Wednesday, August 26, 2015

Improving Odds of Robot Crossing the Atlantic

After building and learning what we can from our 12 foot solar robot boat it really seems like it would be fund to send it across the Atlantic.  There have been a number of other attempts to send small home made robot boats across the Atlantic but it seems so far all have failed.  So maybe we could be the first!

Our boat is not self righting.  If it turns over that is probably the end of our attempt.  There is some chance that another big wave tips it back over.   With the daggerboard down in the water it is more stable in the upright position so it could end up that way even after tipping a few times.   I am sure many people reading this think we will never make it across the Atlantic without the ability to self right.    I think we still have a good chance.    Let me try to explain my thinking.

First it would really take a breaking wave to flip this boat.  This is something that comes from storms.   If we are going from around 18 degrees North in the non-hurricane season we can mostly avoid storms.   If we can send the robot new waypoints over a sat-link to avoid a storm it may never run into a storm.

The boat is 4+ feet wide and has enough stability for a sail.  Without a sail and the leverage that gives the wind, it would not tip over easily.  Also, the battery or batteries will be very low in the boat.

I think the main danger will be at the crest of a large wave.  The phone has motion sensing so we could write code so that it knows when it is coming to the crest of a wave.   It could then point the boat into the wave and slow down to minimize the risks.  We will generally be headed into the wind so most of the time we will be pointed into the waves.

If one motor breaks we can still use the forward/reverse capability of the other motor to control which way the boat is pointed.   With the wind mostly blowing the boat back toward Anguilla if the boat is pointed the right way the daggerboard should be enough to bring the boat home.  So even if a trolling motor fails we should get the boat back and be able to try again.

We have played with vision software on our phone enough to recognize a yellow tennis ball inside our house.   Recognizing a large patch of brown seaweed floating on a blue ocean is probably something we can do.  So we may be able to avoid large patches of seaweed.   A small clump on the surface probably does not stop us.  The propellers are down in the water a ways as is the daggerboard.

We could put in lots of batteries and try to run the motor all night or we could have the boat turn sideways in the night and just drift slowly.

Some of the things that could still go wrong:
   1) Phone or other critical electronics could die
   2) Both trolling motors could fail
   3) We could run into seaweed in the dark when our vision could not help us
   4)  Camera can get crusted over with salt and the robot go blind

There are websites that report seaweed locations.  We might be able to avoid the seaweed just by going when there is not much seaweed.   In the 20 years I have been in Anguilla most of the seaweed I have seen is in the last year.  I think in another few months we might be back to normal.

We can probably work out something to keep the camera from going blind.

There are still challenges to solve but it seems possible.

Satellite Data Link Ordered

We will test our 12 foot solar robot boat around Anguilla using cellphone Internet but it really would be fun to try to send this boat across the Atlantic to Africa.  For that we will need some sort of satellite data link.  It would be best to test the sat link early on as well.  So I have looked into it more and ordered one.

At a minimum we want the robot to report its location but we would also like to report more than that.  Things like wave height, tipping angles, current speed, battery level, etc.   Also, if the robot can receive data we could send it new waypoints if it was headed for a storm or something.  We could also decide it was having trouble and should just head home so we can improve things and try again.  It seems much better if we can both send and receive data.

There is Iridium Go that gives you a wifi hotspot with real Internet.  This would work great with our Android phone.   The problem is the device is around $1,000 US and the data service is like $150/month.   This is out of our budget range for something we might send off into the ocean and never see again.

There is Spot which uses GlobalStar.   Spot says they have good coverage for data in the Atlantic.   GlobalStar says they don't have voice coverage off of Africa where we might be headed.    But GlobalStar does say their data coverage is good.   Spot can only send messages and not receive.   There are a number of Spot devices easily found on Amazon including ones that work with a cellphone.   So the big drawback is just that it can not receive.

There is Orbcomm which has sats in LEO (low earth orbit) so a small device can reach them.   We can get service contracts for $17/month and only $7/month when we are not really using it.  We can send and receive data.   It is not realtime as the data may sit on a sat for a bit till it is near a groundstation or our boat but that is fine for us.

I have ordered a Quake Q1000 modem to work with Orbcomm.   It has an RS232 port so I also ordered an RS232 to Bluetooth converter.   With this I think we can send and receive data from our Android phone over satellite.   It was hard to find a retail outlet for these and I ended up finding one on Ebay for about $100 US.

Monday, August 24, 2015

Catamaran Model

Our first robot boat was having trouble with wind, so we made a catamaran model that is lower and has less for wind to push on.    Once we realized there was a 12 foot sailboat in the basement, we  started making plans to make that into a robot (see previous 2 posts).  So this model is on the back burner.  

Saturday, August 22, 2015

Circuit Drawings

Our motor controller has outputs for 2 motors.   It gets power from USB.  The phone talks to this over bluetooth.

Here is how we will wire from one port on the controller through the relays to the trolling motor.  This is called an H-Bridge.  When the Forward/green wire is on the battery is connected to the motor red-red and black-black.  If the Back/purple wire is on then they are switched red-black and black-red.  The controller should never have Forward and Back on at the same time.  To be safe we will probably have the software always go to stop between Forward and Back so both are off for a bit during transitions.

This is a drawing of the inside of the trolling motor.   I have some hope that if the speed controller is on max that the input power goes directly to the motor but we may have to open it up and bypass the speed controller.  I am told the red and black going down the shaft are the key wires to the motor.  Not sure what the Yellow and White wires going down the shaft are for.

Since there are 2 motors we have to do the above wiring twice, but even so this is really simple electronics for a rather capable robot boat.

Being able to run motors both forward and backward will give us extra steering power at slow speed and also let us control which way the boat points even if one motor fails.   With the daggerboard if we do our testing up-wind from home the boat could drift home even if one engine fails.  It is upwind all the way to Africa from Anguilla.  There is usually good wind, so drifting should be reasonably fast.

We can do smaller upwind tests to Scrub Island, Flat Island, and St Barts. 

Wednesday, August 19, 2015

Island Boys of Island Harbor

My boy's Grandpa Smitty gave them some land next to the beach at Island Harbor.  Thanks Grandpa!   This is a wonderful development environment for boys and boats.

We are going to make an old Laser sailboat hull into a solar robot boat big enough to ride on or pull Kayaks.  This is 12.5 feet at the waterline.  We ordered 2 trolling motors.  We will use the same phone software we wrote for the small solar robot boat.   We also ordered some solid state relays big enough for the trolling motors.  We already had a bluetooth motor controller that we will use to control the solid state relays.   Our phone software uses internal GPS and compass to decide what commands to send to the motor controller over bluetooth.  Since the phone and motor controller are already working we should be able to make a big robot without too much more work.  More about our plans after the pictures below.

We had too many times where the 4 foot boat would have trouble with the wind.  On the days were there was not wind there was usually clouds, so the solar did not work and the small battery could not last long.  Then there was an amazing amount of seaweed in the Caribbean these last 9 months and the little boat propellers get tangled up easily in even a small amount.  I am hoping the trolling motors have a prop design that does not get tangled so easy.  Being bigger, more powerful, and deeper in the water should mean small bits of seaweed don't bother it.    Making computer vision software to let the robot avoid big patches of seaweed seems possible but avoiding tiny pieces probably would not have been.   Spinning backward or alternating back and forth may be a way to get seaweed off if it does tangle.  With the acceleration sensors and GPS we may be able to tell when we run into a thick bunch of seaweed and backup.    If this seaweed keeps coming we will have plenty to experiment on.

The next size Quadmaran will use 4 of these same trolling motors, so working out how to control them is a step toward that goal.  The computer vision will work for that as well.

We should be able to make the trolling motors go forward and backward using 4 relays for each motor to make an H-bridge but to do that we probably have to open up the trolling motor.   We might start with forward only.

If we can make the motors go forward and back then even when one fails we can still control which way the boat is pointing.  With a daggerboard and control of direction we can get the boat to drift back to home port if it is upwind when there is a failure (there is almost always a strong trade wind around here).   So our testing will be done upwind from home base.  From Island Harbor we can send the boat along the coast and be able to drift back.   This would also work if we are sending it across the Atlantic to Africa as that is an upwind trip from Anguilla.  Fault tolerance is something we want in future designs as well.