2 February 2024 in IEEE Access
Pressure insoles allow for the collection of real time pressure data inside and outside a laboratory setting as they are non-intrusive and can be simply integrated into industrial environments for occupational health and safety monitoring purposes. Activity detection is important for the safety and wellbeing of workers, and the present study aims to employ pressure insoles to detect the type of industry-related task an individual is performing by using random forest, an artificial intelligence-based classification technique. Twenty subjects wore loadsol® pressure insoles and performed five specific tasks associated with a typical workflow: standing, walking, pick and place, assembly, and manual handling. For each activity, statistical and morphological features were extracted to create a training dataset. The classifier performed with an accuracy over 82%, using ten-fold cross-validation, for a time window of 5 seconds, showing the potential for task classification in edge-AI applications in smart manufacturing environments. A re-analysis focused on the five most influential features obtained 83% accuracy. The combination of random forest and in-depth feature analysis (SHAP) provided insights into the importance of features and the impact of their value on each task class. Such an understanding can aid in reducing misclassifications for health and safety purposes and can aid in the design of pressure insoles that are optimized for impactful features. The accuracy achieved is comparable to similar task classification studies but with the benefit of added explainability, which increases transparency and, thereby, trust in the classifier decisions.
24 February 2025 in Applied Ergonomics
30 January 2026 in IEEE Access
Accurately estimating physical fatigue is critical for preventing injuries, enhancing performance and supporting rehabilitation, yet current approaches are often limited to specific tasks with no interpretability. In this work, we present a systematic framework for predicting perceived fatigue in upper limb movements using wearable inertial measurement units (IMU) and surface electromyography (EMG). We processed two publicly available datasets into structured repetition level data with extracting over 3,700 biomechanical features across time, frequency and time–frequency domains. We model perceived fatigue using the Borg Rating of Perceived Exertion (RPE) scale. A six-level hierarchical evaluation was designed to progressively test models’ performance and generalisation, from single-task conditions to cross-family transfer. The results show that the models achieve the highest accuracy in task-specific contexts (R2 to 0.69), but the performance declines with increasing task diversity, with cross-family transfer resulting in a negative R2. Fewshot domain adaptation significantly improved performance by turning negative R2 into positive values across all family combinations. Feature importance analyses found that task specific models relied more heavily on IMU based kinematic features, while EMG features grew in importance as models generalised. Sensor reduction experiments showed that close to optimal accuracy can be achieved with as few as four to six sensors compared to eleven sensors for the full dataset. This study provides a unified methodology for fatigue estimation, exposes the challenges of cross-task generalisation and provides guidelines for sensor configurations that balance performance with minimal burden.
7 August 2025 in IEEE Sensors Letters
Accurately estimating perceived fatigue from wearable sensor data is a challenge, especially across diverse tasks. This letter presents a generalized framework to predict estimated fatigue scores (measured using the Borg scale) using combined electromyography and inertial measurement units data collected from two independent upper limb datasets. Our best model achieved a mean absolute error of 2.35 and a mean absolute percentage error of 18.60% using only five strategically placed sensors. A broad set of biomechanical features was extracted to capture both kinematic and neuromuscular indicators of fatigue. Vertical acceleration of the upper arm and shoulder, along with spectral features from deltoid EMG, emerged as the most consistent predictors across tasks. These findings support interpretable and generalizable fatigue detection and provide a foundation for real-time monitoring systems in sports, rehabilitation, and occupational health.
17 July 2026 in Applied Ergonomics
The SAFE (System for Activity-aware Fatigue Evaluation) framework integrates wearable sensing, biomechanical modelling, and task detection to enable monitoring of operator endurance and ergonomic risk. In this feasibility study, the framework was evaluated offline in a laboratory environment with 10 participants performing industry-type tasks, combining inertial measurement units (IMUs), pressure insoles, a torque-based endurance model, task classification, and a custom application. Validation against motion capture confirmed that IMU-derived joint angles were within acceptable error ranges for endurance estimation. The framework produced endurance metrics that distinguished between ergonomically high- and low-risk tasks (P < .05) and correlated strongly with subjective fatigue ratings (rmm≥.878, P < .001). IMUs captured arm movement variability across dynamic tasks and task classification via pressure insole data enabled the application of task-specific endurance coefficients. These results demonstrate that the SAFE framework is a feasible predictive method for online operator monitoring and ergonomic risk assessment, providing a foundation for future real-time deployment.