ROVs Design

Community Groups Marine Robotics
1

Hi folks!

I’m planning to make a simple and very cheap ROVs to do some experiments with thrusters (homemade), pressure sensors, IMU, underwater vision with simple RGB camera.
I have experience with ROS, my question is about the hardware design.

I didn’t want to use a raspberry, but a tethered communication with my PC to send all the sensors data (I don’t care about some latency).
To manage all the sensors feedback and the thrusters signals I was planning to use a simple Arduino, with the proper shield and DC/DC to control the high-voltage thrusters DC motors.

For the frame structure I was thinking about the simple SeaPerch design.
My question is about how to build a waterproof case with all the electronics , and also the waterproof cabled connection from PC to the robot brain (the Arduino).
I know that often the overall electronics is surrounded by gel, inside of a insulated case, how can I create such a case?
The SeaPerch project is a simple ROVs used only for teleoperation.
What I want to add is the simple sensors to get some feedback, visualize the robot state in Rviz using simple IMU and Pressure odometry, and get monocular camera feedback, maybe to perform some simple object classification during dive.

Anyone has some advice or know some useful blog related to DIY underwater robot design?
I don’t even know if my simple project is feasible with this simple design choice, expecially in terms of electronics and communication (I might be good with ROS and RoboticsTheory but a principiant with electronics).

Thank you all for your help and advices!
Consider also the most stupid electronic comment useful for me (I’m completely noob with that!)

1 Like
2

Hey that sounds like a pretty cool project! I’ve been down this road three times now :smiley:

I can recommend this really old site which is a fantastic resource: https://homebuiltrovs.com. There’s a section about potting cables which has always worked perfectly for me. Blue Robotics is a good resource of best practices in terms of ROV design.

Sewer pipes are generally a good option, they seal well and resist reasonable external pressure and you can pot each end in epoxy to seal in cables. The name of the game is making your electronics use as little volume as possible, since all the displaced air will need to be offset with lead weights to make it neutrally buoyant. Water is heavier than it looks.

Before you go designing your own, I would suggest checking out the off the shelf T60 thrusters, they’re a slightly smaller but vastly more reasonably priced version of the completely unaffordable T200 from BR. T60s have basically solved all of my underwater propulsion needs for cheap and they shrug off salt like it’s nothing.

A $15 bilge pump with a propeller is still quite a bit cheaper but the efficiency loss from it (both electrical and hydrodynamic) means you’ll need a larger battery and beefier ESCs to compensate, and they only go down to like 20m before the shaft seals seize from the pressure. Properly epoxy coated, water lubricated BLDCs have no such problems. I’d only recommend going DC if you have a very minimal budget.

As a rule of thumb, these two don’t really go well together. Even the ESP32-CAM struggles with streaming camera data at a reasonable rate and resolution, and can’t really do much else while it’s occupied by that. You’ll likely need some form of small SBC to handle video processing and compression, even if it’s just a Zero 2 with an ethernet hat. But if you go full Pi 4 you can run native ROS so that would be much more practical.

Depending on what you need (range? speed? agility? endurance?), looking into different standard Etehrnet protocols might be a good idea, e.g. 100BASE-T1 which can do 100Mbps over just one twisted pair which would mean a very light tether and better manoeuvrability. The max is just 15m though, I think the BR tether can do 300m by boosting up the voltage with a custom converter.

Practically all Ethernet devices support automagically dropping down to at least 100BASE-TX if they only detect 4 wires, which is already 50% lighter than your standard ethernet cable. I wish I’d known that years ago…

Not sure how they hold up in the long run, but I can say that regular analog garden hose pressure sensors work perfectly down to at least 19m of salt water :grin:

MEMS IMUs are always a pain in terms of quality control, but I’ve been using the Nano RP2040 Connect boards with the LSM6DSOXTR quite a bit recently and I can say they’re decently reliable. And if you want to splurge on a proper magnetometer, the RM3100 is probably the best under $40. Each motor will throw out random magnetic fields sideways, so it’s a good idea to mount the IMU as far away from them as possible.