Life on Earth is inevitably tangled to biodiversity preservation. Habitat monitoring is the main tool that human beings have to avoid biodiversity loss. Monitoring consists in the repetitive and continuous evaluation of the habitat conservation status. However, this operation is expensive from both a time and a cost perspective.

Robotics could enhance human monitoring capabilities, and this work makes a first step towards this direction, focusing on the grassland habitat “6210 - Semi-natural grasslands and scrubland facies on calcareous substrates”, which is among the most species-rich habitat in Europe. Specifically, we openly share data collected during monitoring tasks by the quadrupedal robot ANYmal C. The dataset creation has been performed by a team composed of robotic engineers and plant scientists, and it contains information on the robot state, videos, and images acquired during surveys, and a collection of videos and pictures about two typical and one early warning species of habitat 6210.

The data are available open access via the repository Zenodo ( https://doi.org/10.5281/zenodo.7385369 ), and their goal is to push forward robotic habitat monitoring research. From one point of view, computer scientists could employ data gathered by the robot to train or validate their classification algorithms. Alternatively, robot state information or point clouds could be used by engineers to test their methodologies. Finally, plant scientists could exploit these data to propose new metrics or simply check robot performance in monitoring activities.