The research paper discusses the use of robots in tomato greenhouses to automate labor-intensive tasks like harvesting and de-leafing. The main challenge is for robots to accurately perceive plant nodes that are often hidden or occluded by other plant parts. The study proposes a local next-best-view (NBV) planning task where the robot plans an efficient set of camera viewpoints to overcome occlusion and improve perception accuracy. A gradient-based NBV planner is proposed, which has proven to handle occlusions and improve the 3D reconstruction and position estimation of nodes more efficiently than previous methods.
Publication date: 29 Nov 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2311.16759
