Author(s): WU, K., KUO, C., MA, H., WU, S., CHANG, J., Institution: NATIONAL TSING HUA UNIVERSITY, TAIWAN, Country: TAIWAN, Abstract-ID: 775

INTRODUCTION:
Climbing performance relies heavily on body posture and weight distribution, with finger force application playing a critical role. Despite its importance, hand pressure has rarely been addressed in wearable sensor technologies due to limited space for sensor placement on the fingers, the delicate and complex nature of finger movements, and varying regulations regarding sporting gloves across disciplines. In our previous work (Kuo & Wu, 2025), a prototype smart sporting glove embedded with four thin-film force-sensitive resistor (FSR) sensors demonstrated the feasibility of automatic grip classification using convolutional neural network (CNN) models.
METHODS:
In the present study, the smart glove prototype was expanded to incorporate nine FSR sensors distributed across the hand to enable finer characterization of diverse gripping styles. Experiments were conducted with six male and female climbers across amateur, experienced, and professional levels. Hand pressure signals were collected during standardized gripping tasks and normalized prior to analysis. CNN-based models were trained to classify grip types, while repeated trials were used to assess fatigue-related changes in grip behavior.
RESULTS:
After signal normalization, consistent pressure patterns were observed for identical gripping motions across participants, supporting robust CNN-based grip classification. Fatigue effects were evident through progressive shifts in grip strategies and technique deterioration over repeated trials. Comparisons across skill levels revealed a negative correlation between climbing experience and signal noise associated with slipping. Amateur climbers exhibited higher finger force output and greater variability in pressure signals compared with experienced and professional climbers.
CONCLUSION:
The results suggest that excessive finger force in amateur climbers is linked to suboptimal posture and reduced reliance on whole-body balance strategies, contributing to higher fatigue rates. This study demonstrates the potential of a multidimensional, wearable sensing approach for objective performance and fatigue analysis in climbing. Future work will focus on expanding the dataset to establish generalized signal models of climbing ability across diverse anthropometric profiles, thereby improving the broader applicability of the proposed system.