BioRob 2026 Paper

This page provides an overview of my research paper accepted for the IEEE International Conference on Biomedical Robotics and Biomechatronics (BioRob) in June 2026. The work is titled “Real-Time 3D Proprioception for Soft Robots Using a Single Capacitive Bend Sensor”.

Overview

A core challenge in soft robotics is determining the exact position of the robot in a continuous state space. Traditional open-loop control methods struggle with continuous deformation and nonlinearities like cable slacking, while external camera systems are generally limited to laboratory settings.

To address this, we developed a lightweight proprioceptive approach to reconstruct the 3D pose of a cable-driven soft actuator.

Instead of relying on extensive data-driven training or complex material modifications, our method embeds a single commercial capacitive bend sensor directly inside the actuator’s central core. This sensor measures the net tip bending, which is then mapped to the full actuator shape using a Piecewise Constant Curvature (PCC) model.

The code is accessible on GitHub.

Key Results

• Accuracy: The sensor-based method maintains high tracking fidelity and effectively decouples state estimation from mechanical hysteresis. It achieves millimeter-level Cartesian accuracy (mean tip error of 2.48mm) and high angular precision (0.35°).
• Efficiency: The reconstruction algorithm executes in under 1ms, making it highly suitable for real-time applications.
• Robustness: The method significantly outperforms open-loop motor-based estimation, especially when mechanical disturbances (like slack) are introduced.

Demonstration Video

The supplementary video below demonstrates the experimental setup, the real-time reconstruction process, and the validation against the optical ground-truth system.