JT Jefferson Talbot
Autonomous Robot Fish
A Work in Progress
Created on 2018.06.15 1690 views
- Hello, my name is Jefferson Talbot. I am a recent Aerospace Engineering graduate from Embry-Riddle Aeronautical University. I have had the opportunity to create many projects over the course of my time at college, so I would like to share my most recent and involved project.
- I first came up with this idea while sitting in Calc 3 class in the Fall of 2016. I sometimes did little doodles in the margins of my notebook when class got slow, and one day I drew a fish skeleton with smooth, artificial-looking bones. Nothing much happened soon after that.
- Fast forward to last summer and the idea for a robot fish popped back into my head. I had recently obtained a 3D printer, so I started researching the anatomy of fish and began modeling. Before I knew it, I had a small, semi-complete skeletal structure of a basic fish (only the ribs, vertebrae, and spines were made). This is the short, white, printed skeleton in the pictures. I then redesigned the fish with a more complex spinal shape and rib geometry definitions and printed that out (this is the longer, gray, printed skeleton in the pictures). These skeletons were designed with tubes on the interior of the vertebrae for tendons to pass through to actuate the tail. The time and effort it took to modify the geometry at each body station led me to redesign the fish with splines defining important curves along the body, such as for the bottom tips of the ribs. Also, the tendon-based tail actuation system was giving me poor results, so that system had to be redesigned.
- The next major iteration can be seen in the pictures (the mass of red muscle). This was my attempt to replicate, as closely as possible, the actual shape of fish muscles. These were the models I ended up printing out to use as molds to test the muscle hydraulic actuation system. These models proved too complex and time-consuming to print and have latex applied to it, so the next iteration was greatly simplified.
- This next iteration is the design I was able to fill out considerably. On top of the skeleton, I was also making fins, hydraulic accumulator muscles, a skull with some mounting locations for servos and sensors, and several of the internal components, like pumps, electronic boards, batteries, etc. This is one of the more complete models seen in the pictures (the one with yellow, latex-colored skin). The skin, muscles, and fin membranes are made of liquid latex spread over printed molds with mesh included for extra strength.
- After playing around testing the muscles and fins, I determined that I wanted to make more robust, smooth, and fast-acting fin and tail actuation systems. So, I abandoned that iteration of the fish and moved on to a new one with significantly improved characteristics that should solve all of the issues I had, which included having difficulty implementing pelvic fins cleanly. This is the iteration I am on now, and it is the other nearly-complete-looking assembly with gray-colored skin.
- In addition to the structural aspects of the fish, I have also begun basic work on the electrical and control systems. An assortment of touch sensors, accelerometers, gyroscopes, cameras, LiDAR modules, pumps, servos, motors, Arduino boards, and Raspberry Pi boards was considered and a mix of them will be implemented in future iterations of the fish. The most recent iteration of the fish (with gray-colored skin) has black bands running along the ribs, which are my attempt at a quick representation of velostat-based pressure sensors. These sensor will be one of the last I implement since they aren't as important as other systems.
- In the picture with the number labels: (1) is the dorsal fin; (2) is a test for the flex sensor that would measure the deflection of the tail for a closed loop-control system; (3) is the most recent pectoral (side) fin design (iteration 7); (4) is the first pectoral fin assembly with ray-based structure that was coated in latex for underwater testing; (5) is the first iteration of the pectoral fin that was used to demonstrate a high level of fin control with only two servos; (6) is the first pectoral fin assembly to feature an internal mesh; (7) is the first successful caudal (tail) fin print test; and finally (8) is a better view of the complex myotome (muscle) assembly that was used to test inflation capabilities.
- The first few pictures show the skeleton of the most recent fish iteration. The black-background image shows the actuation abilities of the tail of the fish, the brown-background shows comparisons to previous iterations of the skeletons, and the CAD images depict the most up-to-date version of the fish as of 8/30/18, with most of the internals hidden. The rib angles and unpaired fin deformations were established using a law-defined curve.
Discover the team
Who’s behind this project
Discover the solution
Software used for this project
Post updated with pictures and info for the progress on the most recent iteration