Extracting those LiDAR pointclouds – a ML search for shapes

Abstract

Emerging in the 1960s, LiDAR technology has experienced significant development due to its undeniable advantages and is now widely applied across various fields. Its most successful integrations have been seen in geodesy, cartography, archaeology, environmental monitoring, and the automotive industry. Today, LiDAR has become an integral part of modern systems such as unmanned aerial vehicles (UAVs), drones, and augmented reality platforms.

LiDAR has supplemented—and in some cases replaced—traditional methods of spatial data acquisition. Providing high measurement accuracy, these technologies enable the rapid collection of large volumes of data.

Alongside the development of LiDAR itself, there has been substantial progress in methods for interpreting the resulting data. Traditional approaches such as statistical analysis, geostatistics, and geospatial analysis are widely used. However, intelligent solutions based on machine learning (ML) have recently become more prominent for deriving valuable analytical insights from LiDAR data.