Underwater Vehicle Project

Hello all,

I am currently working on a final year project about an AUV (Autonomous underwater vehicle). I am facing difficulties due to my lack of knowledge in electronics. (I am a mechanical student by the way.)

The purpose of my project is to come up with an AUV that detect different frequencies and move towards the selected frequency. I have a budget of 400 usd.

I've had a look online. My initial plan was to have 4 active transmitters in the north-south-east-west directions with the AUV having the passive transmitter. Is there any device (similar to sonar) to use other then transmitter?

I would like to incorporate the following: -DCCduino / CH340 Arduino NANO -Motion sensor to avoid obstacles -Datalogger to record sensor responses -Propulsion via 4 x 12V DC motors

I'm thinking of using legos to build the AUV.

To sum it up, I need some general advice mainly in the configuration portion and arduino coding. (Is MATLAB feasible?)

I deeply appreciate any suggestions for this project. Thank you for your time.

To be honest, this project would require quite a lot of specialist materials...from scratch for $400 a unit is a push!

Need clearer spec list before anyone could advise.

I've heard good things about BlueRobotics.

I think they use brushless motors on their thrusters. I've read brushless motors can be used underwater without needing special waterproofing. I've read you just want to make sure you rinse the motors off after they've been used.

Xena: My initial plan was to have 4 active transmitters in the north-south-east-west directions with the AUV having the passive transmitter. Is there any device (similar to sonar) to use other then transmitter?

What are these devices you're calling "transmitters".

I imagine the type of "transducer" you want to use depends on the frequency you're interested in monitoring.

What is a "passive transmitter"? Is it like an unpowered microphone?

I'm interested in hearing what others suggest. I think I recall seeing some interesting underwater robots. I'll try to find these projects and add a link to them.

Have you checked out DIY Drones? I believe they have an underwater section to their forum.

Pardon me, it should be receiver instead of passive transmitter. I need something that could provide frequency, preferably 4 different frequency in all directions. AUV being a receiver, could sort out the different frequencies and move towards the selected frequency.

I am aware the budget is extremely tight. Would it be better if I cut down to just one DC motor and place it at the back? Speed is not an issue so long the AUV could reach its destination. The basic specification is to detect frequency and lead the AUV toward the selected frequency. Other specifications could be add on to improve the AUV, but i think it would be hard considering the budget.

Thanks for the recommendation. I have tried on their discussion board.

How about placing sonar beacons in all 4 directions and a sonar on the AUV??

Although it is easy to say "move towards the selected frequency" (you mean toward the sound source), this is in practice surprisingly difficult, and in many cases, nearly impossible if sound absorbing and reflecting obstacles are nearby.

To identify the direction to a sound source you will need at least two microphones forming a phased array, and sophisticated signal processing algorithms. If this is an essential aspect of the project, develop the technique on land first.

Try it yourself. While your eyes are closed and one ear is plugged, have a friend set up a single speaker emitting a simple tone and see if you can tell where the sound is coming from, and whether you can move toward it. Keep in mind that this is a survival trait!

The part of this that interests me is the direction finding. Assuming you locate some good, low cost underwater mics (or adapt some simple mics to the purpose), you could place one mic on each side of your vehicle. I have started with a simple, very cheap electret mic, and three op amps, and created a very versatile amplifier/filter which converts the incoming frequency to square waves. Square waves are generally easier to work with for measuring frequency, and phase.

So when you determine that the frequency you are interested in is present, compare the two mics for signal strength, and turn toward the weaker one until the two are equal. ( At this point the phases of the incoming signals should be the same also.) This may require some baffling to make sure a signal from one side is adequately weaker in the opposite side mic, and of course, careful calibration of the mics.

The orientation of the vehicle now might be directly toward or directly away from the signal, but you would discover that quickly as you began to move.

This approach I believe to be well within the computational ability of the Atmega 328 arduinos.

Theoretically, the above approach sounds simple, but it is impossible to convert a signal with four frequency components into a "square wave".

This sounds like you are building a torpedo that can track towards an enemy ship with a specific sound signature whilst ignoring friendly ships with their own specific sound signatures.

To navigate within the water column without breaking the surface you'll need a water rated pressure sensor. Forget commercial pressure sensors, the cost will make you shake your head in disbelief. Make your own using a diaphragm and microswitch.

To avoid bumping into things you'll need a forward facing sonar something like this http://www.hawkeyeelectronics.com/depth-finder/ This will consume 25% of your budget.

Making 4 sound sources to emulate ships could be as basic as driving 4 unbalanced electric motors at different speeds.

The AUV seeker is a single microphone, amplifier and arduino measuring the amplitude of a specific frequency.

For propulsion use four 12v brushless centrifugal fountain pumps like this. http://www.ebay.com.au/itm/12V-DC-Brushless-Magnetic-Drive-Centrifugal-Submersible-Amphibious-Water-Pump-/201125676966 Steer using differential thrust instead of rudders and through hull mechanical linkages.

Build everything else using fibreglass/epoxy. Forget Acrylic, UPVC and solvent welding. it's just no good for rapid prototyping.

Program the torpedo to move in a random direction that that is weighted towards whatever direction gives the loudest signal at the specific frequency.

Oh my, this seems difficult. How about this: For auv, 1 x DCCduino / CH340 Arduino NANO 1 x Accelerometer 3 x 12V brushless DC motors 3 x propellers (xyz directions) 1 x receiver 1 x waterproof case Lego to build

3 x transmitters (xyz direction). Although the frequency is weaker, but is less complex than sonar.

Some background reading for you.


Xena: Lego to build

I'm a huge fan of Lego bricks and I use some Lego pieces in my robots but it's been a long time since I've tried to make a robot with Lego pieces. It's just so much easier and less expensive to use PVC tubing, plywood, aluminum "L" beam, Polymorph, foamed PVC and lots of other building materials than it is to make robots from Lego pieces.

Lego robots are very hard to keep from falling apart.

Xena: 3 x transmitters (xyz direction). Although the frequency is weaker, but is less complex than sonar.

What are these "transmitters"?

What you may want to do is start with a surface boat with a "sonar" package underneath. If the "detect different frequencies and move towards the selected frequency" is the critical part of the project, then figure out that part first before worrying about motors, housings, and such. May save you $$$ in the end.

What is this about impossible to convert to a square wave? I built a "smoke detector detector" by using an electret mic and purpose built amplifier. The smoke detector output was a 6 KHz., 110 db. signal. On the oscope, it looked about like a sine wave, but that's not really important. Converting it to a square wave took a simple opamp comparator circuit with an adjustable threshold, which allowed me to ignore noises a good bit lower than the smoke detector level.

The square wave, fed to an interrupt pin is easy to use to determine frequency with high accuracy. I would collect 100 square waves, check the total time, and compute the frequency. Gives very high accuracy.

Now once I've got the frequency, how do I do direction finding? The speed of sound in water at about 70 deg. F is about 4800 fps. So a wave length of, say, a 3 KHz frequency is about 15 inches. Two mics on a 1 foot stick, aimed right at the frequency source would produce two square waves right in phase. If I fed these two waves into a difference amp (simple circuit to build), I would get a null output. Now if I turn the stick a little, say ten degrees, one mic might be about 1/2 inch closer to the source, the other 1/2 inch further away. That difference of 1 inch at a wavelength of 15 inches would kick those two waves apart by 24 degrees so that my difference amp would register it own square waves, with a duty cycle related to the phase difference. So follow the difference amp with a peak detector, and when the peak is above zero, you're not turned the right way. There will be a little work involved in creating an efficient direction finding algorithm, but it isn't that hard.

When you try to equate this problem to phased-array antennas, you are over-theorizing it. A phased-array antenna is trying to put a maximum signal level at exactly one place, generally in three dimensions. Direction finding in two dimensions is far less complicated. I worked at Harris Corporation many years ago, and we pioneered in the phase-array business.

The comparator to create high/low signal is very easy to implement. Could also take the 2 outputs and treat like rotary encoder outputs and process that way.

What is this about impossible to convert to a square wave?

A simple sine wave can be converted into a square wave, for example by detecting zero crosses, but I maintain that a wave summed from four different components at different frequencies, such as the OP proposes to detect, locate and drive toward, cannot be so converted.

Here is a quick example: a sound wave consisting of equal amplitudes of frequencies 42, 33, 21 and 9, phases of 0.

How do you propose to use the “square wave” technique to detect, locate and drive toward the source of the signal with frequency 33?

Can perhaps spread the frequencies out a little and add bandpass filters so only one frequency is reacted to.

Hi, I agree with CrossRoads, spread the frequencies out, non harmonically related. Use bandpass filters to detect them. The position may have to be either an array of directional microphones that are sampled in turn, or a single microphone that is mechanically scanned sensing the maximum signal and direction.

Tom..... :)