I. Introduction
A point cloud is a collection of points spread across a 3D space, which can be used to represent a local environment. Point cloud data collected by LiDAR sensors is currently being used for driving environment representation by many autonomous driving systems [1], including those used by Google and Uber [2], and the LiDAR systems used to collect this data are also becoming cheaper. For example, Velodyne’s HDL-64, one of the most popular 3D LiDAR sensors, used to cost around 100,000, but a less expensive, less powerful version, the VLP-16, now costs only about 4,000, and Velodyne predicts the cost may fall to $50 in the future. These trends suggest that the use of point cloud data is becoming an industry standard and that it will continue to be used in autonomous driving applications in the foreseeable future. Shared and stored streaming point cloud data are also likely be important components of accident investigation and future V2V/V2X systems.