Periodic Reporting for period 2 - DrapeBot (Collaborative draping of carbon fiber parts)
Reporting period: 2022-07-01 to 2023-12-31
Draping is the process of placing soft and flexible patches of textile material (in this case carbon or glass fiber fabric) on a 3D shape during the manufacturing of carbon fiber composite parts. Draping remains very important as an alternative to automated fiber placement and tape laying due to its flexibility and its compatibility with a wide variety of fabric materials. As a result, 30% of aerospace composite parts are still produced through draping, while many automotive and almost all marine structures are made this way. At present, most draping is done manually, with only 5% of aerospace composite parts using some automated draping process.
Robotic draping has made significant progress with respect to handling of large patches of material and accuracy of the draping, but there are a number of challenges that still remain to be addressed:
• Many parts include surface elements of high curvature, whose automatic draping is beyond the capabilities of robot draping systems.
• Draping is a precision task, with tolerances going down to ±2 mm for positioning and ±3° for fiber orientation. .
• Large patches of materials of varying shape need to be handled, which can be up to 10m in length..
•The robotic systems involved in draping are usually large scale robots and interaction methods are needed in which such robots can safely and efficiently collaborate with humans.
In order to facilitate efficient human-robot collaboration in complex draping processes, such as those defined in the use cases, the DRAPEBOT project aims at the following technical objectives:
(A) To develop an efficient human-robot collaborative environment, in which a robot and a human can collaborate safely during the transfer (pick&place) of the fabric and cooperate during the draping of the material. The focus will be on AI-based real-time sensing and action-planning capabilities to achieve an increase in productivity and to capture the complexity of human-robot interaction.
(B) To design and build gripper systems with additional instrumentation to help during the interaction with the human worker and to ensure the accuracy of the draping process in terms of positioning, wrinkles and fibre orientation of single patches. AI-based real-time control will be integrated to ensure proper draping results.
(C) To create an environment in which a human worker can safely collaborate with the large robotic systems that are required to handle the materials. Trust and usability are key non-technical aspects that will be addressed in the DRAPEBOT project, especially when AI-based methods are involved.
Robotic draping has made significant progress with respect to handling of large patches of material and accuracy of the draping, but there are a number of challenges that still remain to be addressed:
• Many parts include surface elements of high curvature, whose automatic draping is beyond the capabilities of robot draping systems.
• Draping is a precision task, with tolerances going down to ±2 mm for positioning and ±3° for fiber orientation. .
• Large patches of materials of varying shape need to be handled, which can be up to 10m in length..
•The robotic systems involved in draping are usually large scale robots and interaction methods are needed in which such robots can safely and efficiently collaborate with humans.
In order to facilitate efficient human-robot collaboration in complex draping processes, such as those defined in the use cases, the DRAPEBOT project aims at the following technical objectives:
(A) To develop an efficient human-robot collaborative environment, in which a robot and a human can collaborate safely during the transfer (pick&place) of the fabric and cooperate during the draping of the material. The focus will be on AI-based real-time sensing and action-planning capabilities to achieve an increase in productivity and to capture the complexity of human-robot interaction.
(B) To design and build gripper systems with additional instrumentation to help during the interaction with the human worker and to ensure the accuracy of the draping process in terms of positioning, wrinkles and fibre orientation of single patches. AI-based real-time control will be integrated to ensure proper draping results.
(C) To create an environment in which a human worker can safely collaborate with the large robotic systems that are required to handle the materials. Trust and usability are key non-technical aspects that will be addressed in the DRAPEBOT project, especially when AI-based methods are involved.
The project started with a deep analysis of use cases for draping of carbon fibre parts in aerospace, automotive and boat-building applications. This included the analysis of the single processing steps, the parts, their lay-up, the shape of the related moulds and any other requirements that are important in the respective domain. Based on this information the lay-up process was analysed and different categories of human-robot interaction were identified, including hand-over tasks, collaborative transport of the material and collaborative draping of the material. To address these use cases and to implement the different interaction modes, two robotic workcells called "TEZ" at DLR and "DrapeCell" at Profactor were designed and built. TEZ is exclusively dedicated to teh aerospaces use case and is mainly focuse on human-robot interaction, whereas DrapeCell will be used for the automotive and the boat-building use case and has a stronger focus on human-robot collaboration. After assembling the hardware, several software components were integrated, including the central node for running the whole process, the task and motion planning, the low-level control to adjust the robot's trajectories and human perception to recognize relevant human actions. Over a period of several months extended tests have taken place which provided input to improvements of the single software modules as well as of some of the hardware components. After a first integrated test in April 2023 the single modules were developed further to increase the smoothness of the human-robot collaboration and also to reduce the time that the robot needs to react to human actions. By December 2023 a demonstration of all interaction modes was possible on the example of the automotive use case. To demonstrate the capabilities and assess the performance a set of five plies was draped multiple times. Also, usability experiments have been done to record how trust between the human and the robot evolves during the collaboration. In general the feedback was that the interaction is quite natural and the human only has to adapt in a minimal way to the fact that his collaborator is a robot.
XThe main progress that the project intends so make, is to demonstrate that human-robot collaborative draping can be done in an efficient end economically feasible way. It will thus provide an important alternative to robot-only draping and human-only draping. It will fill a gap in the technological methods for draping and lay-up processed, by providing a human-centered method for draping. To facilitate this step forward the project will develop:
* Automated task and motion planning methods for human-robot collaborative processes
* New grippers, including instrumentation for monitoring of the draping process
* Human perception, action recognition and prediction to enable a smooth interaction between the robot and the human
* Low-level, real-time control methods based on model predictive control to facilitate efficient human-robot collaboration during complex tasks such as collaborative transport of larger material patches
* Live trust estimation that assesses the interaction between the human and the robot, to make the interaction more natural.
All of these elements will be integrated in two robotic workcells, which will then demonstrate a substantial impact on draping processes.
* DrapeBot will increase the potential for robotics in lay-up processes. Medium-size parts, for which collaborative draping is a technically and economically interesting approach, are still the a significant fraction of all parts. An analysis of the market indicates that there is a potential of 20.000 installations of robots for such tasks.
* The deployment risks of collaborative robotic systems will be reduced through the demonstrations in different application scenarios. These demonstrations and evaluations will include long-term tests in realistic environments of production lines for a total of 12 months in the project.
* DrapeBot will promote the use of robots in areas, where they are rarely used today. Aside from the draping process, the demonstration of large scale robots (payload 200kg, reach 3m) in a collaborative scenario will facilitate human-robot collaborations in many other areas, where larger robots are used.
* With its focus on efficency, the DrapeBot project will provide important input to standards development, especially in the area of robotic safety, which is currently hampering the installation of large scale collaborative robotic workcells.
* Automated task and motion planning methods for human-robot collaborative processes
* New grippers, including instrumentation for monitoring of the draping process
* Human perception, action recognition and prediction to enable a smooth interaction between the robot and the human
* Low-level, real-time control methods based on model predictive control to facilitate efficient human-robot collaboration during complex tasks such as collaborative transport of larger material patches
* Live trust estimation that assesses the interaction between the human and the robot, to make the interaction more natural.
All of these elements will be integrated in two robotic workcells, which will then demonstrate a substantial impact on draping processes.
* DrapeBot will increase the potential for robotics in lay-up processes. Medium-size parts, for which collaborative draping is a technically and economically interesting approach, are still the a significant fraction of all parts. An analysis of the market indicates that there is a potential of 20.000 installations of robots for such tasks.
* The deployment risks of collaborative robotic systems will be reduced through the demonstrations in different application scenarios. These demonstrations and evaluations will include long-term tests in realistic environments of production lines for a total of 12 months in the project.
* DrapeBot will promote the use of robots in areas, where they are rarely used today. Aside from the draping process, the demonstration of large scale robots (payload 200kg, reach 3m) in a collaborative scenario will facilitate human-robot collaborations in many other areas, where larger robots are used.
* With its focus on efficency, the DrapeBot project will provide important input to standards development, especially in the area of robotic safety, which is currently hampering the installation of large scale collaborative robotic workcells.