Modular Soft Robotic Catheter

• Modular soft robotic catheter is a multifunctional endoluminal tool, designed with a compliant 1.47 mm silicone body and integrated modules for anchoring, sensing, manipulation, and targeted drug delivery.
• It employs wireless magnetic actuation, tendon-driven gripper control, and real-time closed-loop sensing (camera, FBG, pressure sensor) to enhance navigation in complex anatomies.
• Advanced fabrication methods, including 3D-printed components and precise microchannel integration, support robust in vivo performance and compatibility with standard endoscopic procedures.

A modular soft robotic catheter is a multifunctional endoluminal instrument engineered to integrate actuated modules for anchoring, sensing, manipulation, and targeted therapy delivery within a compliant, miniaturized (1.47 mm outer diameter) body. The system advances the domain of soft-robotic instrumentation by enabling customizable, independently controlled modules to optimize navigation and intervention in delicate, tortuous anatomies inaccessible to standard catheters. Through wireless magnetic actuation, tendon-driven manipulation, closed-loop sensing, and in-situ therapeutic release, this platform supports precision therapies with enhanced compliance, safety, and functional scalability (Calmé et al., 21 Jan 2026).

1. System Architecture and Modular Composition

The modular soft robotic catheter (mSCR) architecture consists of a monolithic silicone body (1.47 mm OD) designed to accommodate up to four functional units in series. Modules, sequentially embedded along the catheter's length, are connected by three 100 µm microchannels/lumens:

• A pneumatic lumen for balloon inflation
• A tendon lumen for gripper actuation
• A central lumen for camera cabling, LED fiber, and multicore FBG

Primary demonstrated modules comprise:

• Anchoring balloon: Hyperelastic, magnetically responsive, inflatable actuator to stabilize the catheter at target loci.
• Manipulation gripper: Tendon-driven, flexure-based L-arm gripper for compliant object retrieval and manipulation.
• Sensing modules: Onboard NanEye camera (250 × 250 px at 30 fps), FBG shape sensor (27 Hz interrogation), and inflation-pressure sensor (VSP1130, ±0.5 kPa accuracy).
• Drug release shell: Micropatterned, spin-coated PLA coating designed to fracture upon 1 MHz ultrasound, enabling on-demand payload delivery.

Mechanical compliance is preserved using low-modulus elastomer (DragonSkin 30) embedded with NdFeB powder and through the optimized geometry of each module (e.g., flexure and balloon wall thickness). The complete assembly is sleeved through a 0.9 mm ID PTFE tube (1.6 mm OD), maintaining flexibility comparable to standard soft catheters.

2. Materials, Fabrication, and Integration

The construction of the mSCR employs advanced materials and fabrication methods:

• Elastomeric matrix: A 1:1 mass ratio of NdFeB (5 µm) and DragonSkin 30, mixed in vacuum and injected into high-resolution molds (microArch S140).
• Silicone and resin shells: Two-part microfabricated HTL resin structures, defining precise air-channel cores and guides for integrated electronics and optics. These are assembled using sacrificial cores dissolved in acetone.
• Gripper: Ultra-high-resolution, 3D-printed (HTL resin) flexure structure, bonded around the imaging module.
• PLA drug-release shell: Spin-coated and micropatterned with 20 µm squares (5 µm depth, 10 µm spacing) by molding on a 3D-printed negative.
• Sensing and transmission elements: NanEye camera, LED fiber (Thorlabs), multicore FBG, nylon tendons (80 µm), and fine-ID/OD inflation tubing.
• Assembly procedure: Sequential demolding, camera/FBG installation, gripper bonding, PLA shell slip-fit, routing of actuator lines, and final coupling to control units.

This integration strategy enables monolithic compliance without local stiffening, supporting both module flexibility and dense packing of active elements.

3. Actuation Principles and Mathematical Modeling

Magnetic Steering

Tip actuation is achieved wirelessly by an external permanent magnet (EPM, N52; 100 mm diameter, 100 mm thickness, remanence ≈1.45 T) mounted on a 7-DoF robot arm. The resulting field at the catheter tip ($|B_{EPM}|$) spans 16–25 mT, transmitting torques up to $\tau_{tip} \approx 6.7~\mathrm{N{\cdot}m}$. The EPM–mSCR interaction is described by a dipole–dipole physical model:

$$B_{EPM}(p) = \frac{\mu_0}{4\pi|p|^3} \left[ 3(p\cdot m)\frac{p}{|p|^2} - m \right]$$

$T(x) = \frac{\mu_0}{2\pi}|B_r||B_{EPM}|r^2(L-x)$

where $r$ is radius, $L$ is magnetized length, $B_r$ is remanence, $\mu_0$ the vacuum permeability, and $p$ is EPM pose. Maximal tip deflection ($\theta_{max}$) is modeled via Euler–Bernoulli theory:

$$\theta_{max} = \left(3E\mu_0 r^2 \pi L^3\right)^{-1} |B_r||B_{EPM}|$$

Anchoring Ballon Actuation

Balloon expansion is pneumatically regulated (syringe pump up to $p_{in}=60$ kPa), with strain–pressure response modeled by a third-order Yeoh strain-energy density function:

$W = \sum_{i=1}^3 C_i (I_1 - 3)^i$

where $I_1$ is the Cauchy–Green invariant and $C_i$ are empirical constants. Measured anchoring force is $F_{anchor}=3.3 \pm 0.2$ N.

Tendon-Driven Gripper Dynamics

The compliant gripper is modeled with a shifted Legendre polynomial curvature expansion:

$\kappa(s)=a_0 + a_1 P_1(s) + a_2 P_2(s)$

and tip response is determined by tendon force ($f_{tendon}$). Pre-twisting elevates fundamental resonance above 2 MHz, avoiding 1 MHz actuation cross-talk. Maximum tip gripping force is $1.28 \pm 0.16$ N when anchored ($0.43 \pm 0.06$ N unanchored, a $\sim 300\%$ increase).

4. Sensing Modalities and Real-Time Calibration

Onboard sensing subsystems are fundamental for navigation precision and autonomous feedback.

• FBG Shape Sensing: A three-core FBG records distributed strain ($\epsilon_i$) at 27 Hz. Curvature conversion: $\kappa=(\epsilon_{+}-\epsilon_{-})/d$, $d$ core spacing; tip orientation $\psi(s)=\int_0^s \kappa(u)du$. Calibration errors are under $2^\circ$ relative to ground-truth.
• Balloon Pressure Sensing: A VSP1130 sensor provides balloon pressure with $\pm 0.5$ kPa accuracy; PID-based closed-loop—$u(t) = K_p[e(t) + (1/T_i)\int e + T_d\,\dot{e}(t)]$, managing actuation within safe limits ($p_{in}\leq150$ kPa burst pressure).
• Visual Marker Tracking: Two heat-shrink markers and papilla (anatomical landmark) are detected by YOLOv8 at 30 fps (precision 0.81, recall 0.76, F1 = 0.78). Angular correction:

$$\Delta\psi = \arccos\left( \frac{v_{tip} \cdot v_{goal}}{|v_{tip}||v_{goal}|} \right)$$

where $v_{tip}$ and $v_{goal}$ are inter-marker and target vectors, respectively.

5. Closed-Loop Control and Automation

The mSCR employs a closed-loop, ROS-integrated pipeline combining perception, model-based planning, and actuation:

1. Endoscopic video analyzed via YOLOv8 for marker and target detection, yielding $\Delta\psi$.
2. Magnetic model computes desired tip curvature $\kappa_{des}=f_{model}(\Delta\psi)$.
3. Inverse field solver translates $\kappa_{des}$ to a desired EPM pose increment ($\Delta p$).
4. 7-DoF robot manipulates the EPM accordingly.
5. FBG shape sensing confirms execution ($|\psi_{FBG}-\psi_{model}| < 2^\circ$).
6. Catheter is advanced using an external feed unit.

Control architecture features PID regulation for balloon inflation, damped least-squares for magnet Cartesian control, and curvature-to-field mapping ($\Delta B \propto EI\kappa_{des}$). Human override is enabled at all times.

6. In Vivo Validation and Performance Evaluation

Key procedural validation was performed in two Yorkshire–Landrace porcine subjects (33 kg, 35 kg) via standard duodenoscopes, with the following protocol:

• Papilla cannulation using duodenoscope
• Semi-autonomous magnetic tip alignment and 75 mm navigation into pancreatic duct (depth not accessible with conventional catheters)
• Balloon deployment to anchor mSCR
• Object retrieval via gripper
• Ultrasound-triggered (1 MHz) drug release

Performance metrics:

Task	Quantitative Metric	Comments
Navigation depth	75 mm	Pancreatic duct; inaccessible to standard devices
Autonomous cannulation time	$t_{auto}=223\pm3$ s	12% faster vs. manual ($t_{manual}=253\pm23$ s)
Anchored gripper force	$1.28\pm0.16$ N	$0.43\pm0.06$ N unanchored
Balloon holding force	$3.3\pm0.2$ N	Enough for robust actuation
Drug release (T$_{80}$)	$104\pm22$ s	80% payload after 1 MHz insonation
Tip orientation error	$<2^\circ$	FBG vs. model
Clinical complications	None	No papilla damage, no bleeding

The platform demonstrated compatibility with duodenoscope, gastroscope, and bronchoscope access, underscoring generalizability.

7. Clinical Significance, Limitations, and Future Directions

The modular mSCR supplies a scalable architecture for hospital-based and research endoluminal interventions, achieving expanded functionalization of compliant catheters without sacrificing mechanical softness. Key clinical benefits include first-person visualization, reduced X-ray exposure, semi-autonomous tip navigation (shortening the procedural learning curve), and stimulus-triggered drug delivery precisely at target sites.

Identified limitations include unresolved depth estimation (addressable via miniaturized localization modalities), lack of multi-dose capability (potentially via multi-resonant segmented PLA shells), and the need for distal tip biohybrid actuators to further enhance manipulation accuracy. Long-term biocompatibility for chronic deployment remains under investigation.

Through integration of wireless magnetic steering, pneumatic anchoring, compliant manipulation, onboard feedback, and ultrasound-mediated drug release, the mSCR defines a new standard for precision therapies in anatomies and disease states where conventional catheters lack efficacy (Calmé et al., 21 Jan 2026).