Cooperative Modular Manipulation with Numerous Cable-Driven Robots for Assistive Construction and Gap Crossing

Kevin Murphy, Kevin Murphy 0006, Joao C. V. Soares, Justin K. Yim, Dustin Nottage, Ahmet Soylemezoglu, João RamosIROS 2024 · DOI

TL;DR

This paper introduces Co3MaNDR, a novel modular robotic system using multiple cable-driven robots to collaboratively manipulate heavy payloads. It demonstrates a new approach to force control and human-robot interaction, enabling a single operator to safely and efficiently assist in tasks like construction and gap crossing with significantly reduced physical effort.

What Problem Does This Paper Address?

Military operations often require soldiers to cross difficult terrain, such as rivers or canyons, to reach objectives. This typically involves constructing temporary bridges, a process that is currently dangerous, labor-intensive, and requires specialized heavy machinery. The existing methods are slow, logistically complex, and pose significant risks in hazardous or contested environments. The paper aims to address these challenges by developing a robotic system that can assist in these tasks, making them faster, safer, and more efficient, even for a single operator.

Methods

The core of Co3MaNDR is its modular design, composed of multiple Cable-Driven Parallel Robots (CDPRs) that collectively control a central payload. Key methodological contributions include:

  • Modular and Reconfigurable Design: The system uses multiple independent robotic modules that can be easily rearranged and scaled to match different payload sizes and operational requirements. This allows for flexibility and adaptability in diverse environments.
  • Payload-Centric Force Control: Instead of traditional position or velocity control, Co3MaNDR focuses on controlling the net wrench (force and torque) applied to the payload. This is crucial for interacting with unknown environments and enabling compliant, safe operations.
  • Proprioceptive Actuators: The system utilizes proprioceptive actuators, which are capable of high-bandwidth force control, essential for dynamic motions and precise force regulation.
  • Quadratic Programming (QP) Optimization: To manage the complexity of multiple cables and potential over-constraint, a QP-based controller is used to determine optimal cable tensions, minimizing deviations from the desired wrench while respecting tension bounds.
  • Human-in-the-Loop Force Amplification: The system is designed to allow a human operator to physically guide the payload. By leveraging force control and transparency, the system amplifies the operator's applied forces, significantly reducing the physical effort required to move heavy objects.

Datasets

Not discussed in the paper. The experiments were conducted using a custom-built Co3MaNDR system and custom-designed experimental setups.

Key Insights and Ideas

  • Force Control is Key for Human-Robot Collaboration: The paper strongly emphasizes that focusing on force control, rather than just position, is critical for enabling intuitive and safe human-robot collaboration, especially when dealing with heavy payloads or unknown environments.
  • Modularity Enables Scalability and Adaptability: The modular nature of Co3MaNDR is a significant advantage, allowing the system to be scaled up for heavier loads by simply adding more modules, and reconfigured for different tasks or environments.
  • Force Amplification Significantly Reduces Operator Effort: The ability for a single operator to manipulate a heavy payload with minimal physical exertion (averaging only 14.5% of the payload's weight) is a major breakthrough for assistive robotics.
  • CDPRs are Well-Suited for Large Workspace Manipulation: The inherent characteristics of CDPRs, such as their large workspace and high payload-to-mass ratio, make them an excellent choice for applications like bridge construction.

Key Results

  • Successful Payload Manipulation: The Co3MaNDR system demonstrated the ability to manipulate a 27.2 kg (60 lb) payload.
  • Reduced Operator Effort: A human operator was able to manipulate the payload with an average applied force of only 14.5% of its weight, and a maximum of 39.4%.
  • Scalability Demonstrated: Experiments with two and four modules showed that the average tension required per module was halved when using four modules, indicating scalability.
  • Fast Response Times: Module step responses had rise times under 75 ms, and the QP controller converged in 2 ms, running at 500 Hz, which is fast enough for human-assisted operations.
  • Accurate Trajectory Following: The system achieved desired macroscopic payload trajectories with a maximum error of 13.7 cm over a 4x2 meter workspace.

Limitations and Future Work

  • Limitations:
    • Reduced Lifting Capacity Compared to Traditional Systems: The prototype's lifting capacity is limited compared to motors with high gear ratios, though the paper suggests this can be overcome by adding more modules.
  • Cable Sag and Elongation at Larger Scales: For significantly heavier payloads, larger and thicker cables would be required, introducing issues like sag, bending limitations, and elongation that are not fully addressed by the current model.
  • Actuator Idealization: The controller assumes ideal torque sources, and deviations from this ideal (e.g., friction, damping) can affect performance, especially with less ideal actuators.
  • Static Operation: The current experiments are conducted with static modules, limiting field utility for tasks requiring continuous movement of the workspace.
  • Future Work:
    • Controlling Additional Degrees of Freedom (DoF): Expanding control to include payload position and orientation for more complex manipulation tasks.
  • Module Mobility: Developing mobile CDPRs to enable transportation of payloads across larger distances and more dynamic environments.
  • Incorporating Vision Systems: Integrating vision for enhanced perception and autonomous operation.
  • Advanced Actuation Approaches: Exploring different actuator types and their impact on system performance.
  • Full-Scale Design and Testing: Scaling up the system to handle heavier payloads and more demanding real-world scenarios.

Summary updated

BibTeX
@inproceedings{murphy2024cooperative,
  author = {Kevin Murphy and Kevin Murphy and Joao C. V. Soares and Justin K. Yim and Dustin Nottage and Ahmet Soylemezoglu and João Ramos},
  title = {{Cooperative Modular Manipulation with Numerous Cable-Driven Robots for Assistive Construction and Gap Crossing}},
  booktitle = {IEEE/RSJ International Conference on Intelligent Robots and Systems},
  year = {2024},
  doi = {10.1109/iros58592.2024.10801616},
}
Figure