Proprioceptive Soft Robot Sensor Modelling
Compensation of nonlinear sensor dynamics in soft robotic actuators.
Principal Investigator
Dr. Xiaobo Tan
Collaborator
Preston Fairchild
Project Summary
This project tackles a fundamental challenge in soft robotics — the nonlinear and time-dependent response of stretchable strain sensors. We developed a lightweight compensation model using Time Delay Neural Networks (TDNNs) to correct for hysteresis and dynamic lag in conductive rubber-based sensors.
The approach was applied to a soft continuum manipulator, enabling real-time proprioceptive sensing with high spatial accuracy. The compensated output achieved a mean tip position error under 4% of the manipulator’s total length — demonstrating strong promise for closed-loop control applications.
Publication
📄 Naik, V.K., Fairchild, P.R. and Tan, X., 2025.
“Nonlinear compensation of stretchable strain sensors with application to proprioceptive sensing of soft robotic arm.”
Smart Materials and Structures, IOP Publishing.
Read the full paper →
