Recently TIAGo has been showing new abilities in events such as IROS 2015 Conference or AYRI11 Forum. Some of them were developed as part of Factoy in a Day, an European project framed in the EU Seventh Framework programme in which PAL Robotics collaborates since 2013. Factory in a Day wants to optimize resources on little product lines in factories that produce short-run.
What is the purpose of this EU FP7 project? Factory in a Day’s goal is to reduce the system integration time of a robotics solution to one single day. The project is focused on making profitable the automatization of temporary product lines for SMEs that constantly need to modify or change the production, and find it difficult to capitalize the investment.
Robotics have a big role on this: this futuristic product lines need very simple robots that can be quickly modified and rapidly adjust to the new needs. Factory in a Day’s aim requires a big simplification of the machinery to make it possible. This is applied to the software too: it must be compatible with any platform and easy to self-calibrate.
The implementation process of a Factory in a Day product line would be the following:
First, there is an analysis of the workflow of the factory and the tasks that robots could perform in a short batch production are defined.
Second, the specific components are designed and 3D printed with Additive Manufacturing. Those parts are attached to highly adaptive gripper modules or to parts of the robots.
Third, the robots are shipped to the factory and afterwards, all robots and systems are deployed in about two hours. All the machines are connected to the same software.
Fourth, the robots are configured and learn the actions they have to do, with a minimal information required. This process aims to be accomplished in only one day. The idea is that robots do most of the repetitive work, and that humans focus on the remaining hard-to-automate tasks.
Easy and fast robotic learning
One important thing is to find ways in which robots can easily learn the task to execute so that the human co-workers can “program” them. Some of the 15 European partners are working on that, finding new ways to make humans cooperate with robots.
One example is the “learning-by-demonstration” application developed by PAL Robotics, performed by TIAGo on min. 1:31 of the video below. The user only has to teach the robot to perform motions moving TIAGo’s arm while it is on gravity compensation mode. Then, the robot learns it and is able to just repeat it as many times as wanted. The ability to change mode of operation for a robot, for instance switching from position control to effort control on the fly, has been developed inside Factory in a Day project and it is part of ros_control, an open source framework for hardware abstraction layer and control.
Another partner, the Institute for Cognitive Systems (Technical University of Munich), has done a research on another learning method for robots. On this video they show the steps to make the robots learn from a YouTube video how to cook some recipes – including REEM-C!
Safe robots at Factory in a Day
To turn Factory in a Day into reality, a precise control of the robots must be ensured. Workers and robots will collaborate together on the same space, performing joint tasks and sharing tools, without fences. Therefore, part of the efforts are put on reinforcing the safety of the workspace.
A development that PAL Robotics has achieved for Factory in a Day is a safety application to protect both the robot and the user, or any other person around. PAL Robotics implemented it to TIAGo: when detecting a collision that blocks the arm movement, instead of keeping pushing, TIAGo would stop it and automatically switch from position control to current control (min. 1:59 of the video). Then the robot would be in gravity compensation mode, which means that the user could move the arm applying a minimum effort on it, while the robot will hold its position.