engleski

Pepper to fall: a perception method for sweet pepper robotic harvesting

In this paper we propose a robotic system for picking peppers in a structured robotic greenhouse environment. A commercially available robotic manipulator is equipped with an RGB-D camera used to detect a correct pose to grasp peppers. The detection algorithm uses the state- of-the-art pretrained CNN architecture. The system was trained using transfer learning on a synthetic dataset made with a 3D modeling software, Blender. Point cloud data are used to detect the pepper’s 6DOF pose through geometric model fitting, which is used to plan the manipulator motion. On top of that, a state machine is derived to control the system workflow. We report the results of a series of experiments conducted to test the precision and the robustness of detection, as well as the success rate of the harvesting procedure.

Robotic harvesting, Transfer learning, RGB-D, Convolutional neural networks, Depth camera, Sim2real

nije evidentirano