A Soft Supernumerary Robotic Limb with Fiber-Reinforced Actuators

Jiajun Xu, Tianyi Zhang, Kaizhen Huang, Mengcheng Zhao, Xuyan Hou, Youfu Li

Introduction

Supernumerary robotic limbs (SRLs) are wearable robots designed to provide humans with an "extra hand" for industrial assembly, overhead tasks, or daily assistance. Currently, the field is divided between rigid and soft systems. Rigid SRLs offer power and precision but are heavy, bulky, and pose safety risks during human interaction. Conversely, soft SRLs are lightweight and inherently safe but often suffer from limited "monotonous" motion and are difficult to model and control accurately.

This research bridges the gap by creating a soft SRL that combines biological versatility with mathematical rigor. By utilizing fiber-reinforced actuators (FRAs) and a novel optimization framework, this wearable extra arm achieves a remarkable payload-to-weight ratio of 15 and high-precision control, outperforming existing alternatives in both safety and maneuverability.

Overview of the wearable soft supernumerary robotic limb system.

Design and Methodology

The team took inspiration from the "muscular hydrostats" of octopus arms—complex arrangements of transverse, longitudinal, and oblique muscle fibers. They replicated this using Fiber-Reinforced Actuators (FRAs): hollow silicone tubes (DragonSkin 30) wrapped in high-strength Kevlar fibers.

Motion Programming and Modeling

By varying the angle and pattern of the fiber windings, a single actuator can be "programmed" to perform specific deformations—such as extension, expansion, bending, or twisting—when pressurized. This multi-modal capability allows the limb to navigate complex 3D environments that were previously the sole domain of rigid robots.

Motion programming of fiber-reinforced actuators: (a) extension, (b) expansion, (c) bending, and (d) twisting.

To achieve precision, the researchers established a continuum mechanics model using a Neo-Hookean material framework to predict how input air pressure translates into specific deformations. This was validated using Finite Element Modeling (FEM), showing less than a 3% discrepancy between theory and simulation.

Optimization Framework

Rather than building a general-purpose limb, the authors propose an "inverse design" framework where the task defines the limb. Two optimization layers were developed:

  • Trajectory Optimization: Uses Particle Swarm Optimization (PSO) to determine the limb's geometry and pressure sequences required to match a desired path.
  • Payload Optimization: Uses the BFGS algorithm to determine the best attachment points and pressures to maximize lifting capacity.

Key Results

The limb demonstrated exceptional performance across several metrics:

  • Precision: The system achieved a Root-Mean-Square Error (RMSE) of less than 0.4 mm in trajectory tracking experiments, a significant feat for soft, pneumatic systems.
  • Workspace: The limb demonstrated a maximal reach of 0.81 m and a vertical range of 1.20 m, comparable to rigid counterparts.
  • Payload Capacity: The limb achieved a payload-to-weight ratio of 15 (a 2kg limb lifting a significant load), largely due to the fiber-reinforcement preventing inefficient material expansion.
  • Usability: In pick-and-place trials, users with no prior experience showed a rapid learning curve, reducing task completion time from ~50 seconds to ~10 seconds over just five attempts.

Limitations and Future Work

While the limb is technically impressive, the authors acknowledge several hurdles for future development. Currently, the model treats the limb in isolation; future work must account for the dynamic interaction between the soft limb and the moving human body. Additionally, the researchers aim to integrate EMG (muscle signals) or EEG (brain signals) to allow the limb to move intuitively based on the wearer's intent, rather than following pre-planned trajectories. Finally, improving modeling for "nonsmooth contact"—where limb segments interact with each other or external obstacles—remains a priority.

Summary updated

Abstract

Supernumerary robotic limbs (SRLs) have great potentials to assist human in daily activities and industrial manufacturing by providing extra limbs. However, current SRLs have heavy and rigid structures that may threaten the operator safety; moreover, their limited degrees of freedom and movement modes are not suitable for complicated tasks. Although soft SRLs have exhibited advantages in structure compliance and flexible manipulation to address these problems, it remains challenging to accurately design the geometrical parameters to adapt to specific tasks, and accurate control is also required to realize the expected movement. Inspired by the biological characteristics of the octopus arm muscle fibers, fiber-reinforced actuators (FRAs) are employed to realize various motions, including extension, expansion, bending, and twisting; multiple FRAs are assembled to implement the SRL to achieve complex movement trajectories. The analytic model of the FRA is established to reveal the relationship between its deformation and geometrical parameters as well as input air pressures, which is validated with finite element simulation. Trajectory and payload optimization algorithms are proposed to optimally design the SRL and its control strategy with meeting the prescribed requirement of movement trajectory and payload capacity. Finally, experiments are conducted to validate the proposed robotic system.
BibTeX
@article{xu2024soft,
  author = {Jiajun Xu and Tianyi Zhang and Kaizhen Huang and Mengcheng Zhao and Xuyan Hou and Youfu Li},
  title = {{A Soft Supernumerary Robotic Limb with Fiber-Reinforced Actuators}},
  journal = {J. Intell. Robotic Syst.},
  year = {2024},
  doi = {10.1007/s10846-024-02102-6},
}
Figure