The sensing work package will deliver a robust pragmatic computer vision package to localize mature yellow and red fruits and its surrounding obstacles. An early milestone will ensure that a basic sensing framework can be tested and thereafter continuously improved after the first few months of the project start. This early stage will be based on results obtained in the CROPS project. For advanced sensing, a Lightfield camera will be added as primary sensing hardware on the end-effector. Basic software will be provided in an early stage and is based on CROPS algorithms adapted to data collected with the new SWEEPER system.  Necessary adjustments to it are made to meet optical and greenhouse’s constraints. Additional advanced computer vision algorithms will be developed to ensure progress in terms of robustness, accuracy and detection rate. To improve detection rates, sensing from the gripper will be performed from multiple viewpoints. Part of these algorithms will be elaborated on the results from the CROPS project.

Workpackage leader

Ben-Gurion University of the Negev – BGU (IL)