Iteration 2: First complete spine of a fish. This iteration used a micro stepper motor in a pull-pull system to actuate tendons running the length of the tail. The tendons were allowed to move inside tubes on the interior of each vertebra and were attached to the most caudal vertebra. Tension in a tendon resulted in rotation of the vertebrae. The issue with this design was that the tendons applied force to the very end of the tail first. This created a delay in actuation of the whole tail and meant that all of the force was applied at one point. Further, the source of all the power delivered to the tail could only be derived from one tendon at a time for half of the tail actuation period. These issues would be fixed starting with iteration 4 when the tendon-based, pull-pull system was replaced with a better system that delivered power along the entire length of the tail on both sides.

Iteration 2: Fully printed and assembled spine. Each vertebra, rib, and dorsal spine are seperate printed parts that are glued together. This was done to produce the flattest dorsal spines possible and to reduce the required support material. This would be changed in iteration 8 for ease of printing and for quicker assembly time.

Iteration 3: Second complete spine of a fish. Tendons were still used for this iteration. The purpose of this iteration was to explore spine curvature, pectoral fins, and also to make a longer fish. The issue with this design was that the parameters for the shape of the ribs and dorsal spines were not well defined and had to be redone every time a change was required. This would be changed in iteration 7 with curves that defined the tips of each rib and dorsal spine and allowed for quick modification of each vertebra.

Iteration 3: Fully printed and assembled spine. Pectoral fins were not attached this iteration in favor of producing a smaller test assembly exclusively for the fin.

Iteration 7: The new tail actuation system with hydraulic accumulators located at each vertebra. This system uses a pump to differentially inflate the accumulators on either side of the fish. This new method allows power to be delivered along the entire length of the tail. Further, by pumping water from one side to the other, the deflation of one side and spring force in the skin on that side contribute to the actuation. An issue with this iteration is that the technique uesd to create the muscle and spine elements did not deal with the most anterior and posterior stations efficiently and made it impossible to create muscles at these locations. Iteration 8 solved this issue by using more careful design considerations. Additionally, creation of the myotomes in real life were made possible with molds that were generated at each vertebra station. The muscles were made by layering liquid latex inside a clamshell mold. A single layer of mesh fabric lined the larger faces of the muscle to distribute the pressure across the entire surface and to avoid the formation of bubbles at sites where the latex was too thinly layered.

Iteration 7: Sagittal cross-section. Intervertebral discs to be printed in TPU can be seen between a few of the vertebra. These allow for freer actuation of the tail and provides a bit of squish to ensure the vertebrae are seated well together. Additionally, a slot can be seen cut into the dorsal spines of some vertebrae for flex sensor installation. The flex sensor allows for accurate measurement of the position of the tail. This system currently uses long, off-the-shelf flex sensors, but switching to a more advanced tail actuation system will require multiple small, DIY flex sensors spanning only a couple vertebrae

Iteration 7: Transverse cross-section

Iteration 8: Complete spine of the fish. This iteration features more complex rib and dorsal spine geometries. Further, each vertebra section is a single part and is easily printed in under an hour with minimal support material. Finally, the myotomes are more easily created because the complex curves of iteration 7 were eliminated, making it easier to lay down mesh.

Iteration 8: Side view with muscles showing the simpler design for easier muscle molding

Iteration 8: Isometric view

Iteration 8: Pectoral Fin

Iteration 8: Dorsal fin featuring spines and rays

Iteration 8: Test prints and assemblies for various parts of the fish.

1) Dorsal fin with both spines and rays.

2) Segment of vertebrae to test flex sensor for tail position feedback.

3) Newest iteration of the pectoral fin; the printed spring was removed because it was too rigid. The vertical bar seen here is cut off once the fin rays have been contained in the latex/mesh composite membrane.

4) The first ray-supported pectoral fin. Tests with this assembly determined that a mesh was necessary to avoid weaknesses in the non-composite latex.

5) The first pectoral fin actuation test. This consists of two spines hinged along non-coaxial axes and are actuated with low-power linear servos. This assembly (along with accompanying circuitry) determined that the complex pectoral fin motions of a fish could be easy replicated using only two independently actuated spines.

6) The most recent full assembly of a pectoral fin. Tests here determined that the printed spring greatly interfered with the required flexibility of the fin but also demonstrated the best membrane layup technique (CA glue applied between mesh and fin rays, then the liquid latex is brushed on).

7) The first successful caudal fin print. Actuation tests have yet to be performed.

8) A test assembly of three hydraulic accumulator muscles connected with small-diameter latex tubing. Results here demonstrated expected inflation and deflation characteristics but determined that liquid latex does not adhere well to store-bought latex tubing. Difficulty making each muscle led to simplification of the muscle geometry and eventually to the decision to use electromagnetics instead of hydraulics to actuate the fins and tail.

Iteration 10: This is the most recent iteration. It features a tail actuation system that still needs to be evaluated for its patentability, so no information about it will be provided here. The pectoral fins are actuated using small DC motors with trim potentiometers for position feedback. The spines of the first dorsal fin are extended by spring tension in silicone strips and retracted using a micro servo. The pelvic fins are extendable and retractable with micro servos. Silicone mold making is simpler and more robust than latex mold making, so muscles, fins membranes, and skin will be made from silicone. The process to select electronics, like the batteries and controllers, has begun and some parts have been introduced into the model to test dimensions. The entire skeleton, except for the head, has been printed and assembled. Pectoral fin actuation testing and redesign processes are currently underway. The Pelvic fin actuation mechanism has been tested and finalized. Fin material selection is still ongoing.

Iteration 10: View of the most recent progress of the 10th iteration. Body ribs were constrained and unpaired fins were deformed using law-defined curves. The deformation of the dorsal fins can be easily seen.

Iteration 10: Top view showing the body wave.

Iteration 10: Isometric view showing the body wave.