Artificial lateral line based relative state estimation between an upstream oscillating fin and a downstream robotic...

Area of Science:

• Robotics and Bio-inspired Systems
• Fluid Dynamics and Hydrodynamics
• Sensor Technology

Background:

• Fish utilize lateral lines for environmental sensing and flow-related behaviors.
• Artificial lateral line systems (ALLSs) are inspired by fish lateral lines and applied to underwater robots.
• Estimating relative states between underwater objects is crucial for robotic navigation and interaction.

Purpose of the Study:

• To investigate the use of ALLS-measured hydrodynamic pressure variations (HPVs) for estimating relative states between an oscillating fin and a robotic fish.
• To develop and compare regression models for predicting relative states from HPVs.
• To identify optimal sensor configurations and regression methods for accurate state estimation.

Main Methods:

• Flume experiments were conducted to measure HPVs and relative states (frequency, amplitude, offset, distance, yaw, pitch, roll).
• Criteria were proposed to assess sensor sensitivity, insufficiency, and redundancy for optimal sensor selection.
• Four regression methods (Random Forest, Support Vector Regression, Back Propagation Neural Network, Multiple Linear Regression) were employed to build models.

Main Results:

• The Random Forest (RF) algorithm demonstrated the best regression performance among the tested methods.
• The RF-based method successfully estimated relative yaw angle and oscillating amplitude with excellent accuracy.
• Sensor selection criteria helped optimize the regression analysis by identifying sensitive and non-redundant sensors.

Conclusions:

• ALLS-based HPVs can be effectively used to estimate relative states between underwater robotic systems.
• The RF algorithm offers a robust and accurate approach for such estimations.
• This research contributes to the advancement of underwater robot perception and control through bio-inspired sensing.