Synergy-based model predictive interaction control for robotic hand-arm systems
Third party funded individual grant
Start date : 01.01.2026
End date : 31.12.2028
Project details
Short description
Robotic hands are often motivated by the fact that they can be used to perform a wide range of manipulation tasks. The extension to a robotic hand-arm system (e.g., as part of a humanoid robot) makes it possible to significantly reduce the workload of people in shared workspaces. Clear examples of this can be found both in industrial manufacturing processes (pick and place, assembly, machining, etc.) and in everyday human life (opening doors, cutting food, handling objects, etc.). In addition to complicated movements, such tasks require the precise adjustment of forces. Nevertheless, they are usually performed by humans or various highly specialized machines. Alternatively, the environment could be adapted to the technology, but this is hardly practicable. In addition to safety, a particular challenge in shared workspaces is the acceptance of the robot system by the people present. An important aspect of this is that people must be able to recognize and anticipate the robot's movements intuitively.
The research project aims to develop a synergy-based model predictive control (MPC) for robotic hand-arm systems that is integrated into a modular control architecture. The challenges of numerous actuated degrees of freedom are considered and the methods for nullspace control are co-developed. With the help of a model-predictive approach, the aim is to achieve a modular architecture that also integrates task planning. Various approaches are being developed to efficiently integrate linear and non-linear kinematic synergies into a modular MPC formulation. Therein, the synergies are intended to reduce the complexity and thus the computation effort on the one hand and to realize human-like motions on the other. This approach also aims to increase the human acceptance of such systems.
