Stable Adaptive Neural Filters for Teleoperations

ICRA 2023 presentation and IEEE RA-L publication on adaptive neural filters that stabilize teleoperation systems under uncertain and time-varying delays.

Project Highlights

This project introduces an adaptive neural filter to improve stability and performance in teleoperation systems under random, time-varying delays. Key challenges addressed include:

Neural filter approach: An adaptive NN dynamically adjusts filter coefficients, ensures stability via Lyapunov-Krasovskii analysis, avoids local minima, and provides accurate signal estimations under uncertain delays.

Experimental results: Demonstrated outstanding performance, improved stability, and smoother teleoperation operation on a real-world platform.

Read Paper on IEEE RA-L