Research Analyzer
← Back ICRA 2026

Learning-Based Kinematic Modeling for Concentric Tube Robot: Addressing Its Nonlinearity and Snapping Behavior

Gowoon Jeong, Seong Young Ko

PDF

AI summary

Key figure (auto-extracted from paper)
A lightweight LSTM-MLP neural network accurately predicts forward and inverse kinematics for concentric tube robots, capturing complex nonlinearities and snapping behavior with sub-millimeter precision and real-time speed.
Concentric Tube Robot Learning-Based Kinematics Snapping Behavior LSTM-MLP Minimally Invasive Surgery Real-Time Control

Problem

Accurately modeling the nonlinear, history-dependent kinematics of concentric tube robots, particularly snapping and hysteresis effects, remains a major challenge for real-time microsurgical control.

Approach

The authors developed a lightweight LSTM-MLP hybrid neural network with a history input buffer and directional rotation parameters to learn and predict both forward and inverse kinematics from experimental data.

Key results

  • Forward kinematics RMSE of 0.69 mm and 0.16°
  • Inverse kinematics RMSE of 1.22 mm and 2.46°
  • Sub-millisecond computation time (~0.83 ms) for real-time control
  • Accurate prediction of snapping and hysteresis outperforming conventional models

Why it matters

Enables safer, high-precision real-time control for minimally invasive microsurgical applications where unpredictable tube snapping and hysteresis currently limit reliability.

Abstract

The Concentric Tube Robot (CTR) has great promise for minimally invasive surgery. However, accurately modeling nonlinear and history-dependent behaviors remains a significant challenge. This letter proposes a learning-based forward and inverse kinematics model that accounts for the history dependence and nonlinearities of CTR, including the snapping behavior. A lightweight LSTM-MLP hybrid neural network with an input buffer and directional parameters was used to train forward and inverse kinematics models for 4-degree-of-freedom (DOF) CTR. The model was validated by comparing its predictions with actual values and results from a conventional torsional-compliant model (TCM) across random points, rotational trajectories, and arbitrary paths. This validation successfully demonstrated the model’s ability to capture snapping behavior. For forward kinematics, the model achieved a Root Mean Square Error (RMSE) of 0.69 mm and 0.16° with a computation time of 0.831±0.200 ms. The inverse kinematics model achieved an RMSE of 1.22 mm and 2.46° with a computation time of 0.816±0.200 ms. The proposed method improves the accuracy and speed of kinematic modeling by capturing nonlinear behaviors, such as snapping and hysteresis. The lightweight system ensures accurate real-time control and offers a safer and more reliable solution for microsurgical applications.

Index terms

Machine Learning for Robot Control Surgical Robotics: Steerable Catheters/Needles Kinematics

Related papers