Fleets of heterogeneous autonomous robots

GPA has a broad expertise in developing fleets of heterogeneous autonomous agricultural robots. Over the years, the group has developed and worked with robots of various sizes and purposes: inspection, pesticide treatment, harvesting assistance, among others. To this end, GPA’s research has focused on vehicle automation, perception, sensor fusion and control systems for safe autonomous navigation; and on route planning and fleet supervision, generating optimal routes, taking into account factors such as distance to be traveled and consumption, preventing collisions and ensuring continuous monitoring.

Keywords: fleets of robots, autonomous navigation, route planning, fleet supervision, sensor fusion

3D reconstruction

One of the fields in which GPA has conducted research in recent years is the 3D reconstruction of crops. GPA has worked with various RGB-D cameras, LiDARs, and traditional cameras, using the data provided by these sensors to generate 3D models. GPA has successfully integrated these sensors on board various vehicles, enabling the automatic 3D reconstruction of entire crop rows. For this purpose, it employs techniques such as NeRF, utilizing neural networks to generate photorealistic 3D scenes. Furthermore, GPA has developed techniques to extract information from these models, which have been studied and validated in numerous studies, and which facilitate improved decision-making in crop management

Keywords: 3D reconstruction, RGB-D cameras, sensors on board vehicles, NeRF, measurement estimation