Over the past five months, the integration of the Goizper mock-up has been taking place at STAM laboratory in Genova, Italy. The mock-up has the main objective of being a prototype of the final demonstrator for the assembly of rotary tables before its installation in the industrial environment at Goizper factory floor in the Basque Country, Spain.
At STAM laboratory a square area of four by four meters has been reserved for the integration of the mock-up. This area is delimited by a metal structure about 2.5 meters high on which four RGBD cameras are mounted for monitoring the working environment and in its center the working station for the collaborative assembly of the rotary tables has been installed.
The working station consists mainly of:
- A hydraulic table on which the rotary table to be assembled is positioned
- A collaborative robotic arm equipped with a screwdriving tool for the automation of the bolts tightening process
- A servo motor for the automatic shaft rotation
- An HMI for human-robot communication
The first activities carried out regarded the hardware integration of the components, including the installation of a control cabinet that houses the controllers, drivers and all electrical materials and wiring. Once the hardware integration and calibration activities were completed, it was possible to proceed with the integration of the Sharework software modules.
First scenario: rotary table bolts tightening
The objective of this first scenario is the collaborative tightening of the rotary table bolts: the worker inserts the bolts into the holes, the system recognises where the bolts have been inserted and the robot tightens them avoiding the collision with the operator.
For this scenario additional hardware components are required: a camera that captures the rotary table from above and additional lights to ensure good and uniform illumination of the table. After testing different configurations, it was decided to mount the camera directly on the robotic arm, in order to have it as close as possible to the table, thus increasing the resolution of the acquired images, and to have the possibility of self-centering the camera with the rotary table.
As foreseen by the ShareWork methodology, among the 14 software modules developed in the project, the most useful ones for the purposes of this use-case have been selected. In particular, in this scenario five Sharework modules play a significant role.
The Environment Cognition module allows to recognise which bolts have been inserted on the rotary table holes and to estimate their position. Additionally, thanks to this module, it is possible to extract the center of the rotary table from the images, allowing to self-center the camera on the same axis.
Then, the Task Planning module manages the information about the inserted bolts and, considering the tasks outlined in the Knowledge Base module, dispatches the commands for robot movement. The robot motion is handled by the Robot Motion Planner module which selects the best trajectory to reach the desired bolts avoiding the collision with the human operator, whose location is provided by the Human Tracking module.
STAM, as required by its role as Goizper’s technical coach, is adapting, integrating and tuning all these modules in the Goizper mock-up, also developing a dedicated procedure for the tightening of the bolts, once the tool is correctly positioned above them.
Second scenario: cam followers’ visual inspection
This second scenario aims to facilitate the visual inspection of the rotary table cam followers by controlling the automatic rotation of the shaft with arms gestures or HMI commands.
Regarding the first method, four different arm gestures have been defined to control the speed level and the direction of rotation of the shaft. Thanks to the Human Tracking module, it is possible to continuously track the location of the operator which is analysed by the Primitives Learning module to recognise if the worker is performing one of the predefined gestures.
Regarding the HMI commands, the Human-System and System-Human Communication module has created an interface that allows both to give rotation commands and to provide the operator with valuable visual information on how cam-followers are moving.
Then, thanks to a procedure developed by STAM, it is possible to monitor the commands received by both modules (gestures or HMI) and control the servomotor accordingly to rotate the shaft as requested by the operator.
What comes next?
In the next couple of months, the mock-up will be fully completed and some tests to assess the performances of the modules will be performed. In December 2021, the mock-up will be transferred to Goizper facilities, in order to integrate it in the real industrial environment. Once at Goizper, it will be possible to train operators on how to interact with the system and perform the new collaborative process of assembling rotary tables.
Project Engineer at STAM S.r.l.
She received the Master Degree in Robotics Engineering at the University of Genova, Italy, in 2020. During her studies she was involved in several projects where she experienced different aspects of robotics, including robot programming, computer vision, machine learning, control of robots, software architectures. In 2020, she started working as Project Engineer in STAM, an Italian engineering SME very active in the European Research and Development frame through the participation to several H2020 projects. Francesca is currently involved in the undertaking of technical activities from different R&D projects in the field of robotics and mechatronics.
STAM is a private engineering company based in Genova, Italy, providing engineering services to industries. Specialized in design and manufacturing of innovative robotics and mechatronics mechanical systems, STAM leads the system conceptualization and use case definition and is responsible for the demonstrator integration and monitoring.