Mixed Dimensional Multiple View Geometry and Integration of Different On-Vehicle Sensors

在先进的车辆安全系统中，多个车辆获得的传感信息的协作是非常重要的。在这项研究中，我们研究了不同的传感器之间的多视图几何，并开发了一种方法，用于计算多个车辆的3D位置和方向的传感信息，由不同类型的传感器，如相机和激光雷达，安装在多个车辆。特别是，我们表明，通过使用另一辆车上的摄像头观察激光雷达投射在一辆车上的激光点，我们可以计算这些车辆之间的相对位置和方向。通过使用我们的方法，我们可以报告接近车辆的司机在周围的盲目交叉口。

In advanced vehicular safety systems, cooperation of sensing information obtained in multiple vehicles is very important. In this research, we studied multiple view geometry among different sensors, and developed a method for computing 3D position and orientation of multiple vehicles from sensing information obtained by different types of sensors, such as cameras and laser radars, mounted on multiple vehicles. In particular, we showed that by observing laser points projected by a laser radar on one vehicle using a camera on the other vehicle, we can compute relative position and orientation between these vehicles. By using our method, we can report approaching vehicles to drivers at around blind intersections.