Multi-Depth Cross-Calibration of Gaze Tracker and LiDAR Systems

Discovering where a driver looks and what they look at while driving can provide many benefits to improve driving safety in Advanced Driver Assistance Systems (ADAS). Toward this goal, we utilized a research vehicle equipped with a remote gaze tracker installed inside the vehicle and facing the driver and a LiDAR mounted on the roof of the vehicle. We introduce a novel approach for the cross-calibration of the gaze tracker with the LiDAR's system. We also propose a method to estimate the driver's Point of Gaze in the sparse LiDAR point cloud. In order to provide more information about the driver's attention, we introduce a technique to calculate the driver's visual attention area in the 3D space. We empirically demonstrate that our proposed multi-depth calibration-calibration approach yields excellent results to find the transformation matrix between a remote gaze tracker and LiDAR coordinate systems.