Reproduce “Leader-follower formation enabled by pressure sensing in free-swimming undulatory robotic fish” 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
Kundan Panta, Hankun Deng, Micah DeLattre, Bo Cheng
Researchers at The Pennsylvania State University experimentally demonstrated leader-follower formation swimming in free-swimming undulatory robotic fish, leveraging only two bilateral flow pressure sensors. An imitation learning framework with recurrent neural networks enabled a follower robot to track a randomly moving leader by interpreting complex hydrodynamic cues, achieving tracking performance comparable to an expert policy.
Key insights
- A minimal setup of two bilateral pressure sensors, complemented by internal state data, is sufficient for a free-swimming undulatory robot to perceive and react to a leader's hydrodynamic wake.
- Recurrent neural networks (LSTMs) can effectively extract actionable temporal patterns from sparse and noisy pressure sensor data to enable robust control in dynamic hydrodynamic environments.
- Strategic positioning in a staggered formation enhances the detectability of leader-induced pressure signals, allowing the follower to leverage fluid interactions for stable tracking.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.