Horse gait analysis using wearable inertial sensors and machine learning

Abstract

Equestrian sports require horses to possess physical and mental attributes such as agility, strength, balance, and gymnastic skills. Performance analysis is critical in evaluating a horse’s performance, which involves assessing athleticism, gait quality, jumping ability, and general health. Assessing the kinematics of horses is crucial for selecting, training, and managing sports horses. Understanding a horse’s gait pattern and detecting Ground Reaction Forces (GRF) help diagnose lameness in the horse. Traditional gait analysis methods are performed visually, which can be biased due to subjectivity and human error. Optical motion capture (OMC) technology for equine gait analysis is expensive and ideal for indoor use. Wearable inertial measurement units (IMUs) offer a cost-effective alternative for analyzing kinematic parameters. This study has devised novel wearable sensor devices for horses and riders to measure forces acting on the legs and body of the horse and the orientation of their legs during field performance. Ground Reaction Forces (GRF) were measured using 100g accelerometer data from each leg to assist owners and riders in analyzing the magnitude of forces and detecting any anomalies. Machine-learning models were developed to classify horse movements, such as jumps, stands, gallops, and trots, using features extracted from the data collected by wearable sensor devices. These models were compared to identify the most effective model for accurately classifying horse movements. This approach provides a valuable tool for recognizing patterns and trends in the data, enabling owners and riders to make informed decisions about training and management strategies.