GRIPS and Light Cafe Conduct Demonstration Experiment of Automated Tomato Sorting System Utilizing Small Robots and AI
This demonstration experiment is being conducted at various facilities in Aichi Prefecture to promote the social implementation of service robots in Aichi Prefecture, sponsored by Aichi Robot Transformation (ARX). was broken The system configuration consists of two small robot arms DobotMagician, one conveyor belt, and a USB camera. It performs processing and appearance judgment. In addition, the judgment results are uploaded to the cloud server one by one, and the collected tomatoes are converted into data along with environmental information, making it possible to apply it to ensure the traceability of harvested agricultural products. The cloud system was developed by Light Cafe, and realizes the integration of the cloud server and the robot control system.
Hundreds of tomato images are taken in advance as a dataset for machine learning, and a CNN model is generated using them. For the algorithm, we added TensorFlow and Keras to the Python environment implemented on the RaspberryPi, and implemented RUTILEA's SDTest as an application. With a system configuration that combines open source technology, it is possible to realize an automatic sorting system using AI technology at a low cost. In addition, as a technology to "gently" grip tomatoes, Light Cafe's uniquely developed soft gripper has been adopted, realizing picking operations that can handle tomatoes of different shapes and weights. The shape of the gripper part was manufactured by a stereolithography 3D printer. In the future, based on this knowledge, we would like to contribute to the improvement of productivity by applying systems and elemental technologies with improved stability, durability, and speed to smart agriculture. GRIPS has already contributed to the proposal of low-cost automation by applying the small robot DobotMagician, which was also used in this demonstration experiment, to the production site. In the future, we will continue to propose automation in various industrial fields such as agriculture and medicine with partner companies through open innovation and ecosystems.
