I have started a project to build an underwater rover to take a look at what lies beneath the surface.  I've been interested in using technology to see what is on the bottom of the lakes, and have looked at many options including side scan sonar and even fish finders.  Instead I wanted to make something from scratch so I could drive around and record video of what the rover finds. 

Yes, you can use a GoPro to look at video under the water, and there are many other waterproof cameras and UAV's/Rovers already made, but this will be a fun learning experience to see what it takes to make it all work.

I will be posting my progress, pictures and video as I go.

12"x12"x4" Watertight electrical box with 0.25" thick walls and a thick rubber gasket.

GIGABYTE GA-E350N-USB3 AMD Dual-Core E-350 1.6GHz Mini-ITX

4GB DDR3 1066 RAM

500GB Seagate 410mA Hard Drive

Microsoft Lifecam Cinema 720P USB Webcam

Microsoft Lifecam VX-3000 (internal monitoring)

180 Lumen LED Light (Flashlight cut off)

Windows 7 (Remote desktop)

VirtualDub (Video Capture/Monitoring)

CAT6 Tether for communications

Pico PSU-80-WI-32V DC to DC power supply

2 x 12 V Sealed lead acid batteries ( 2.5" x 6" X 3.75" )

JULY 2011 - Design and build

The internal acrylic frame and external body frame was designed in AutoCAD first so I could work out the best placement and fit for the components.  Plus it helped me build a materials list to make it all work.

The internal frame was built using a 12"x24" sheet of 0.25" acrylic which is pretty strong and easy to glue together.  I paid around $18 for that piece and $5 for the front 3.5"x10.5" piece which I picked up earlier from P&A Plastics here in downtown Hamilton.  The Acryweld glue and needle applicator was a few extra dollars.  The glue melts the pieces together and flows like water between the joints.  To glue the pieces together you just line the two pieces up where you want them to be then run the needle along one side of the joint and the glue will seep under to fill the void and permanently weld it together, it sets in a minute or two and takes a bit longer to cure fully.

I drilled out the holes on a drill press before I glued everything together to make it easier, and used a router with a straight bit to cut out the notches for the bolts after everything was finished.

The batteries I used were from a UPS I had bought for $100 because it was fairly inexpensive to get a charger, batteries, an inverter and a way to monitor the status of the batteries through USB on my laptop.

As for sealing the front panel to the case I tried a cork gasket, but i couldn't get the face tight enough to stop leaking so I just used clear silicon caulking and tightened it in place with the bolts creating a seal that was about 1/16" thick.

JULY 23, 2011 - Arrive at Little Hawk Lake on vacation

Today we arrived at the lake for vacation, and once everything was unpacked I took out the rover for some initial testing.

First I removed all the components from inside to start testing as I didn't want to risk damaging anything right off the bat.  I added rocks inside the box as ballast to replace the weight of the batteries and other parts.  I also added the lead ballast I had made earlier to the bottom of the frame with tie straps.

Once everything was put back together and sealed I put the unit in about 2 feet of water.  After 15 minutes I took it out of the water and opened it up, and there was no water!

JULY 24, 2011 - Second day of testing

Today we, my brother-in-laws Mat & Tim and myself, conducted a few more tests at a deeper location off the dock.  First we attached a rope to the frame in order to lower it down.  I also had a test piece of CAT6 attached to the box through the waterproof connector which was a few feet long.

Once we were ready we lowered it down to a depth of 33 feet which really surprised me as we were only about 30 feet off shore on the dock.  The lake at its deepest point is over 300 feet deep!

As we lowered the rover you could feel it getting heavier and sinking at a faster rate than at the surface.  After leaving it on the bottom for a minute or so we brought it up to the surface.  when the rover was close to the surface we noticed a few air bubbles coming from the CAT6 cable, after opening the case we found a small amount of water (5ml) inside. 

We figured that the water had entered the box through the cable end that was cut off and open to the water, didn't really think that one through.

As a result I attached the complete CAT6 cable from inside the rover to the surface to eliminate any water from entering through the cable end.  I also attached a secondary rope line to the frame and taped everything together every few feet to form a single tether cable.

LOST TRACK OF THE DAYS - DEPTH TEST 2

We attempted a second depth test at 33 feet.  We didn't notice any air coming from the box through bubbles in the water.  As the rover sat on the bottom we could feel air being forced out of the end of the CAT6 cable. This concerned us as we figured the box must be filling with water, so we brought it to the surface.

After opening it up there was still only trace amounts of water inside. So we were puzzled as to what was going on.

Next we decided to put a GoPro video camera inside with its waterproof case in order to try and capture what was happening.

DEPTH TEST 3

Same test at 33 feet, upon surfacing we found the same amount of water inside as expected.  Upon reviewing the footage from the GoPro we realized that the box was being compressed so much from the pressure of the water that it was compromising the seal around the edges.

We never realized that this would happen as I calculated that the pressure at that depth was ~15psi. The part I forgot to calculate was that the pressure was applied to 'every' square inch of the box, resulting in a top surface pressure of around 2160 lbs!

The formula for the pressure calculation is:

psi = ( depth(ft) * 15(psi/atm) ) / 33 (ft/atm)

psi = ( 33ft * 15psi/atm ) / 33ft/atm

psi = 33psi

DEPTH TEST 4

We wanted to see how much the box compressed this time, so we left the camera inside to capture the footage and we also placed a semi crushed pop can inside so that it would crush further and remain crushed to tell us the total 'crush factor'.

With this test we measured an approximate 'crush factor' or around 1.5". So internal reinforcement is a must to keep this box in one piece.

Next I put the batteries and the acrylic internal frame back inside so that it would provide a little more support to the lid and bottom during the descent.

DEPTH TEST 5

With the batteries and frame inside and the GoPro pointing outside the box, we lowered it for another test at 33 feet. We couldn't wait to see in the water at that depth, so the GoPro captured one of the videos linked on the side.

The box only leaked slightly this time as the extra internal support really helped, but we still noticed a deflection of around 1".  With the external footage it seemed the visibility was around 15 - 20 feet and surprisingly there was plenty of light from the surface.

We were pretty excited to look at the video to say the least...even though it wasn't from the actual rover, but from the GoPro.  The guys were giving me a hard time about just using the GoPro instead :)

DEPTH TEST 6

This time we tested the same setup at around 10 feet, and there were no leaks.  So I finally replaced the rocks with the full setup of the motherboard and camera.

DEPTH TEST 7

Tested at ~10feet with all components capturing video and streaming live through a remote desktop session. No leaks occured and I was a happy camper!

We continued tests up to around 18+ feet for over 30 minutes with only a few drops inside and capturing lots of video. I made a few adjustments to the camera, angling it down so that I could get a much better view of the bottom.

From the night test, using the light which was placed about 3 inches off center from the camera we found that it was illuminating too many particles in the water to see much of anything.  The video is available on the sidebar.

Adjustments to the camera for night shooting were done manually.

From testing and trying to make it all work here are a few of the observations made:

1. Boot problems everytime the battery was removed.  (Maybe bad memory. Not sure yet)
2. Video capture sometimes jittery @ 1280x720 30fps MJPG
3. CPU during 1280x720 30fps capture sat at ~ 55%
4. Preview rate in VirtualDub only ~8fps
5. Internal temperature without the light hovered at 125 F
6. Idle operating temperature 105 F (Without lid)
7. With the light on the system was still stable at 150 F
8. Batteries last for around 4 Hours
9. Full battery charge 12.7 V
10. Light draws 410 mA @ 5 V
11. Draw from batteries was ~1.2 A @ 24 V

1. DONE-Build a better support structure for internal frame
2. Internal sensor for water
3. Relays to control lights
4. Flexible wrap/sock/protector for tether to protect the cables
5. Add 2 lbs of ballast to front of frame for proper buoyancy
6. Multiple cameras for better view
7. Mount lights externally under and above frame
8. Add thrusters for driving control
9. Build C# application to control and receive/log data from rover
10. Possible CO₂ pressurization system for greater depths
11. Addition of the Polulu Mini Maestro 24-Channel Controller from HERE

Using a square case is not ideal for deep water exploration as the pressure applied is too great, so the next version will be built using an acrylic cylinder.  Lighting also needs to be more powerful and placed below/above and away from the camera.  Video captured during daylight hours produced great visibility up to ~15 feet, with the sharpest and clearest objects being within 5 feet of the rover. Adding thrusters to control the rover will also provide a much better way to get footage instead of 'hanging' the rover from a rope.

Added 180 lumen light to the right of the camera