robotboat

Good article in The Economist on the need for Robotboats

2009-12-05T17:03:00.000-08:00

http://www.economist.com/sciencetechnology/displayStory.cfm?story_id=15009907

Test of MkII A Robotboat 24 July 2009, north of Ocean City NJ

2009-08-20T13:02:00.000-07:00

The test was mostly successful. The boat was sailing into the wind trying to make the waypoints. It made waypoint 0 and 1 but did not attain 2. After attainment of 1 the boat was twoed back to shore with the kayak.

The tide was going out in an easterly direction at about 2 Km/hr. (not helpful!)

The wind and boat speed are shown.

The test took about 1.5 hours from off the truck to back on.

-WH

Planning a big test

2009-07-23T14:34:00.000-07:00

Last weekend I looked at launch spots in New Haven Connecticut (USA). The 30 mile route would be almost due south to the north shore of Long Island, NY. The advantage of this test is that its fairly hard to lose the boat to the open ocean and I can take the ferry (from Bridgeport) to meet the boat on Long Island.

Test of MkIIA at Bellehaven Marina, VA

2009-07-14T12:11:00.001-07:00

I did the first water test of the MkIIA robotboat. It sailed well until it ran into a small island. The wind changed and the boat was trying to tack. The tack led straight for the island. Other wise quite successful.

-WH

Progress on controls

2009-07-09T08:33:00.000-07:00

For the MkII A (the yellow robotboat), I am using a netbook with Labview (National Instruments). This is very power inefficient for operational boats, but it is fantastic for de-bugging and operating a prototype. Most of our tests lasted much less than the 12 hour life of the MkIIA, so power efficiency is not important right now. Being able to enable and dis-able subsystems is very useful in de-bugging the code. Labview is robotic control for dummies.

The MkIIA is an upgrade of the MkII. I tested the MkIIA successfully in the parking lot next to AMS in Silver Spring MD

-WH

The latest test

2009-04-01T02:23:00.000-07:00

We tested the boat in the water on 22 March 2009 on the Potomac River in Woodbridge, VA (USA)

We are a company

2009-01-01T08:12:00.000-08:00

We incorporated our endeavor as Autonomous Marine Systems (AMS) this week. We intend to produce advanced robotic boats for the global oceanographic surveillance market.

Recently we developed some more sophisticated manufacturing techniques for the boat. The most exciting of these is the use of 3D printing to make the hulls and sails. This new manufacturing technique allows us to also "print in" the major elements of the control actuators (sail and rudder). The values of this are lower manufacturing costs and better performance.

Steady Progress

2008-11-09T22:03:00.000-08:00

The boat is all painted. We verified the solar panels charged the batteries sufficiently well. The micro-controller works.

The micro controller is being upgraded to an FPGA. The psuedo code to operate the boat is nearly ready.

We may even have a small mission with the government to operate as a ground segment of a prototypical boat tracking device to be used for maritime security.

First test of MkIII Robotboat

2008-06-23T14:13:00.000-07:00

Lake George, NY, USA 21 June 2008

The MkIII Robotboat was transported and assembled successfully. Solar panels powered the battery and the computer.

The control actuators work as expected. The test ran about 24 hours. It was very useful to have the boat dis assemble into small lightweight parts. The low weight (45 pounds) allowed transport by one person.

Next steps include finishing the structures with fiberglass and white paint; conditioning the batteries; adding the navigation code. Then it will be on to more extensive tests.

-Walter Holemans

The Mark III Robotboat

2008-01-09T12:14:00.001-08:00

The Mark III Robotboat will be operational in May 2008. It should be a major improvement over the Mark II (the yellow one shown previously)!

-Walter Holemans

robotboat goes to the adirondaks

2007-09-10T07:30:00.000-07:00

here is some video from lake george

microtransat is getting some press:
from the NY times
the register
and the BBC

Final test of Robotboat MkII at Lake George, NY USA

2007-09-05T15:03:00.000-07:00

Final test of Robotboat MkII at Lake George, NY USA

The testing of the MkII is complete. She has taught us all we she can which, in summary is 95% of the design has to change:

Must have low power reliable computer, no more power hungry laptop running Labview and windows

Must have low power reliable data acquisition as the NI SCXI1000 was power hungry and failed a lot

Hulls need to be streamlined

Sails need to taller and more slender

Need stronger dagger boards

Need ~50W of power from solar panels at full sun (even though steady state load is only 2.5 W)

Add PB100 Weatherstation and delete the RM Young Wind monitor, GPS, and electronic compass (the PB100 has all these built in)

Mass has to decrease from 43 Kg (MkII) to 23 Kg (MkIII)

Delete the "pelican case"

Rudder paddle needs to be centered so it can't back-drive the actuator.

Need to be able to stow the rigid sail so the boat runs with the wind in bad weather

-Walter

robot boat loves kids

2007-07-20T13:01:00.001-07:00

This week we got a new integrated sensor array that is going to provide some major advancements vis-a-vis power consumption & code complication generally. Thank god for RS-232; its so easy! Walter got it working on the first try which is rather astonishing. We've learned so much from the mark II. It is clear that software constitutes much of the remaining challenge. Anyway lest y'all think we only care about Big Money, here are some pics of a trip we took to a charter school in D.C.

Ms. Dolan's 4th grade class was gracious enough to host us and listen as we tried to explain some of the design choices we made.

We just set her up on the street outside of the school.

Walter explains bouyancy. We even got some adults from the street to stop and listen.

2007-07-07T16:34:00.001-07:00

We tested the boat today in a short, 3 waypoint course off of Bellehaven Marina in Virginia (USA). The total distance of the course was about 260 meters. The boat made the first two waypoints but failed to acquire the third without human help. The wind was coming from waypoint 2 directly towards the boat, so it tacked. However once it no longer needed to tack it jibed, but by the time the jibe was complete, the boat had to tack again.

Jibing works. The boat never tried to turn into the wind. The boat ran an hour, so the issues with the laptop crashing in about 10 minutes seem to have gone away. This may be because it has a new battery. The laptop lasts about an hour before the battery is drained. The max speed was 3.5 Km/hr. The wind was about 12 Km/hr. Tack angle was set to 50 degrees off of the desired true angle and the tack threshold was set to 45 degrees.

I have not implemented averaging in actual angle or wind angle yet.

We began designing the MkIII robotboat this week. We'll be redesigning about 90 % of the boat. The laptop and DAQ systems will be replaced with much smaller, reliable and lower power systems.

Software

2007-06-30T11:25:00.000-07:00

We are using vectors in the software to force the boat to jibe when the wind is between the actual angle and the desired angle. We believe jibing is a more reliable means of acquiring the desired angle because this manuever precludes the possibility of stalling if the boat were to try to turn into the wind.

Without the use of vectors the roll over from 358..359...0...1 degrees generates complicated code. Averaging around these values is challenging.

Averaging the actual magnetic angle and the true wind angle will reduce the duty cycle of the actuators. We will impliment averaging on the next test.

-W

Test in Occoquan Bay, Virginia (USA), 09 June 2007

2007-06-11T12:28:00.000-07:00

The dagger boards worked well, though one came off when we were towing the boat. The magnetic variation is erroneously programmed. Still need to improve tacking and smooth out the sail and rudder response.

water test 2

2007-06-06T17:12:00.001-07:00

Test in Occoquan Bay, Virginia (USA), 01 June 2007

2007-06-04T11:19:00.001-07:00

Eamon Carrig and Walter Holemans tested the Robotboat on Occoquan Bay. It is about 20 miles south of Washington DC. Three 10 minute tests were completed.

In each case the test ended when the laptop controlling the boat locked-up.

The data show the center of pressure of the hulls in the water is too far back. A contributing cause may be the mass of the boat which cause the boat to lean back effectively moving the center of pressure backwards. The large moment arm between the center of pressure of the sail and the hulls cause the boat to stall when trying to turn into the wind. The wind is also causing the boat to slide sideways.

The boat is responding too much. The control of the sail and rudder should be low pass filtered to eliminate unnecessary corrections to high frequency disturbance.

The maximum speed was 2.1 Knots in a 7.0 Knot wind.

Summary
The bandwidth issue has been solved. Now side slip and stalling must be fixed.

Actions
1-Side slip and stall can be reduced by adding dagger boards forward of the sail. This will reduce stall and side slip, allow slightly higher velocity-made-good and allow sailing closer to the oncoming wind.
2-Add a low pass filter to the control of the rudder and sail.
3-Fix the laptop so it does not lock-up.

lab test

2007-05-25T08:07:00.000-07:00

Chronologically the test documented below occurred before the water test. but is instructive to see how the boat responds out of the water. Here Tom acts as the laws of physics by rotating the boat as the rudder would, Walter monitors the controlling software, and I fix the absolute wind direction.

As noted in the previous post, the response time for input changes is slow. A newer version of the controlling software will be tested tomorrow morning.

First robotic test on the water

2007-05-19T10:32:00.000-07:00

Today Eamon Carrig, Tom Edwards and Walter Holemans completed the first test on the water. It was at a small lake near Angler's Inn in Potomac Maryland (USA). The winds were about 15 mph but they varied in direction substantially. Two test were completed totalling about 45 minutes of operation. The total distance between the way points was about 1200 feet. The second test was more successful: the boat made one of the three desired way points. The boat was in all cases trying to reach GPS way points. We observed several episodes when the boat moved fast (~3.5 mph).

The power system worked without issue. The laptop ran fine. The SCXI 1600 (power relay to actuators and analog in from sensors) ran fine. It appears the boat has enough electric power to achieve a 1.5 hour mission.

However, at this time the truth is: under ideal conditions the boat barely works.

Several issues became apparent:

1) The software is too slow because it controls sequentially. For example it gets GPS, then controls the rudder, then controls the sail. Each one of these operations can take ~4 seconds. So each cycle of the program is taking ~12 seconds. In that time the wind varies and the boat rotates. Next time we will run as much of the software in parallel so the software cycle decrease to about 4 seconds.

2) The angle of attack is too small. The angle of attack is the angle between the apparent wind and the chord line of the sail. In today's test, the angle was 20 degrees. Next time we will set it to 30 degrees and we'll tighten up the sail angle set point from 4.5 degrees to 3 degrees. Additionally, the natural variability of the wind direction begs the use of a filter...

3) Tacking did not work because the boat stalled as it tried to sail into the wind. Next time when tacking is required, the boat will be programmed to first go with the wind and approach the tack direction from an orientation of high velocity and stability.

4) Setting the sail to zero degrees when the wind is coming at the boat from the direction it is heading is un-productive. It should be set to 90 degrees or perhaps nothing should be done.

5) The actuators are too slow (~4 seconds to reach a set point). In future actuator designs, having the rudder reach its set-point twice as fast will increase control. The torque of the actuators is more than required, so decreasing the reduction of the gears from the motor to the actuator is a readily available solution.

The bottom line: this test was very successful. We learned a lot and should be able to make major improvements

Substantial progress on Software

2007-05-06T19:01:00.000-07:00

We made progress writing the software. The use of the compass (0 to 360 degrees) was challenging. The rudder and sail are controlled. Tacking was also coded. Given desired GPS coordinates, the boat should be able to autonomously sail to the target. This assumes relatively calm weather and no obstructions and a duration less than about 1.5 hours since the batteries won't last much longer. We'll add more duration when we are sure the control system is working.

A connector was added to the cable connecting the electronic enclosure to the wind monitor. This allows us to discoect the sail from the rest of the boat for shipping ad storage. The strenciling was completed.

Next activities are to try out the code in the lab, then in a parking lot and then in the water.

Calibration activities

2007-04-29T12:34:00.000-07:00

This week we calibrated the compass, wind monitor, battery voltage and current. Control system software is being written in "Labview" (National Instruments). The electrical and software efforts are taking much longer than anticipated.

prettiest belle at the ball

2007-04-16T13:39:00.000-07:00

We gave robotboat a fresh coat of paint.
The white on the control box is thermally motivated while the blue accents are purely aesthetic. I got the command modules working for the rudder and sail. The control code is coming along nicely. We should be ready for more waterborne tests soon.

video from the first water test

2007-04-10T08:23:00.000-07:00

Robotboat looks kinda small, but otherwise this video is very instructive.

Yesterday we got the GPS and magnetometer working at the same time as the DAQ system.

First test in water

2007-04-05T15:16:00.000-07:00

Transport via a 1996 Toyota Tacoma went well. We designed it to fit in the back of a small pick-up. I could easily get the boat in and out myself.
The boat was tethered to the dock near Leonardtown Maryland (USA) 5 April 2007. This first test was just to see if all the parts would go together, to see if we could easily transport it, and to verify that it could acquire and save data via the on board computer and power. The boat works well in this regard.

The next steps are to calibrate all the instruments and write the navigation code. We ought to be able to try for a full system test late next week.

Electrical and mechanical assembly

2007-04-02T06:43:00.000-07:00

The yellow fiberglass / styrafoam pontoons were joined to the tubes with fiberglass and epoxy. The rudder (rotated 90 degrees from the design orientation) and mast actuators are already attached to the fiberglass tubes.

The laptop is connected to the SCXI 1600 data acquisition system. Both will reside in the black 'Pelican' suitcase. The actuators, battery, wind monitor and electronic compass are outside the suitcase. Wiring harnesses are pass into the suitcase via plastic compression fittings.

How it will Look

2007-03-28T11:26:00.000-07:00

Here are some graphics. The dimesions are in inches. The estimated weight is 110 lb.

Fiberglass is very messy

2007-03-04T15:47:00.000-08:00

We finished glassing the hulls. Complex shapes have proven to be the hardest to deal with. Tight radii and small voids are very problematic. To make matters worse, filling in with epoxy hasn't worked as well as we had hoped. You really have to cut the glass right the first time. Epoxy sans glass is very brittle and runs interminably - it requires three or four iterations of filling to seal well.

I completely rebuilt the motor assembly:

It looks like it should be a good seal. One thing I have learned since the first time I built this assembly: bonds with a little give are better than stiff bonds (eg. RTV makes a better marine adhesive than super glue). Seems obvious in retrospect, but it took me knocking off a limit switch to learn.

Up next... more G10 machining and sail integration.

sail and rudder actuators

2007-02-01T18:10:00.000-08:00

we started fatigue testing:

next up: more fabrication the hulls aren't quite done yet...

R&D

2007-01-29T21:12:00.000-08:00

So far we have bent a motor shaft, broken electrical terminals, and crushed limit switches. Expensive mistakes. But hey, that's why they call it R&D.

cliff came through bigtime

2007-01-26T14:20:00.000-08:00

we got the motor housings for the rudder and sail.

sail

2007-01-25T16:35:00.000-08:00

Wind power seemed like the easiest way to move robot boat. It is abundant and relatively easy to harness.

ASV mk II

2007-01-24T22:37:00.000-08:00

this week we built the hulls and the control box.
Teej made templates and we cut styrofoam with an electrified nicrome bow (hotwire). Epxoy and fiberglass. The idea is to make robotboat out of materials that could be confused with garbage; to reclaim useless artifacts of modernity and transform them into a system capable of providing unique and valuable information.

the sail was harder... still working on that. when the radius of curvature gets small the precision is very important. Syncronicity is no myth here.