Affordable Underwater Distance Sensor Problem

[BaBaBeet]

Hello everyone.

I'm working on a small AUV (8kg) for a project at my university. I want to be able to determine the distance from the AUV and underwater objects around the AUV (say in a 120 - 180 degree arc infront of AUV) to a range of 3m to 5m. Please note that the AUV only need to operate in one depth plane, ie only lateral information is needed of the terrain. I would also prefer if the sensors are relatively cheap (maybe not much more expensive than about $100 - $200). I would also prefer the sensor to be mounted on the AUV, although this is not a must.

Basically what I've thought of using so far:

• hacking a laser rangefinder or building my own using a laser and webcam. Hacking a relatively expensive rangefinder in the hope that it might work seems silly to me and I'll rather do as little as possible building-work myself.
• Using InfraRed sensors. I understand that the IR band is absorbed heavily by water. I don't think IR will be able to give me distance sensing up to 5m (I might be wrong?)
• Hacking a Fishfinder. For the same reason as the laser rangefinder and that I'm not too sure about the resolution of the sensor I'll prefer not to persue this option.
• Waterproofing and modifying ultrasonic sensors like the MaxSonar as suggested by Admin ( http://www.societyofrobots.com/robotforum/index.php?topic=10435.0 ), however other posts have stated that it might not be worthwhile ( http://www.societyofrobots.com/robotforum/index.php?topic=7383.0 ). I am actually quite interested in this option as it falls nicely in my price range and I would like feedback on this idea if its possible (especially from Admin??!)

I also thought of mounting the sensor on a stepper motor or servo to enable it to sweep 180 degrees. This should ofcourse also be able to work underwater - I've seen a few posts on underwater servos and waterproofing your own on this site ( http://www.societyofrobots.com/robotforum/index.php?topic=485.0 and http://www.societyofrobots.com/actuators_waterproof_servo.shtml ). Can I also get feedback on this if its possible?

About the AUV and the testing environment I'm using:

• its a small ~8kg submarine Neptune SB-1 ( http://www.tiger.com.tw/product/5220-F.html )
• Used by the previous student to implement path following using rudder control and been modded somewhat.
• I am planning on testing the AUV in our university's water tank - I can give you the exact details if you need it, however I'm estimating that its dimensions are: 10-15m wide, 5-7m deep, and very long (50m?). If this is insufficient I can find other testing tanks which are larger, however, this would not be preferable.

Please give any feedback or comments so that I can atleast have somewhat of an idea what I'm getting myself into or whether this idea is a bit too ambitious...

Please Note that I'm actually more interested in an "quick-fix" method, the measurements do not have to be too detailed as this setup is only needed to practically demonstrate my project (my project is actually on the obstacle avoidance and resolution part and not on the sensing part.) However, I would prefer it to work somewhat nicely atleast.

Sorry for this long post - hopefully I mentioned everything needed to give you a vague idea of my situation. Thanks in advance for any help/suggestions.

Regards Hendrik

[corrado33]

Just a quick post before I go to work, but have you thought about sonar?  There are plenty of ultrasonic range finders on the market, and dolphins use it for a reason (it works).

[BaBaBeet]

AFAIK underwater sonar sensors are too expensive for my price range (please correct me if I'm wrong - would much rather buy a manufactured product than hacking/building my own tacky equivalent). Because of this reason I'm leaning more to buying and waterproofing a MaxSonar Ultrasonic sensor, but I have no idea how feasible this idea is other than the fact that Admin mentioned himself toying with the idea (see link in OP). thanks

[Admin]

The home-waterproofed MaxSonar that I did worked really really well. I tested it in a large cylindrical old nuclear test plant vat of water that's 50 feet deep and 50 feet diameter (5 stories both directions). The MaxSonar was able to "see" the entire tank. The sonar cone was like 10 degrees wide, if I remember right. Be aware that pings will reflect off the water surface. Also remember the speed of sound underwater is different, so your microcontroller will need to do the proper math to convert the range.

[corrado33]

AFAIK underwater sonar sensors are too expensive for my price range (please correct me if I'm wrong - would much rather buy a manufactured product than hacking/building my own tacky equivalent). Because of this reason I'm leaning more to buying and waterproofing a MaxSonar Ultrasonic sensor, but I have no idea how feasible this idea is other than the fact that Admin mentioned himself toying with the idea (see link in OP). thanks

Buying and modifying was exactly what I was thinking of.   

[BaBaBeet]

Just knowing that something like that is possible sets my heart at ease I'll do some further research before making a final decision though (the South African Rand doesn't convert into $100 per sensor that easily ), although I believe that this might be the best option currently available to me. So please - other suggestions and ideas are still welcome!

A couple of questions:

• Is the home-waterproofing method a permanent solution if done correctly, or will the level of "waterproofiness" deteriorate / stop working after a while (for the sensor and the method described in http://www.societyofrobots.com/actuators_waterproof_servo.shtml for the servo)
• How many sensors do you recommend I should buy (in case my first few attempts at waterproofing fails)? My study leader is in America atm and it will be cheaper if I can buy everything now via him (no oversea shipping fees etc.)
• How idiot-proof is the waterproofing method (sort of ties in to the previous question)
• would you recommend me using a different testing tank (the performance of the sensor underwater sounds IMBA ), will a tank with the current dimensions maybe affect the readings too much?

Thanks again for all the help I really appreciate it (for both of you)! (Sorry for some of the silly questions - I feel I should probably ask before just jumping into a solution willy-nilly)

Hendrik

[Admin]

Waterproofing was very easy (much easier than with servos). I had one last me for many many hours, than it stopped working . . . not sure why, but perhaps because I left some underwater wires exposed . . . my second one is still going strong.

Using geometry and knowing the cone angle degrees plus the dimensions of your water area, you can determine the max measurement range before pings strike the water surface and water bottom.

[corrado33]

Admin really knows what he's talking about when it comes to underwater stuff.  If you read through enough threads you'll find out he works with underwater bots for part of his living. 

Question for either of you, if you know how deep you are, and you are using sonar, can you just ignore the pings that would come back from reflecting off of the surface since you'd be able to calculate how long it'd take for the ping to get there and back?

[Admin]

Admin really knows what he's talking about when it comes to underwater stuff.  If you read through enough threads you'll find out he works with underwater bots for part of his living.

Sorta . . . my robot fish is very non-traditional, meaning much of what I do isn't entirely applicable to a typical submarine type underwater robot. It took us two years to reliably stop the water from leaking in all the time

Question for either of you, if you know how deep you are, and you are using sonar, can you just ignore the pings that would come back from reflecting off of the surface since you'd be able to calculate how long it'd take for the ping to get there and back?

What happens when the object in front is the same exact distance that the robot is from the surface?

In my situation, my robot fish doesn't go more than 5-10 feet deep in the 50 foot tank. As such, I point the sonar downward at a ~20 degree angle so it rarely ever reflects off the water surface, yet it can still see stuff in front. It does on occasion pick up strange echoes when it's close to the surface (I never figured out why), so I wouldn't trust any sonar values as accurate if it's less than 1.5 feet deep.

[corrado33]

What happens when the object in front is the same exact distance that the robot is from the surface?

I thought about that, but how often will you be EXACTLY (within sensor error) the same distance away from an object?  And besides, if you're moving, the object will be getting closer while the surface will be staying put.  Sure you'll have a blind spot of whatever your sensor error is, but that's nothing another sensor (pointed downward maybe) wouldn't fix. 

Can you get sonar that works at different frequencies so you can have multiple running at the same time?

[Admin]

Can you get sonar that works at different frequencies so you can have multiple running at the same time?

MaxBotix, to my knowledge, only sells sonar at the same frequency. You can however do whats called sonar ping patterning, or put one on a rotating motor.

[BaBaBeet]

Thanks for the replies. I've decided to order a few MaxSonar sensors.

Admin: is there a reason why you chose the compact version of the sensors (WRC vs WR)? Does the directing-horn of the WR series sensors make it harder to waterproof, or did you choose the WRC series because of less water resistance and size?

[Admin]

They only had one model that was water resistant at the time, so that's what I chose.

[BaBaBeet]

Ok so I've ordered the MaxSonar sensor. Just a quick question: Admin mentioned in the one post that he used a soft RTV to coat the transducer. I also see the two RTV's that Admin recommends for waterproofing (the DC 1345 and 1340) are rated as 50 and 32 A hardness. Are these too hard for what I plan to do (my limited knowledge of such matters are that you would want to match the water's density as much as possible and 32 A might be a bit too much).

What RTV would you recommend or would the Dow Corning be fine?

thanks

[Soeren]

Hi,

It took us two years to reliably stop the water from leaking in all the time

Not trying to jack this thread, but while sodastreaming a bottle of water for my scotch, got this thought... If there's room aboard, A CO2 cartridge (they come in various sizes) with an electronically controlled valve and a similar valve in the hull, could be used to adjust the pressure going down and back up, to match the surrounding pressure and hence loosen the demands on waterproofing.

[Admin]

Dow Corning engineers can help you choose a different RTV. Just send them an email.

That's how I selected mine.

[corrado33]

Not trying to jack this thread, but while sodastreaming a bottle of water for my scotch, got this thought... If there's room aboard, A CO2 cartridge (they come in various sizes) with an electronically controlled valve and a similar valve in the hull, could be used to adjust the pressure going down and back up, to match the surrounding pressure and hence loosen the demands on waterproofing.

Wow, I love that idea!  It'd be a good emergency assent thing too.  You could even use it as emergency propulsion.  **Imagines mini sub jetting out of the water and flying hundreds of feet into the air** lol