COLECT: COntrol and Learning of Contact Transitions
The project aims to develop contact detection and force control methodologies that can be leveraged across different industries (e.g., industrial assembly and medical robotics) to ensure safe and robust control of tasks involving compliant robot control. The latest state of the art will be leveraged and new methods will be developed to handle challenging problems from our industry partners.
The COLECT project
The project aims to pioneer the development of secure and resilient methods for compliant robot control that can be used in a wide variety of contexts, where such control methods are not currently utilised or where the latest cutting-edge research techniques have yet to be implemented in industry settings.
Our industry partners involved in the project will gain valuable insights into innovative technologies and algorithms that can be integrated into their specific cases requiring compliant control of robotic manipulators. Our emphasis will be on addressing intricate challenges, such as minimising contact forces, seamlessly transitioning from unconstrained to constrained movements, and controlling soft surfaces like human tissue.
As a result, project partners can anticipate receiving new ideas and solutions on how to apply force control methods to problems, where they were unable to or limited in applying before.
Background
The control of robotic manipulators for tasks that are force-sensitive is a pertinent issue across various applications.
In the manufacturing sector, several processes, such as deburring or grinding, make it necessary that the robot applies a consistent and precisely controlled force to a workpiece.
In the medical field, robots used for tasks like scanning procedures often need to establish gentle and continuous contact with a probe on a patient's body while performing scanning movements.
In both scenarios, it is of utmost importance that the contact is initiated smoothly and without surpassing specified force limits. Throughout this process, the force should be maintained within predefined safety thresholds, and the robot's control system must adapt its compliance to the surrounding environment. Achieving these behaviors is a complex task, often demanding a deep understanding of both the specific task domain and robotic control. In many instances, the expertise of a highly trained operator cannot be seamlessly transferred to the robot during the operation.
In the COLECT project, we will develop an end-to-end framework for setting up force-sensitive tasks leveraging robust contact estimation, control and learning from demonstration.
Purpose / Vision
The COLECT project aims to develop secure and resilient methods for compliant robot control that can be universally applied, including in scenarios where such control methods are not currently utilised or where the latest cutting-edge research techniques have yet to be implemented in industry settings.
Our industry partners involved in the project will gain valuable insights into innovative technologies and algorithms that can be integrated into their specific cases requiring compliant control of robotic manipulators. Our emphasis will be on addressing intricate challenges, such as the minimisation of contact forces, seamless transitions from unconstrained to constrained movements, and control on soft surfaces like human tissue. As a result, project partners can anticipate receiving new ideas and solutions for the application of force control methods to problems that were previously difficult or impossible to address.
Expected results
Academic partners will develop research infrastructure in robot compliant control and learning from demonstration, leading to future research projects and publications.
Success may lead to an extension or spin-off projects, with funding options such as Grand Solutions or co-financing with partners like the I.4.0 Lab at SDU. A concluding workshop will share project insights and explore applications in different contexts.
Project participants
The University of Southern Denmark (SDU)
Technical University of Denmark (DTU)
Ropca
Nordbo Robotics
Odense Robotics
Funding
The Ministry of Higher Education and Science has financed this project with 500.000 DKK.
Start and finish
The project runs from January 2023 to March 31 2024.
Contact
Do you want to get involved or hear more? Reach out to Ole Georg Andersen.
Contact us to learn more
Curious to find out more about the project and how you can get involved? Get in touch with Ole.
Knowledge-based innovation and the Fehmarn Belt project