INESCOP is researching into the robotisation of shoe sole demoulding
In the framework of the SOFTMANBOT project, INESCOP is teaching the robot to execute the movements for the demoulding of soles, learned from human operators, and then to improve its skills by means of learning techniques.
As we enter 2021, it is a good time to recapitulate INESCOP's participation in the SOFTMANBOT project. As an integration partner, INESCOP is involved in one of the four use cases: the demoulding of PLASTINHER’s injected shoe soles. Our focus is to help the development of the robotic workcell and to guarantee that it is proven and complies with the industrial requirements defined early in the project.
During the last year INESCOP has been continuously involved in various tasks with other partners, such as the development of the robotic gripper, the integration of sensors and the carrying out of first simulation tests of the robotic workcell. Currently, we are developing technologies for data acquisition of the demoulding task, both to help to determine the best grasping points for the sole and to determine the necessary path to extract it from the mould.
The idea of INESCOP’s robotic team is to use a combination of 3D vision information and an inertial measurement unit (IMU) to record and tag this data while the operator performs the task the usual way in his/her daily routine. We will analyse this data to better understand the task from a different source of knowledge and to obtain quantitative results from the different extractions carried out by humans.
In this context, the development of technologies to acquire data from the operator’s work is of great interest, because it will help to narrow down the high complexity of different skills needed to perform the task, such as the initial pull, the dexterity for continuously adapting the force, or the ability to react to different object deformations. Our approach is to use the tagged human-operator data to teach the robot by demonstration and then to further improve its skills by trial and error.
Stay tuned for more upcoming news on this exciting project!