Underwater Autonomous Project Help

[blacksheep]

I'm taking a graduate level autonomous machines course that consists of a completely open ended project to create an autonomous robot that completes a task. The idea I came up with is a submersible vehicle that will drive on the bottom of a kiddy pool (to simulate a mall type fountain) and collect the change from the bottom.

I'm having some issues locating sensors for obstacle avoidance that can sit out in the water -- obviously IR is out of the question, bump switches are only to be used as emergency back-up, so sonar is all that's left. The only sonar unit I can find spec'd for underwater use is $130 and I'm working on a student budget (the board was $100 as it is and motors + motor controllers will already make this project hurt my wallet, hah). I see plenty of sonar units for much much cheaper but I have no idea what on the sensor or connection requires waterproofing.

BTW, the unit will be very small - no more than a 1ft diameter. I've already tested the available tap water on campus and can run DC motors at a low voltage (4.5VDC motor on a 9V battery) without waterproofing (there was a drag issue with the cheapest brushed motor I tested however) so I'm not sure how adverse a reaction other components may have. It also doesn't have to work beyond May, so corrosion isn't a huge factor for me either.

Does anyone here have any experience with development of underwater sensors of any kind or waterproofing devices? As of right now the only idea I have for even sensing the coins is another sonar unit. It seems the miniature UAV community is rather lacking and most devices aren't on such small a scale as my robot. I did some digging and reading through this forum, so I don't think I'm reposting any questions - but sorry if I did.

Regardless, this is a cool little forum! Thanks for any help!

[airman00]

How do you plan to pick up the change? Can you suck it up into one side of the robot and out the other side, and all the pennies and whatever get stuck behind a mesh piece.

Cute idea , i like it

[blacksheep]

As of right now I plan to design the robot sort of like a garbage truck. So there will be a very thin perforated plate infront of the robot with some gate mechanism if needed that collects the change and then it will rotate by some motor to the rear overhead into a removable bin. The bin will have bump sensors to see if the bin gets overloaded so that the motors don't have to move the robot and more than maybe 1/4lb of change or something. Then I'll activate a speaker or LEDs or something that lets the user know to remove the bin from the robot while it's still underwater (a fountain is only 1ft of water anyways). The robot will have a few behaviors during this time that let it know if the bin is reattached, allowing it to continue searching and collecting.

The actual mechanism that collects the change probably won't be well defined until I get to do some testing. It's going to be tough!

I'm more worried about finding change without roaming the entire fountain. heh.

[blacksheep]

BUMP!

I'm still pretty stumped on this. I might be able to use a floating buoy but I have no experience with those fishing bobber sonars available either.

How about metal detection? Could that be set up to go underwater? I'd rather do a sweep though so the robot will know when it doesn't need to look for coins anymore.

Hmmmm 

[fuzzyt]

Although being able to detect coins would be an awesome idea, is it a requirement?  Could the robot map the inside of the fountain and simply run a pattern sucking up change more by chance?  If coin detection is a requirement, how far away does the coin have to be detected?  Maybe some form of metal detector would be more appropriate than sonar.  A sonar ping is usually dependent on the cross section it has to bounce off of.  You won't have much return from a coin.

Then again, I may have completely misunderstood the problem ( wouldn't be the first time ).

EDIT - So after re-reading the post, I misunderstood the problem.  Sorry....

[blacksheep]

Although being able to detect coins would be an awesome idea, is it a requirement?  Could the robot map the inside of the fountain and simply run a pattern sucking up change more by chance?  If coin detection is a requirement, how far away does the coin have to be detected?  Maybe some form of metal detector would be more appropriate than sonar.  A sonar ping is usually dependent on the cross section it has to bounce off of.  You won't have much return from a coin.

Then again, I may have completely misunderstood the problem ( wouldn't be the first time ).

EDIT - So after re-reading the post, I misunderstood the problem.  Sorry....

Actually, you really didn't. lol.

The requirement is in the robot's ability and use of sensors/mechanisms - not so much in how exactly it does it or what the task is (that was determined by me). It most definitely could map the fountain and just go back and forth inside it - but when I developed this idea, the professor running the class was saying that we needed some kind of "special sensor" beyond bump and IR (they're too easy to incorporate). He told me that building a robot that performs underwater would count as my "special sensor", so your idea has much more merit than you knew! I'm just not sure how I would do that.

I'm starting to think more and more about the use of a dc water pump to suck them up instead. It seems like it might be easier than creating a mechanism. I'm just worried about the amps that will be required. heh.