 |
CooPick Collaborative robot-robot-human interaction for fruit laying Depending on flexibility and capacity requirements, placing fruit on conveyors is performed either completely manually or fully automated in large plants. Affiliated to the process are quality control and final packaging. Against this background, large rationalization potentials for medium flexibility and capacity requirements can be identified by partial automation. Therefore, a robot-robot-human collaborative fruit placing system was developed by the BIBA together with the TU Kaiserslautern and the company MWZ for control engineering. The fruits are transported in a special conveyor belt, which is controlled by the main system, so that the velocity and position of the belt is known. A camera system detects the pose and defects of the fruits, and send this information to a robotic system with two robotic arms with specially adapted grippers to pick up the fruit without affecting its quality. Both arms work in collaboration with the employees at the conveyor belt to pick up the fruit and place it correctly on a tray, so that it can be properly packed. As a result of the project, the camera system was able to detect the position and rotation of different types of apples using state-of-the-art convolutional neural networks. Additionally, a near-infrared camera was used to detect bruises and surface defects in apples using different neural network architectures such as SSD (Single Shot Detection) and U-Nets. This way it was possible to perform a quality inspection for each apple. This information was sent to the robotic system using the robotic operative system framework, ROS. The robotic system was able to properly pack and sort the apples in collaboration with human workers. Contact persons: M. Lütjen   Funded by: BMWi Duration: 01.01.2018 - 15.04.2021 See project's publications List all projects |
