Apicella & Tarabella, 2025) [5,8] .

Assistive Wearable Technologies

Wearable assistive technologies have recently been investigated for reducing physical workload during repetitive digging and planting activities. Personal assist suits and lower-back exoskeletons provide mechanical assistance to the lumbar region during repeated bending, decreasing activation of the lumbar erector spinae muscles while promoting more stable movement patterns throughout work activities (Apicella & Tarabella, 2025) [5] . These wearable assistive systems demonstrate the growing role of human-centered technologies in minimizing musculoskeletal loading during labor-intensive agricultural operations. Continued improvements in comfort, flexibility, and usability are expected to further enhance their practical applicability within modern agricultural work systems.

Benefits of Digital and AI-Based Ergonomics in Agriculture

Digital technologies and AI are transforming agricultural ergonomics by enabling continuous monitoring, predictive risk assessment, and data-driven decision-making. These advancements enhance worker safety, improve equipment design, and increase operational efficiency. They also support training and long-term occupational health management in agricultural environments (Apicella & Tarabella, 2025; Naranjo et al., 2025) [5,37] .

Enhanced Occupational Health and Injury Prevention

Wearable sensors, IoT devices, and exoskeletons enable continuous monitoring of posture, fatigue, muscle activity, and environmental risks such as heat stress. Early detection of ergonomic hazards allows timely corrective actions and reduces musculoskeletal disorders. These systems significantly improve worker safety and long-term health outcomes in agriculture (Kapse et al., 2024; Yang et al., 2024) [30,55] . [27,34] .

Objective Ergonomic

Digital Human Modeling and Virtual Prototyping

Digital Human Modeling (DHM) and CAD-based simulations enable ergonomic evaluation of tools and machinery before physical production. These virtual environments improve safety, usability, and comfort while reducing design costs. They also allow adaptation to diverse worker anthropometry and agricultural conditions (Slob et al., 2023; Peladarinos et al., 2023) [39,51] .

Operational Efficiency and Decision Support

AI systems, collaborative robots, and Digital Twin technologies optimize agricultural workflows and reduce physical workload. They enhance resource allocation, improve productivity, and support real-time decision-making. These technologies ensure efficiency while maintaining worker health and safety (Adamides & Edan, 2023; Naranjo et al., 2025) [1,37] . [27,51] .

Real

Challenges and Limitations 9.1 Technical and Environmental Challenges

Agricultural environments expose digital ergonomic systems to dust, rain, heat, uneven terrain, and variable lighting, which reduce sensor reliability and computer vision accuracy. Occlusion from crops, tools, and machinery further affects posture detection and ergonomic risk estimation. Wearable sensors may also experience motion artifacts and unstable wireless connectivity during field use. These factors limit continuous and accurate monitoring in real-world conditions. Therefore, robust sensor fusion and improved AI models are required for reliable performance (Apicella & Tarabella, 2025; Naranjo et al., 2025) [5,37] .

Artificial Intelligence and Data Constraints

AI-based ergonomic systems require large, diverse, and wellannotated datasets, which are often unavailable in agricultural contexts. Deep learning models also demand high computational resources, limiting deployment on portable or edge devices. Many general pose estimation models include unnecessary body landmarks, reducing efficiency for task-specific analysis. Developing lightweight AI architectures and agriculture specific datasets is therefore essential. These improvements will enhance scalability and practical usability in farming environments (Kapse et al., 2024; Yang et al., 2024) [30,55] .

Physical and Usability Limitations

Wearable ergonomic technologies, such as exoskeletons, are often heavy, uncomfortable, and may restrict natural movement during prolonged agricultural tasks. Virtual Reality (VR) systems can also cause motion sickness, dizziness, and visual fatigue with extended use. Additionally, there is a lack of standardized guidelines for implementing these technologies across different farming operations. Improving ergonomics, comfort, and usability is essential to increase worker acceptance. User-centered design approaches are therefore critical for successful adoption (Hilmi et al., 2024; Slob et al., 2023) [27,51] .

Socioeconomic and Infrastructural Barriers

High costs of AI systems, robotics, wearable devices, and VR technologies significantly limit adoption, especially among small-scale farmers. Many rural areas also lack reliable internet connectivity and digital infrastructure required for real-time monitoring systems. Cultural resistance, privacy concerns, and low technological trust further hinder acceptance. These challenges highlight the need for affordable, scalable, and farmer-friendly solutions. Policy support and inclusive implementation strategies are essential for wider adoption (Apicella & Tarabella, 2025; Naranjo et al., 2025) [37] .

Methodological Limitations in Existing Research

Most existing studies rely on short-term laboratory experiments, which do not fully represent real agricultural working conditions. Seasonal variations, long-term fatigue, and cumulative musculoskeletal load are often not considered. There is also limited large-scale field validation of digital ergonomic systems. This restricts the generalizability of current findings. Future research should focus on long-term, real-world, and multidisciplinary field studies to validate effectiveness in agriculture (Massiris-Fernández et al., 2020; Peladarinos et al., 2023) [34,39] .

Future Research Directions

Future research should focus on transforming digital ergonomics from experimental technologies into practical, field-validated solutions for agriculture. Although advances have been made in AI, wearable sensors, computer vision, Digital Human Modeling (DHM), and robotics, most studies remain laboratory based. Future investigations should emphasize long term field validation, improved system reliability, and integration of intelligent technologies into occupational health management (Apicella & Tarabella, 2025; Naranjo et al., 2025) [5,37] .

Advanced Artificial Intelligence Modeling and Technical Refinement

Future AI research should develop advanced deep learning models such as CNN-LSTM and GCN-LSTM to better capture dynamic agricultural movements and repetitive tasks. Improvements in 3D pose estimation, multi-camera systems, and comprehensive agricultural posture datasets will enhance assessment accuracy under field conditions. AI should also evolve into decision-support systems capable of recommending personalized ergonomic interventions before injuries occur (Kapse et al., 2024; Yang et al., 2024; Massiris-Fernández et al., 2020) [30,34,55] .

Human-Centered and Inclusive Design

Future ergonomic technologies should consider worker diversity, including age, gender, anthropometry, and physical capability, to improve assessment accuracy and inclusiveness. Research should also address worker acceptance, privacy concerns, and cultural factors influencing technology adoption. Affordable AI systems, lightweight wearable devices, and userfriendly VR technologies will promote wider adoption, especially among small-scale farmers (Apicella & Tarabella, 2025; Hilmi et al., 2024; Slob et al., 2023) [5,27,51] .

Systems Integration and Robotics

Future studies should integrate wearable sensors with farm management systems, Decision Support Systems (DSS), and Enterprise Resource Planning (ERP) platforms for comprehensive worker and farm monitoring. Human-Robot Collaboration (HRC) should prioritize worker safety and ergonomic support, while Digital Twin technology should enable realistic simulation of agricultural operations. Economic analyses, including cost-benefit and return-on-investment (ROI) studies, are also needed to demonstrate the practical value of digital ergonomic technologies (Adamides & Edan, 2023; Slob et al., 2023; Peladarinos et al., 2023; Naranjo et al., 2025) [1,37,39,51] .

Conclusion

Agriculture remains one of the most ergonomically challenging occupations due to repetitive movements, awkward postures, heavy manual material handling, vibration, and adverse environmental conditions that increase the risk of work-related musculoskeletal disorders (WMSDs). This review demonstrates that digital ergonomics, integrated with Artificial Intelligence (AI), offers significant advancement over conventional observation based ergonomic assessment methods. Technologies such as wearable sensors, Internet of Things (IoT), computer vision, motion capture systems, Digital Human Modeling (DHM), Digital Twins, Virtual Reality (VR), and robotics enable continuous, objective, and real-time monitoring of workers' biomechanical, physiological, and environmental conditions. AI and machine learning further enhance ergonomic assessment through automated posture recognition, ergonomic risk classification, and predictive injury analysis, enabling early intervention before injuries occur. These technologies also support the design of ergonomically optimized tools, machinery, and workstations while promoting human-robot collaboration to reduce physical workload and improve operational efficiency. The review further highlights their successful application in harvesting, planting, weeding, pruning, and tractor operation, demonstrating their potential to improve worker safety, productivity, and overall occupational health. However, challenges such as limited field validation, high implementation costs, insufficient agriculture-specific datasets, sensor reliability under harsh environmental conditions, and worker acceptance continue to restrict widespread adoption. Future research should therefore focus on developing robust, affordable, and fieldvalidated AI-driven ergonomic systems through multidisciplinary collaboration and user-centered design. Overall, the integration of digital technologies with AI represents a transformative step toward predictive, personalized, and preventive occupational health management in Agriculture 4.0. Continued advancements in these technologies will contribute to safer working environments, reduced incidence of WMSDs, improved farm productivity, and more sustainable agricultural work systems.

References

1. Adamides G, Edan Y. Human-robot collaboration systems in agricultural tasks: A review and roadmap. Computers and Electronics in Agriculture. 2023;204:107541. DOI:10.1016/j.compag.2022.107541.
2. Adhitya Y, Mulyani GS, Köppen M, Leu JS. IoT and deep learning-based farmer safety system. Sensors. 2023;23(6):2951. DOI:10.3390/s23062951.
3. Ali I, Chen L, Huang Y, Song L, Lu X, Liu B, et al. Humidity-responsive gold aerogel for real-time monitoring of human breath. Langmuir. 2018;34(16):4908-4913. DOI:10.1021/acs.langmuir.8b00472.
4. Anagnostis A, Benos L, Tsaopoulos D, Tagarakis A, Tsolakis N, Bochtis D. Human activity recognition through recurrent neural networks for human-robot interaction in agriculture. Applied Sciences. 2021;11(5):2188. DOI:10.3390/app11052188.
5. Apicella A, Tarabella A. Wearable technologies in Agriculture 4.0: A systematic review of applications for worker safety and ergonomic support. International Journal of Academic Research in Business and Social Sciences. 2025;15(8). DOI:10.6007/ijarbss/v15-i8/26140.
6. Arrais A, Dias D, Cunha JPS. Novel real-time metrics for quantified vineyard workers' operations with wearable devices. In: 2023 IEEE 7th Portuguese Meeting on Bioengineering (ENBENG); 2023. p. 92-96. DOI:10.1109/ENBENG58165.2023.10175321.
7. Barbie A, Hasselbring W, Hansen M. Digital twin prototypes for supporting automated integration testing of smart farming applications. Symmetry. 2024;16(2):221. DOI:10.3390/sym16020221.
8. Benos L, Tsaopoulos D, Bochtis D. A review on ergonomics in agriculture. Part I: Manual operations. Applied Sciences. 2020;10(6):1905. DOI:10.3390/app10061905.
9. Cannady RT, Yoder A, Miller J, Crosby K, Kintziger KW. Understanding and perceiving heat stress risk control: Critical insights from agriculture workers. Journal of Occupational and Environmental Hygiene. 2025;22(3):203213. DOI:10.1080/15459624.2024.2439812.
10. Caputo F, Greco A, Fera M, Macchiaroli R. Digital twins to enhance the integration of ergonomics in the workplace design. International Journal of Industrial Ergonomics. 2019;71:20-31. DOI:10.1016/j.ergon.2019.02.001.
11. Caria M, Sara G, Todde G, Polese M, Pazzona A. Exploring smart glasses for augmented reality: A valuable and integrative tool in precision livestock farming. Animals. 2019;9(11):903. DOI:10.3390/ani9110903.
12. Catania P, Aiello G, Certa A, Ferro MV, Orlando S, Vallone M. The MAGIC project: A tool for promoting safety in agriculture during COVID-19 pandemic. In: Biosystems Engineering Towards the Green Deal. Cham: Springer; 2023. p. 437-445. DOI:10.1007/978-3-03130329-6_45.
13. Cha EH, Kim K, Oh SY, Yu MI, Kwon TK. Analysis of physiological signals on the wearable assist suit for repetitive agricultural task. Journal of Mechanics in Medicine and Biology. 2021;21(9):2140032. DOI:10.1142/S0219519421400327.
14. Chittar OA, Barve SB. Waist-supportive exoskeleton: Systems and materials. Materials Today: Proceedings. 2022;57:840-845. DOI:10.1016/j.matpr.2022.02.455.
15. Chouriya A, Soni P, Hota S, Thomas EV. Ergonomic evaluation of a walk-behind AI-based cotton fertilizer applicator. Journal of Field Robotics. 2025;42(5):18261839.
16. Cividino SRS, Zaninelli M, Redaelli V, Belluco P, Rinaldi F, Avramovic L, et al. Preliminary evaluation of new wearable sensors to study incongruous postures held by employees in [source information incomplete]. Sensors. 2024. DOI:10.3390/s24175703.
17. Lundstrom D, Nevaranta T, Hanson L, Hogberg D, Sundin A. Visualization of comfort and reach in cab environment. In: 40th Annual Conference of the Nordic Ergonomics Society; 2008 Aug 11-13; Reykjavik, Iceland.
18. Das B. Work-related musculoskeletal disorders in agriculture: Ergonomics risk assessment and its prevention among Indian farmers. Work. 2023;76(1):225-241. DOI:10.3233/WOR-220246.
19. Donisi L, Cesarelli G, Pisani N, Ponsiglione AM, Ricciardi C, Capodaglio E. Wearable sensors and artificial intelligence for physical ergonomics: A systematic review of literature. Diagnostics. 2022;12(12):3048. DOI:10.3390/diagnostics12123048.
20. Fathallah FA, Chang JH, Pickett W, Marlenga B. Ability of youth operators to reach farm tractor controls. Ergonomics. 2009;52(6):685-694.
21. Fathallah FA. Musculoskeletal disorders in labor-intensive agriculture. Applied Ergonomics. 2010;41(6):738-743. DOI:10.1016/j.apergo.2010.03.003.
22. Grandi F, Prati E, Peruzzini M, Pellicciari M, Campanella CE. Design of ergonomic dashboards for tractors and trucks: Innovative method and tools. Journal of Industrial Information Integration. 2022;25:100304. DOI:10.1016/j.jii.2021.100304.
23. Hsiao H, Whitestone J, Bradtmiller B, Whisler R, Zwiener J, Lafferty C, et al. Anthropometric criteria for the design of tractor cabs and protection frames. Ergonomics. 2005;48(4):323-353.
24. Hall AR, Hose N, Nakajima H, Kahng SJ, Chung JH, Novosselov I, et al. Design and fabrication of wearable sensors for the measurement of ammonia concentration in air. In: IECON 2024 -50th Annual Conference of the IEEE Industrial Electronics Society; 2024. p. 1-6. DOI:10.1109/IECON55916.2024.10905302.
25. Hayati A, Marzban A. A review of the ergonomic problems of agricultural activities in connection with mechanization: Some Iranian farm works as instances. Work. 2025;82(3):715-726. DOI:10.1177/10519815251351602.
26. Hilmi AH, Rasyidah A, Hamid AW, Abdul W, Assyahid R. [Article title not provided]. Malaysian Journal of Ergonomics. 2024.
27. Deisinger J, Breining R, Robler A, Holfe I, Ruckert D. Immersive ergonomic analyses of console elements in a tractor cabin. In: Proceedings of the Fourth Immersive Projection Technologies Workshop; 2000; Iowa.
28. Chang JH, Fathallah FA, Pickett W, Miller BJ, Marlenga B. Limitations in fields of vision for simulated young farm tractor operators. Ergonomics. 2010;53(6):758-766.
29. Kapse S, Wu R, Thamsuwan O. Addressing ergonomic challenges in agriculture through AI-enabled posture classification. Applied Sciences. 2024;14(2):525. DOI:10.3390/app14020525.
30. Kee D. Participatory ergonomic interventions for improving agricultural work environment: A case study in a farming organization of Korea. Applied Sciences. 2022;12(4):2263. DOI:10.3390/app12042263.
31. Kirkhorn SR, Earle-Richardson G, Banks RJ. Ergonomic risks and musculoskeletal disorders in production agriculture: Recommendations for effective research to practice. Journal of Agromedicine. 2010;15(3):281-299. DOI:10.1080/1059924X.2010.488618.
32. Martins B, Baesso B, EC R, Engineer A, Fischer B, Engineer E, et al. Noise and vibration of tractors: An ergonomic evaluation. International Journal of Applied Science and Technology. 2014;4(4).
33. Massiris-Fernández M, Fernández JA, Bajo JM, Delrieux CA. Ergonomic risk assessment based on computer vision and machine learning. Computers and Industrial Engineering. 2020;149:106816. DOI:10.1016/j.cie.2020.106816.
34. Merino G, da Silva L, Mattos D, Guimarães B, Merino E. Ergonomic evaluation of the musculoskeletal risks in a banana harvesting activity through qualitative and quantitative measures, with emphasis on motion capture (Xsens) and EMG. International Journal of Industrial Ergonomics. 2019;69:80-89. DOI:10.1016/j.ergon.2018.10.004.
35. Mushtaq SA, Gul A, Dixit J. Occupational health impacts of tractor noise and its attenuation through ergonomic interventions. Journal of Biosystems Engineering. 2025;50(2):225-239. DOI:10.1007/s42853-025-00261-4.
36. Naranjo JE, Mora CA, Bustamante Villagómez DF, Mancheno Falconi MG, Garcia MV. Wearable sensors in industrial ergonomics: Enhancing safety and productivity in Industry 4.0. Sensors. 2025;25(5):1526. DOI:10.3390/s25051526.
37. Patel T, Sanjog J, Karmakar S. Ergonomics perspective in agricultural research: A user-centred approach using CAD and digital human modeling (DHM) technologies. Journal of The Institution of Engineers (India): Series A. 2016;97(3):333-342. DOI:10.1007/s40030-016-0162-2.
38. Peladarinos N, Piromalis D, Cheimaras V, Tserepas E, Munteanu RA, Papageorgas P. Enhancing smart agriculture by implementing digital twins: A comprehensive review. Sensors. 2023;23(16):7128. DOI:10.3390/s23167128.
39. Petrosyan T, Dunoyan A, Mkrtchyan H. Application of motion capture systems in ergonomic analysis. Armenian Journal of Special Education. 2020;4(2):107-117.
40. Pirozzi M, Donato LD, Tomassini L, Ferraro A. Possible innovative technical measures for risk prevention during the use of mobile machines with remote guide/control. Procedia Manufacturing. 2020;42:457-461. DOI:10.1016/j.promfg.2020.02.049.
41. Pistolesi F, Lazzerini B. Assessing the risk of low back pain and injury via inertial and barometric sensors. IEEE Transactions on Industrial Informatics. 2020;16(11):71997208. DOI:10.1109/TII.2020.2992984.
42. Ying Q, Hui L, Zhenghe S, Zhongxiang Z, Enrong M. Design of agricultural equipment cab based on virtual reality. In: International Conference on Advanced Technology of Design and Manufacture (ATDM); 2010. p. 62-67.
43. Ross JA, Shipp EM, Trueblood AB, Bhattacharya A. Ergonomics and beyond. Human Factors. 2016;58(5):777795. DOI:10.1177/0018720816645457.
44. Rybnikár F, Kačerová I, Hořejší P, Šimon M. Ergonomics evaluation using motion capture technology-Literature review. Applied Sciences. 2023;13(1):162. DOI:10.3390/app13010162.
45. Safari M, Naserbakht AH, Badri Kouhi A, Varmazyar S. Artificial intelligence and emerging technologies in assessing ergonomic risk factors in the workplace: A systematic review. Work. 2025;82(3):727-739. DOI:10.1177/10519815251349793.
46. Salisu S, Ruhaiyem NIR, Eisa TAE, Nasser M, Saeed F, Younis HA. Motion capture technologies for ergonomics: A systematic literature review. Diagnostics. 2023;13(15):2593. DOI:10.3390/diagnostics13152593.
47. Schuman SH. Ergonomics in agriculture: Commentary and literature review. Journal of Agromedicine. 2002;8(1):9-18. DOI:10.1300/J096v08n01_02.
48. Sharma NK, Tiwari M, Thakur A, Ganguli AK. Evaluating biomechanical risks in manual material handling: An ergonomic intervention approach. Ergonomics. 2025. DOI:10.1080/00140139.2025.2473678.
49. Singh LP, Kumar P, Lohan SK. Development of a real-time work-related postural risk assessment system of farm workers using a sensor-based artificial intelligence approach. Journal of Field Robotics. 2024;41(7):21002113. DOI:10.1002/rob.22215.
50. Slob N, Hurst W, van de Zedde R, Tekinerdogan B. Virtual reality-based digital twins for greenhouses: A focus on human interaction. Computers and Electronics in Agriculture. 2023;208:107815. DOI:10.1016/j.compag.2023.107815.
51. Son H, Lim JW, Park S, Park B, Han J, Kim HB, et al. A machine learning approach for the classification of falls and activities of daily living in agricultural workers. IEEE Access. 2022;10:77418-77431. DOI:10.1109/ACCESS.2022.3190618.
52. Vitale E, Vella F, Matera S, Rizzo GC, Rapisarda L, Roggio F, et al. Precision agriculture: Assessment of ergonomic risks of assisted driving system. Applied Sciences. 2024;14(9):3738. DOI:10.3390/app14093738.
53. Xiao-yan Y, Zheng-he S, Zhong-xiang Z, En-rong M. Expert system for tractor cab man-machine interface evaluation. In: International Conference on Computational Intelligence and Software Engineering (CiSE); 2009. p. 1-4.
54. Yang Z, Song D, Ning J, Wu Z. A systematic review: Advancing ergonomic posture risk assessment through the integration of computer vision and machine learning techniques. IEEE Access. 2024;12:180481-180519. DOI:10.1109/ACCESS.2024.3509447.
55. Yousefi S, Aryal B, Shutske J, Issa SF. Enhancing occupational safety through AI: A review of key AI technologies. Journal of the ASABE. 2026;69(1):165-179.