Kinematic and Physiological Synchronization of Horse-Rider Dyad using Wearable Sensors

Equestrianism is the only Olympic sport in which two living beings of different species, the horse and the rider, must coordinate their movement to pursue a single athletic goal. Yet the harmony of the horse-rider dyad is still judged mainly by instructors’ and judges’ visual assessment or by the rider’s own feeling, both easily biased by stress and fatigue. Meanwhile, the formal recognition of the horse as an athlete calls for objective tools to monitor both performance and welfare, and literature rarely links the kinematic and physiological domains. This thesis therefore aimed to develop a framework based on wearable sensors for the quantitative and integrated characterization of kinematic and physiological synchronization within the horse-rider dyad. The relationship between humans and horses has evolved from military horsemanship to moder Olympic disciplines governed by welfare-oriented rules. Together with the ethology of equine communication, this evolution shows that performance relies on cooperation rather than coercion, and that it requires the motor and autonomic systems of both partners to synchronize. In the kinematic domain, Inertial Measurement Units make it possible to track the vertical displacement of horse and rider in the field across Walk, Trot and Canter. Coupling was quantified through Discrete and Continuous Relative Phase (DRP, CRP), harmony indices derived from Poincaré Maps (Length-of-Vector, LV, and percentage Average Deviation, AD%), and complexity and spectral metrices. In the physiological domain, interspecies cardiac coupling was assessed on baseline-normalized Heart Rate (HR) through normalized Dynamic Time Warping (DTW), windowed Spearman correlation and Directed Coherence (DC) in the Low Frequency (LF) and High Frequency (HF) bands. Eleven show-jumping horse-rider dyads with female riders were recorded during a standardized flat-work session that included resting baseline, warm-up, free-work and cool-down, with pre- and post-session questionnaire on the rider’s perceived synchronization. The horse wore an NGIUM and a Fourth Frontier® X2, and the rider wore a Zephyr® BioHarness 3.0. Four cascaded MATLAB® pipelines handles device synchronization, displacement reconstruction, automatic gait classification and stride matching, computed the metrics, and ran a cross-domain Spearman analysis with False Discovery Rate (FDR) correction. The pipelines provided robust to field contingencies such as sensor detachment and IMU disconnection, and also reflected a horse-rider conflict in one horse-rider dyad. At the single-domain level, each gait showed a distinct coordination signature, with trot giving the most stable kinematic coupling, while hear-rate similarity between horse and rider decreases as exercise intensity increased. The cross-domain analysis (N = 10) produced the central finding of this work: only two correlations survived FDR correction, both at walk. The harmony indices LV (ρ = −0.867) and AD% (ρ = −0.879) were negatively correlated with rider-to-horse DC in the HF band (pFDR = 0.048). This means that more harmonious motion was associated with a stronger physiological influence from the rider in the band linked to parasympathetic activity. No significant relationship emerged with questionnaire scores, rider experience or partnership duration. Therefore, this thesis provides a replicable, non-invasive and filed-robust framework, together with preliminary evidence that the kinematic and physiological domains interact, particularly at Walk. These findings remain exploratory because of the small and homogeneous sample, the single-session design and the lack of validation against gold-standard systems. Future work should involve larger stratified samples, equine-specific sensors and the separation of the emotional HR component.