This project focused on building a robotic fish tail designed to mimic the efficient movement of fish. My role centered on mechanical redesign, system integration, and control development for the fin mechanism. I redesigned the model in SolidWorks to improve spatial layout and increase angular motion range, using FEA simulations to ensure structural strength for underwater operation.
Old design vs. new design which improves pitch and roll clearance of the fin
Fish tail assembly with 3D-Printed components to test fitment and functionality
The design was validated through physical prototyping and 3D fit assessments, achieving roll and pitch angles close to 30 degrees, a significant improvement from the previous design which had 15 degrees tilt angles.
Once the design was finalized we machined parts using a mill, lathe, and water jet.
Milling a D-Shaft
Full assembly with machined aluminum/steel components
One of the biggest challenges we faced during this project was implementing all the electronics together and performing various trajectories such as figure 8, ellipse, etc. I integrated a high-torque actuator (Cubemars AK60-6) and developed control software using Arduino and MATLAB. This software converted desired movement paths into motor commands, enabling the fish tail to execute realistic, programmable fin movements, laying the groundwork for future implementation of machine learning-based trajectory optimization. The schematic below shows the control logic, wiring diagram, and desired trajectories.
Robotic fish tail performing an elliptical trajectory
In the last week with the help of numerous online resources, we were finally able to make the fin actuate and perform any desired trajectory. Unfortunately due to the time constraint of the internship, we were not able to perform the experimental portion of the project. The future goals that remain on the project focus on experimentally validating the vehicle’s performance and efficiency, then using those results to improve control and autonomy.
This can be done by conducting water-tunnel testing to measure power and thrust across operating conditions, compare the datasets, integrate a machine-learning–based controller, and characterize key swimming trajectories (forward, maneuvering, hover), ultimately enabling reliable remote control for free-swimming operation.
This project was a group effort of me, my amazing teammate Jacob Schuster, and brilliant mentor Meredith Hooper