Résumé
This article presents a novel approach to karstic exploration using an autonomous drifter for path reconstruction in GPS-denied environments. Karst reservoirs supply drinking water to approximately 25% of the global population, yet their internal structure and flow dynamics remain poorly characterized. Over the past 50 years, advances in hydrodynamic modeling have improved our understanding of these systems, but significant challenges persist. Current monitoring methods rely primarily on artificial tracers, which provide only an average flow velocity between an inlet and an outlet (spring), or on cave diving, which is severely limited in spatial coverage and safety. These constraints hinder the development of more accurate models of karst networks. To address these limitations, this study introduces a proof-of-concept autonomous drifter designed to navigate and map submerged conduits. The drifter operates without GPS, instead relying on onboard sensors to measure its movement and orientation, enabling path reconstruction in unknown environments. A novel algorithm is proposed to compensate for the absence of external positioning data. The results demonstrate the feasibility of autonomous exploration in submerged karst systems, laying the groundwork for future improvements in both technology and methodology for characterizing complex underground water networks.