Having a grip on things with “AeonDroid” BMBF-funded project aims to teach robots how to grip
Complex gripping tasks such as the handling of door handles have not been solvable with robot systems until now due to the high programming effort. Technische Universität Braunschweig and the start-up Aeon Robotics are now jointly researching a robot platform that allows robots to learn gripping movements from humans. This is intended to automate cleaning and maintenance work, for example.
The Institute of Robotics and Process Control (IRP) at TU Braunschweig and the Braunschweig-based robotics start-up Aeon Robotics are jointly researching the automation of cleaning and maintenance work. In the research project “AeonDroid”, both want to make it possible for service robots to learn cleaning and maintenance tasks directly from humans for the first time. To this end, they are receiving project funding totalling around 700,000 euros as part of the START-interaktiv funding programme of the Federal Ministry of Education and Research (BMBF). The funded project was one of five selected from a total of 72 applications by the BMBF and the project management organisation.
Hand movements modelled on humans
In the project, a mobile robot platform is being developed that has robotic hands and arms modelled on humans and is based on the innovative HandEffector technology from Aeon Robotics. The system will be operated intuitively via a hand control without any programming knowledge. The robotic hands have an AI-supported depth-sensitive sense of force with direct force transmission to the human user. This means that cleaning tasks can be controlled with force feedback and intuitively taught by demonstrating the task.
The depth-sensitive sense of force of the “AeonDroid” also allows deformable and sensitive objects such as sponges and cleaning agents to be gripped and moved safely in order to avoid damage to the objects to be cleaned. Due to the high variability of the human models, the development of adapted force and movement primitives for this specific application has a considerable need for research, which is being worked on at TU Braunschweig by Professor Jochen Steil and his team. For example, databases of gripping movements, gripping forces and work sequences are to be created in order to recognise patterns within the movement sequences by means of neural networks and to independently adapt them for new utensils and work areas with AI support.
Interactive technologies for health and quality of life
With the START-interaktiv programme, the BMBF aims to strengthen the innovation potential of start-ups in cutting-edge research on interactive technologies for health and quality of life. To this end, the opportunities for founding start-ups of suitable research teams are to be promoted at an early stage and in a targeted manner at universities. Funding will also be provided for start-ups that have already been founded in high-risk pre-competitive research. The object of the funding is high-risk research and pre-competitive development projects that are cross-technology and application-oriented and have a direct positive influence on the innovative capacity and expected competitive opportunities of the start-ups involved.
Network coordinator: Aeon Robotics GmbH
Aeon Robotics GmbH is a start-up with a focus on modular robotics, artificial intelligence, drive technology and smart components. The company started in 2021 as a spin-off of TU Braunschweig. With the HandEffector robot system, Aeon Robotics offers a solution that makes it possible for the first time to intuitively teach gripping forces and gripping movements to a robot system without prior knowledge of robotics or programming. The robotic hand with AI-supported, depth-sensitive force sense in combination with force feedback control allows direct force transmission to the human user. The integrated gripping and joint force measurement without additional sensors is cost-effective and, in combination with the force feedback unit, enables the automated gripping of sensitive objects. In this way, processes can be automated whose automation was previously not possible.
Partner: Institute of Robotics and Process Control (IRP)
The long-established IRP represents robotics, especially in connection with machine learning, in its basic teaching and research, with a focus on robot learning, human-machine interaction, process modelling of complex systems and the application of artificial intelligence in these areas. Activities also include the integration of robotics into production systems and generally in the context of Industry 4.0 and the digital transformation of society, with Professor Jochen Steil advising various institutions on the future of work and the role of robot ethics and being active in the national platform “Lernende Systeme” (Learning Systems). The IRP is a member of the Wissenschaftliche Gesellschaft für Montage, Handhabung und Industrierobotik (MHI), runs transfer projects, and cooperates with the Zukunftsallianz Maschinenbau Norddeutschland.