GitHub Repository

Exploiting Multiple Echoes in 3D Point Cloud Classification Generated from LiDAR Sensors in Precision Agriculture Applications

The present project aims to study the possibilities offered by the technologies of Light Detection and Determination (LiDAR) in the context of crop monitoring. LiDAR technologies are now the sensor of choice in robotics and airborne applications to provide range data. While most LiDARs only provide a simple point cloud by associating a single distance to each measurement, some can provide more information, such as the multiple echoes generated by the environment illuminated by the LiDAR beam and the associated returned energy. The physics behind LiDAR is in fact more complex than single flight distance measurements, since the beam is not infinitely narrow, and the reflected energy is a function of the nature and orientation of the impacted materials. The higher performance LIDARs return the “full wave” of the echoes, which is the full (sampled) function that measures the energy returned as a function of time. Such LIDARs are quite expensive and have been successfully exploited in classical airborne images at altitudes of about 4000 m, where, in particular, they allow to distinguish the canopy from the ground when flying over forests, and for example evaluate the amount of wood that can be harvested or to classify elements such as nature, buildings, vehicles, people or others in urban scenes.

In the context of precision agriculture, with scanning distances of less than 4,000 m and with variants of use according to their displacement capacity as fixed, land mobile and air mobile, more affordable multi-echo LiDAR can be used to assess crop growth, either by giving information on the “Leaf Area Index” or by segmenting the “crop canopy” of the soil, thus estimating the height of the crop. This idea implies engineering challenges related to the development of a three-dimensional reconstruction and mapping system that can be used for information gathering and for the study of elements of analysis and classification of point clouds.