A Machine Learning Approach to Analyze Rider’s Effects on Horse Gait Using On-Body Inertial Sensors
Authors: H. Darbandi, P. Havinga
Journal: 2022 IEEE International Conference on Pervasive Computing and Communications Workshops and other Affiliated Events (PerCom Workshops)
Summary
# Editorial Summary Wearable inertial sensors offer valuable insight into equine gait mechanics, yet current applications struggle to distinguish between ridden and unmounted locomotion—a critical prerequisite for meaningful biomechanical analysis and welfare assessment. Darbandi and Havinga developed a machine learning classification model using minimally positioned on-body sensors to automatically detect ridden versus unridden states in real time, addressing a notable gap in equine monitoring technology despite the widespread use of sensors for identifying other gait characteristics. The research successfully demonstrated that accurate classification of ridden state is achievable with a streamlined sensor setup, though specific accuracy figures and sensor placement protocols warrant review of the full paper for implementation guidance. This capability matters significantly for practitioners relying on wearable technology for performance monitoring or longitudinal soundness assessment, as automated rider-state detection enables more precise baseline comparisons and reduces manual data processing overhead. For farriers, physiotherapists, and veterinary professionals using gait analysis in clinical or performance contexts, this work suggests a pathway towards automated, objective evaluation systems that account for rider influence on horse biomechanics—particularly valuable given that rider position and weight distribution meaningfully alter movement patterns.
Read the full abstract on the publisher's site
Practical Takeaways
- •Wearable inertial sensors can objectively detect whether a horse is being ridden, providing a foundation for real-time monitoring of rider impact on gait
- •This technology could enable automated coaching feedback or early detection of biomechanical changes caused by improper riding
- •Minimal sensor placement reduces equipment burden while maintaining classification accuracy, making field application more practical
Key Findings
- •Machine learning models can classify ridden vs. unridden horse locomotion using minimal inertial sensors placed on the horse's body
- •Wearable sensor technology enables automatic real-time evaluation of rider effects on equine biomechanics
- •Sensor-based classification of ridden state is a prerequisite for subsequent gait analysis in equine research applications