The MOLLUSCA Project
An estimated 171 trillion plastic particles are currently floating in the world's oceans, and the goal of this project was to find a better way to get them out. Conventional net-based collection disturbs the water surface, mixing plastic particles down into the bulk and capturing large amounts of organic matter along with them. As part of MOLLUSCA my research group and I helped developed a robot modeled after the apple snail, which feeds by undulating its foot to pump particles along the water surface. The robot uses a 3D-printed undulating "carpet" to pump at the liquid–air interface, drawing microplastics toward a collection unit without disturbing the water beneath it.
I characterized the undulator so the team could optimize collection. I amended the designs and built the test apparatus in SolidWorks and analyzed the footage to show that pumping performance peaks at a specific driving frequency that shifts with particle size and depth, allowing for optimization and tuning of the full-scale robot.
I was responsible for building the test apparatus used to characterize the undulator. I modeled the rig in SolidWorks and fabricated the parts through 3D printing and machining (Mill/Lathe.) The design has an undulating carpet driven by a rotating helix on a 12V DC motor, mounted inside a clear-walled water tank alongside a force sensor and a high-speed scientific camera for imaging particle motion at the surface. I wired and programmed the motor driver and sensors through an Arduino so we could sweep driving frequencies and log data repeatably. Getting clean measurements took a lot of iteration because vibration from the motor shook the whole tank and blurred the imaging. Stray room light washed out the particles against the background so I redesigned the mounting to isolate the motor from the tank frame and built an enclosure with controlled backlighting to keep the optical setup consistent between runs.
The Test Apparatus
To measure undulator quality and performance, added microplastic particles of known sizes to the tank. We tracked how the particles moved across the undulator to be able to numerically quantify the flow. I used particle tracking software in MATLAB that follows individual particles frame by frame as they move across the water surface. I repeated test across varied parameters such as driving frequency, particle size, and depth below the surface. Reflections and uneven lighting created false detections and particles occasionally overlapped or dropped out of frame, so I added filtering and gap-handling logic to keep trajectories from breaking or jumping between neighboring particles.
Testing
We measured mean surface particle velocity across driving frequencies from roughly 3 to 15 Hz, for two particle sizes (100 µm and 1000 µm) at three water heights above the undulator (2, 4, and 6 mm). Every configuration followed the same shape, velocity climbed with frequency to a local maximum near 7–11 Hz, peaking around 180 mm/s, then fell off. This means Faster undulation does not mean better collection past that point and there is an optimal frequency. The location of that peak shifted with the other variables such as particle size and water height. This means the full-scale robot needs to be tuned to the particle size and depth it is targeting rather than run at a single fixed frequency.
Result
