Real-Time Evaluation of Human Inertial properties in Immersive Simulators for Moth type Boats

The aim of this work is to present a system designed for real-time markerless evaluation of human inertial properties, to be used in a Human In The Loop (HITL) simulator of a Moth class sailboat. The method is based on human body pose estimation algorithm used for 3D skeleton reconstruction from a multi-camera acquisition. The exploitation of a markerless system enables ease of use compared to marker-based systems, and reduces signal’s disturbances compared to Inertia Measurement Units (IMUs) for body pose reconstruction. The use of multi-camera setup and the integration of YOLOv8-pose model results with image triangulation techniques reduce occlusion occurrence, measurements drift and complexity. The 3D pose obtained is then used to estimate the Center of Mass (CoM) position and the Tensor of Inertia (ToI) of the skipper. The system allows for a real-time computation of the inertial properties of the skipper, thus enabling a real-time update of the mass matrix of the numerical model running the human-in-the loop simulator. This study shows the results achievable by the system in terms of inertia estimation for different skipper’s poses, and compares the cases in which the skipper is modelled as a point mass with real-time position updates with the case in which the full inertia tensor of the skipper is considered. Results show how in contexts like those of Moth class sailboat simulators the possibility to take into account and to vary in real-time the tensor of inertia of the sailor increases the accuracy achievable in the dynamic simulation.