Hello Ruben,
Cool idea! Every time I come back from vacation I start designing one of these. I am yet to get beyond the back-of-a-napkin-sketch phase...
To start, have a look at this ROV (no affiliation with them and haven't used the product myself).
I believe your main challenges are outside of the control schematics / software:
- Creating a sufficiently water-tight setup that you can drop down to 60-80 feet. At depth (pressure) you can't get away with splash-rated electricals. Things get large/heavy/expensive very quickly;
- Getting a quiet propulsion system that is responsive enough to keep with the object, yet allows you to approach them without scaring them away (unless you are content with anemones and sea cucumbers);
- Getting the right power source. Most Li-based batteries create a big mess around water. I've read about people using the cells made by A123 and found in the DeWalt cordless drill battery packs for submerged propulsion pods;
- Finding a way to get rid of the tether (this has been paramount for me but might not be an issue for you). Wireless is a good idea under (salt) water, as long as it is not radio-wave based
;
If you are not afraid to tackle all of the above... I've always looked at this as an "underwater hovercraft" project: create (and maintain) neutral buoyancy and use trust-vectoring (angle/velocity) to move around. Won't win any high-speed chases but will probably get the longest dive times with the least amount of energy and disturbance. Plus it will provide the most stable camera platform - it's like getting a gyro-stabilized camera pod.
I wonder if any of the quad-copter projects out there cannot be used a a starting point when it comes to schematics / software:
- Use a 3-axis accelerometer to determine orientation (mainly for horizontal plane reference);
- Use a 3-axis gyro to determine rate of rotation (propulsion control keep those to a minimum when there is no operator input for camera PoV stabilization);
- Use pressure sensor for current depth detection. Delta registration is probably more important than absolute value;
- Use a pressure cylinder with a step motor for buoyancy control (several other ways to achieve this);
- Hook-up each of your motor pods through a decent bi-directional speed controller;
- Your "hover" algorithm will depend heavily on your thruster layout, but the basic idea should be to work towards zero-readings on the gyro and zero-deltas on the accelerometer. Once you get that, it should be relatively easy to "temporarily break" your hover algorithm to move in the right direction.
Just my 2 cents.
Keep us posted on the progress - really looking forward to seeing some underwater footage
Chris
www.projectpilotfish.com