Black and white is all you need
We trained an artificial Intelligence on dividing underwater footage into two parts. Objects and open Water. This way our fish can avoid all types of obstacles with ease.
It works!
We tested the movement and steering of the full sized fish in a pool. The battery and the controlling electronics were still missing.
That looks complicated!
The layout and all cruical internal components are fixed in size place and orientation.
Did she get bigger?
Our seccond Prototype. This version contains all cruical parts except for the electronics. Belle can move and collect eDNA.
Fishhead V1
Doesn't look so professional? Trust me it is ;) If you ignore the fact that it is all held together by a cardboard box, you can see the first time all the major electronic components were successfully assembled. Both the camera and the nuc ran on battery power.
Is it moving?
Surrounded by a web of cables, tubes, valves and microcontrollers we finally get some movement out of our finn. Unfortunately the trio; Water, Electronics and rapid protoyping isn't ideal for our highly sensitive microcontrollers. This little Tailwag costed us 2 microcontrollers.
Click to play
VIO works! partially
Say hi to Dennis, proudly smiling in the grainy black-white image littered with green refference points. Feature extraction is a main building block for any Visual Inertial Odometry algorithm and depends heavily on lighting, corners and edges.
First Silicone Finn
After some test trials with casting and curing silicone we made our first functional fishtail.
First Steps
We divided the Team in different subsystems, each with their own responsibilities and tasks. Initially we had to get rid of some organisational to-dos, like creating a website and preparing pitchdeks for potential sponsors.
Concept and Team
In the first half of 2022 a concept, to autonomously gather underwater Biodiversity data, was drawn up. And a Team, that was ready to face the upcomming challenge, was assembled.
