Reproduce “A-SLIP: Acoustic Sensing for Continuous In-hand Slip Estimation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Uksang Yoo, Yuemin Mao, Jean Oh, Jeffrey Ichnowski
Researchers at Carnegie Mellon University developed A-SLIP, an acoustic sensing system that continuously estimates in-hand slip presence, direction, and magnitude using integrated piezoelectric microphones. The system achieved 81.8% detection accuracy and 14.1 degrees directional mean absolute error, enabling robust real-time feedback for robotic manipulation.
Key insights
- Textured silicone contact pads significantly improve the accuracy of slip direction estimation by promoting more informative structure-borne acoustic signatures.
- Multi-microphone configurations, particularly a 4-microphone setup, are critical for resolving directional ambiguity in acoustic slip signals, outperforming single-microphone approaches.
- A combined training approach (robot-induced pretraining, externally-induced finetuning) and a robust model architecture enable cross-object generalization and robustness to robot operating noise.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.