Last year I built a negatively buoyant ROV crawler, with the idea of “crawling” along the lake bottom for exploration, rather than using a 3D submersible. An early design concept was to eliminate the crawler being tethered to the shore, with the inherent issue of entanglement in bankside and bottom weeds, and rocks.
I set a limit of 60 feet for depth and attached a 70-foot tether from the crawler to a small (8” X 6”) foam float. On the float was an RC receiver and a video transmitter, plus a battery. On shore a video monitor and RC transmitter was used to control the crawler.
I was most concerned that the RC PWM signals from the receiver to the Arduino in the crawler would be attenuated too much to be reliable. I was wrong on that account. Signal strength at the Arduino pins was good enough to be reliable.
There was one problem, however, that I was unable to overcome. The lake bottom is so soft that the wheels almost immediately bogged down and game over. The wheels were 5” diameter RC crawler tires.
I decided to not pursue the project any further. A possible solution would be to use much larger diameter tires and add a vertical thrust motor to “unstick” the crawler.
As I posted yesterday, I am currently building a 3D ROV for lake exploration. Initially I will use a tether to the shoreline, but eventually I want to try the float tether concept as an option to eliminate long tethers to the shoreline. A thought some of you might want to think about this as a future project.
In case it’s of interest, there’s a similar system that was brought up recently here
I expect that spinning and rolling propulsion mechanisms are in general poorly suited to gradual vertical density changes in the medium being moved through:
propellers pull up denser material from below which can make them slow down and/or get stuck while also reducing visibility
small wheels have little to grip onto, and large wheels are heavy, bulky, and may have similar issues to propellers
Likely a more effective solution would be an undulating, gliding, and/or flapping mechanism (e.g. something like this biomimetic robot). Unfortunately such mechanisms don’t tend to be readily available, or have readily available control algorithms, so making one would likely need a fair amount of custom development of both hardware and software (although it’d also be really cool…).