The programming of an industrial robot to perform a relatively simple repetitive task that only lasts for a few seconds, might take plenty of hours even from expert personnel. If complex geometries or synchronization with other peripheral systems is required, the programming can take days. This entails increased automation costs that are only justified in large industries with a steady production cycle. Focusing on SMEs in food production, there are frequent changes in the production line due to different batches of products coming from the same line. For example, in Unismack -which specializes in the production of healthy snacks- on a single line, many different types of product batches are produced in terms of packaging type and therefore the packaging is mainly performed manually due to high automation costs.
The aim of the project is to develop an innovative robot assistant to which a person can demonstrate kinesthetically complex repetitive or periodic tasks and the robot will learn to execute it with progressive automation. The project will be based on novel methods of learning by demonstration that will be developed to achieve effective human-robot collaboration. In addition, state of the art methods of machine vision with depth cameras will be used and will be developed to provide robot perception of the environment in conjunction with object grasping. In this way, usability, adaptability and safety will be achieved, enabling the robot to learn from the human how to perform various tasks. The result of the project will be a portable and flexible work cell with a robotic arm, which can be placed in a production line and can be programmed very easily and quickly for the task by the human, without requiring expert personnel.
The project will be implemented starting with the identification of the needs of the industry, the determination of the use-cases and the definition of the specifications of the robotic system. Then, solutions beyond state of the art will be developed to demonstrate human movement with kinesthetic guidance of the robot, automatic adaptation of the robot to changes of the environment using machine vision and safety of the cooperation by avoiding obstacles and distinguishing contacts of the robot with the environment. Finally, the overall system will be integrated (hardware and software), tested and evaluated in a real industrial environment during the production and packaging of organic food.