A Decoupling Scheme for Force Control in Cooperative Multi-Robot Manipulation Tasks

Luca De Pascali, Sebastian Erhart, Luca Zaccarian, Francesco Biral, Sandra Hirche

TL;DR

This paper presents a novel control scheme that decouples the control of motion-inducing forces and internal forces in cooperative multi-robot manipulation. This allows for precise control over how robots manipulate an object and how they grip it, independently of each other, leading to more robust and versatile robotic systems.

What Problem Does This Paper Address?

Cooperative manipulation, where multiple robots work together to grasp and move an object, offers significant advantages in terms of payload capacity and dexterity. However, coordinating multiple robots introduces complexities, particularly in controlling the forces they exert. Two critical types of forces arise: motion-inducing forces, which are responsible for moving the object, and internal forces (or "squeezing" forces), which relate to how the robots grip the object. Uncontrolled internal forces can damage the object or the robots, while precisely controlling them is crucial for tasks requiring delicate manipulation or maintaining a stable grasp. Existing control methods often struggle to independently manage these two types of forces, leading to challenges in achieving both precise object motion and controlled grasping. This paper addresses the need for a control strategy that can decouple and independently manage these motion-inducing and internal forces in cooperative multi-robot systems.

Methods

The core of the paper's contribution lies in a novel decoupling control scheme. It builds upon a physically and mathematically consistent model of the interaction dynamics between the object and the manipulators. The approach leverages the concept of internal wrenches residing in the null space of the grasp matrix, meaning they don't affect the object's overall motion.

The proposed method involves:

  • Modeling Interaction Dynamics: The authors derive equations that precisely describe the forces and torques exchanged between the robots and the object, considering the system as a constrained multi-body system.
  • Impedance Control for Motion: An impedance control law is used to render the end-effectors compliant, allowing them to track desired object trajectories. This part is largely based on prior work.
  • Static Allocator for Internal Forces: A key innovation is a "static allocator" that independently controls the internal wrenches. This allocator ensures that the internal wrenches follow a desired time-varying reference, without interfering with the motion-inducing forces or the object's trajectory.
  • Decoupling Strategy: The control signal for internal wrenches is designed such that it lies within the null space of the grasp matrix, thus guaranteeing that it does not affect the object's motion. This effectively decouples the control of motion and internal forces.

Datasets

Not discussed in the paper. The paper focuses on theoretical development and simulation-based validation.

Key Insights and Ideas

  • Internal forces don't have to affect object motion: A fundamental insight is that internal forces, by definition, can be designed to lie in the null space of the grasp matrix, meaning they don't contribute to the object's translational or rotational acceleration. This is the theoretical basis for decoupling.
  • Decoupling is achievable: The paper demonstrates that it's possible to design a control scheme where the motion-inducing forces and internal forces can be controlled independently. This is a significant step towards more sophisticated cooperative manipulation.
  • Importance of a consistent dynamic model: The effectiveness of the decoupling scheme relies heavily on a precise and physically consistent model of the interaction dynamics between the robots and the object.
  • The "static allocator" as a key component: The introduction of a static allocator that specifically targets internal wrenches, ensuring they meet desired setpoints without disturbing motion control, is a novel contribution.

Key Results

The paper presents a theoretical framework and validates it through a numerical simulation.

  • Successful Decoupling: The simulation results (shown in Figures 6 and 7) demonstrate that the proposed control scheme successfully achieves independent control of internal wrenches. The actual internal wrenches closely track the desired time-varying reference.
  • Unperturbed Motion Control: Crucially, the control of internal wrenches does not negatively impact the object's trajectory tracking. Figure 7 shows that the object's position, velocity, and acceleration (both translational and rotational) accurately follow the desired references, even with the application of internal wrenches and external disturbances.
  • Robustness to Disturbances: The system shows good performance in the presence of an external disturbance wrench acting on the object.

Limitations and Future Work

  • Simulation-based validation: The results are currently demonstrated through numerical simulations. The authors acknowledge that experimental evaluation is a necessary next step.
  • Assumptions: The model assumes rigid grasps and a rigid object. Real-world scenarios might involve deformable objects or compliant grasps, which would require extensions to the model.
  • Future Work: The authors explicitly state that future work will involve an experimental evaluation of the proposed technique. This will be crucial for validating the theoretical findings in a real-world robotic system.

Summary updated

Abstract

International audience
BibTeX
@inproceedings{pascali2021decoupling,
  author = {Luca De Pascali and Sebastian Erhart and Luca Zaccarian and Francesco Biral and Sandra Hirche},
  title = {{A Decoupling Scheme for Force Control in Cooperative Multi-Robot Manipulation Tasks}},
  booktitle = {AMC},
  year = {2021},
  doi = {10.1109/amc51637.2022.9729263},
}
Figure