Abstract
Tendon-sheath-driven manipulators (TSM) are widely used in minimally invasive surgical systems due to their long, thin shape, flexibility, and compliance making them easily steerable in narrow or tortuous environments. Many commercial TSM-based medical devices have non-linear phenomena resulting from their composition such as backlash hysteresis and dead zone, which lead to a considerable challenge for achieving precise control of the end effector pose. However, many recent works in the literature do not consider the combined effects and compensation of these phenomena, and less focus on practical ways to identify model parameters in realistic conditions. This paper proposes a simplified piecewise linear model to construct both backlash hysteresis and dead zone compensators together. Further, a practical method is introduced to identify model parameters using motor current from a robotic controller for the TSM. Our proposed methods are validated with multiple Intra-cardiac Echocardiography (ICE) catheters, which are typical commercial example of TSM, by periodic and non-periodic motions. Our results show that the errors from backlash hysteresis and dead zone are considerably reduced and therefore the accuracy of robotic control is improved when applying the presented methods.
Introduction
Tendon-sheath-driven manipulators (TSMs), such as steerable catheters for heart surgery, are prized for being thin, flexible, and compliant. However, they are notoriously difficult to control precisely. Because the tendons rub against the sheath and have inherent slack, they exhibit non-linear behaviors: backlash hysteresis (a delay in movement when reversing direction) and dead zones (a range of input where the tip doesn't move while tendons take up slack).
While previous research has attempted to model these effects, most solutions require adding external sensors like EM trackers or vision systems to the catheter tip. In clinical environments, these sensors increase cost, device size, and sterilization complexity. This paper presents an "open-loop" solution that achieves high precision using only the data already available in the robotic controller: motor current.
Methods and System Overview
The proposed solution consists of a simplified piecewise linear model and an automated parameter identification method using motor current as a proxy for physical sensors.
Piecewise Linear Model
Instead of complex differential equations, the authors developed a model consisting of eight linear equations. This model maps desired input to actual output by accounting for the "flat" regions of dead zones and the "lag" regions of backlash.
Motor Current as a Sensor
The key innovation is using motor current to identify model parameters. Specific "signatures" in the current correspond to physical states:
- Dead Zone: Indicated by a flat, constant current signal as the motor moves without resistance.
- Backlash Interval: Indicated by a sharp current peak followed by a plateau, marking the point where the tendon regains tension.
By sweeping the catheter through its range of motion and analyzing these spikes and plateaus, the system can automatically calibrate itself for a specific catheter before use.
Experimental Results
The approach was validated using commercial medical hardware, including multiple Intra-cardiac Echocardiography (ICE) catheters and Transesophageal Echocardiography (TEE) probes. To verify accuracy, an Electromagnetic (EM) tracker measured the actual tip position as ground truth.
Key Findings
- 1-DOF Precision: For single-axis movement, the Peak-to-Peak Error (PTPE) was reduced by 62% to 66%, and the Root Mean Squared Error (RMSE) dropped by 71% to 76%.
- 2-DOF Precision: In simultaneous two-axis motion, the system improved PTPE by approximately 40% and RMSE by 58%.
- Linearization: Qualitative analysis shows that the "input vs. output" graphs, which typically appear as distorted hysteresis loops, become nearly straight diagonal lines after compensation.
Discussion and Future Work
The motor current acts as a "free" sensor, reflecting tendon tension and friction without adding wires to the catheter. While highly effective, the piecewise linear approach can occasionally cause slight "jerkiness" during segment transitions. Furthermore, while the model accounts for some coupling in 2-DOF motion, complex mechanical interactions between tendons remain a challenge. Future research will focus on how catheter tortuosity (bending through curvy veins) affects the width of the dead zone, requiring more dynamic parameter identification.
Summary updated