Underwater vs Land robots

Hi,
Could someone tell me what are the challenges faced and what different hardware/software is required while building an underwater robot as compared to a wheeled land robot?

This is what I think:

1. Communication - you cannot use RF effectively underwater while with a land robot you can transmit signals wirelessly to long distances with no problem. With an underwater robot you'll have to use wires or probably some kind of ultrasonic method of communication.

2. Visibility - underwater you cannot see far away. this will probably require a light source (extra power) and perhaps different cameras.

3. Power - while on the land, you can use solar panels to recharge - you cannot do this underwater. Also, acceleration underwater is much more difficult because of high water drag.

4. 3D - moving under water always involves 3 axis. While on land you generally move in 2 axis only. This requires more control surfaces.

5. Construction - a deep underwater robot has to be able to withstand high pressures.

6. Moisture and corrosion - environmental conditions underwater requires special protection to you electronics and use of special materials for mechanical parts.

That covered most (and more) of what I could imagine. Also I was wondering how the electrical connections and PCBs function under water...

Do you know of any hardware that is used to solve these problems?

Make yourself familiar with the IP code.
https://www.dsmt.com/resources/ip-rating-chart

Electronic connectors that are directly exposed to water have to be IPx8 (second IP digit is 8).
You'd probably also need fine solids protection (against sand) so the first digit may also be of importance for you.

Now, very important aspect of electronic board design is proper thermal management.
Electronic components (and motors) generate (a lot of) heat. Water has excellent thermal conductivity and will remove heat from the surface of your robot quickly.
However, your electronics will have to effectively transfer that heat to the chassis which may prove to be quiet a challenge sometimes.

Also, components with proper thermal rating will have to be chosen - industrial components are generally rated between -40C and +80C. That may not be sufficient...
I've seen designs with components reaching temperatures of above +100C...

Other then that your assembled PCB will probably require conformal coating.
https://en.wikipedia.org/wiki/Conformal_coating
Low temperature will cause the air around your PCB to condensate and moisture to build up on the metal contacts - causing short circuits and corroding them.

Note: most of the above also applies to land systems.

There are a couple of standard ways to deal with the water (and pressure)issues, one common approch is to build a metal pressure vessel that houses all the doings with wiring passed thru connectors designed to take the pressure (Or cables fitted with water blocks, be careful here below 100m or so the water will quite happily flood the pressure pot by flowing up the inside of the outer sheaf of a cable).

There are a couple of thngs to note about this approach, formost being that if you do have the pressure pot flood due to a small leak and you then bring the thing back to the surface you have just created a bomb as the internal pressure may still be many atmospheres in the small volume of high pressure air trapped above the water, a pressure relief valve is a good thing.

The next approach is to fill the housing with oil which helps with both cooling and can optionally be pressure compensated such that the internal pressure matches that of the water, good for shallow water work (But messy to work on), you quickly find out which parts have voids inside if you send such a thing down to a few km.

The final common approach is to pot everything in PU, same issues with voids but if you do the prep on the wiring right (the PU MUST bond to the wires) it will be watertight.

You will get real good at inspecting O rings and thier grooves if you go into this stuff.

Now, control is best done with a wire unless you are determined to go autonomous (and those guys regularly loose vehicles, you have been warned), but any such wire must be close to neutrally boyant or it will become an issue, also the voltage drop due to the resistance of a km or so of cable can be a real pain, lots of volts up on deck is the typical answer.
You can get off the shelf cables for this stuff which have a couple of power cores and a pair of fibres which is increasingly how the ROV game is going, put an ethernet switch on the vehicle and you have loads of bandwidth for video, sonar (Generally more useful then video), telemetry and telecommand.

Regards, Dan (Who used to do sonar for a living).

Dan,

Almost all underwater vehicles I've seen use cables for power and communication. How common are robots that use wireless ultrasonic communication systems?

Acoustic modems are used on some vehicles but power becomes an issue as does data rate, somewhat more common are pingers used to allow the position of the thing to be monitored and the vehicle to be located if it all goes wrong.

In shallow water the symbol rate is limited by the reverberent field and you find things like COFDM being employed, in deep water you need to lower the operating frequency as range becomes rather limited above a few hundred Khz.

A few hundred bits to a few tens of kilobits per second if fine for control and useful for tracking and dynamic positioning but is useless for sending enough sonar or video to allow an operator to drive the thing.

Usually a cable of some form is the way to go for a general ROV unless there are pressing reasons to need to operate autonomously.

Regards, Dan.