An interpretable approach to estimate the self-motion in fish-like robots using mode decomposition analysis.

Area of Science:

• Robotics
• Fluid Dynamics
• Bio-inspired Sensing

Background:

• Fish-like robots utilize artificial lateral line systems with velocity and pressure sensors for sensing.
• Estimating self-motion in complex flow fields generated by robot movement is a significant challenge.

Purpose of the Study:

• To develop and validate a mode decomposition method for accurate self-motion estimation in fish-like robots.
• To investigate the correlation between decomposed modes and fluid dynamics principles.
• To assess sensor array design redundancy and method generalizability.

Main Methods:

• Application of mode decomposition to artificial lateral line sensor data.
• Correlation analysis with Lighthill's theoretical pressure model.
• Validation using computational fluid dynamics (CFD) simulations of fish models and complex flow scenarios.

Main Results:

• Primary decomposed modes strongly correlate with velocity components.
• Analysis reveals redundancy in artificial lateral line sensor array design.
• Method successfully estimates self-states across varying parameters and in complex flows with vortices.

Conclusions:

• The developed data-driven pipeline is interpretable and generalizable for hydrodynamic sensing.
• This approach can enhance perception in autonomous underwater robotics.
• Potential applications include generating hydrodynamic sensing hypotheses in biofluids.