Multi-layer Caching and Parallel Streaming for Large Scale Cloud Optimized Point Cloud Data Visualization using WebGPU.

As LiDAR sensors become more precise and widely used, effectively managing and rendering large amounts of point data is becoming increasingly difficult. Available web based solutions for large scale point cloud visualization do not take advantage of modern cloud optimized data formats which are well suited for parallel and efficient data streaming. Furthermore, current web visualization tools do not take advantage of new capabilities like persistent file caching, losing their data when a tab/window is closed. In this work, we present the first web based viewer for point cloud data using a multi-layer cache system that allows to store data using both file and memory storage directly from a browser. Furthermore, we also introduce the first open source viewer for the cloud optimized point cloud (COPC) data format employing a parallel workflow using WebGPU to stream, process and render point cloud data. Experimental studies of the multi-layer cache system demonstrates to provide the best performance in different configurations, including offline data visualization.