Cooperative manipulation with dual-arm robots at the payload limit

• Applicant: Dr.-Ing. Andreas Völz
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 493539176

Dual-arm robots offer a high potential for automation technology, as they can be used to implement tasks that are not possible with one arm alone. This includes in particular the manipulation of large or heavy objects that exceed the payload of a single arm. Illustrative examples are the movement of beverage crates, long boards or pipes, which are also preferably grasped by humans with both hands.However, cooperative manipulation is particularly challenging, because both arms and the grasped object form a closed kinematic chain. The corresponding constraints reduce the number of degrees of freedom and must be taken into account on the levels of control, trajectory and path planning. Conversely, the system has the advantage that the load can be flexibly distributed on both arms due to the redundant actuators. This is especially crucial for heavy objects, since it is the only way to comply with actuator torque constraints. The first goal of the research project is therefore the development of a dynamic load distribution that explicitly takes actuator constraints into account and is thus suited for high payloads. To this end, an optimization-based approach is pursued with a focus on efficiency and real-time capability.Moreover, this load distribution must be taken into account on all system levels, since otherwise large payloads can lead to the situation that no admissible trajectory can be computed for a path or that a trajectory is not executable by the controller. Consequently, the second goal is the consistent consideration of the dynamic load distribution. On the control level, this includes not only the isolated solution of the optimal load distribution in each sampling step, but also the approach of a forward-looking model predictive controller. For trajectory planning, on the one hand, a time-optimal trajectory generation with subordinate solution of the dynamic load distribution and, on the other hand, the extension of the model predictive controller to a predictive path-following controller shall be investigated. Furthermore, a path planner for dual-arm robots will be developed for the first time, which explicitly considers the payload and can be extended in a modular manner to take collisions as well as the additional degrees of freedom of a mobile base into account.The third goal is the extensive experimental validation of the control, trajectory and path planning methods in order to practically demonstrate the potential of cooperative manipulation with dual-arm robots. For this purpose, a mobile dual-arm robot with additional motion capturing system from the DFG major instrumentation proposal 438833210 is available at the Chair of Automatic Control. In particular, movements with large and heavy objects shall be performed, whose mass is in the order of magnitude of the combined maximum payload of both arms.

• DFG Programme: Research Grants