homemade autopilot

Did the same. For a hands on steering you need the proportional acceleration as Rick described.

As for the steering vein, you need to dampen the direction sensor's input to the control so it operates on an average value. With electronics it is quite easy to adjust this sensitivity.

 

You would not find that using the steering wheel felt more natural if you used a proper autopilot.

They steer far better than a crewman can.

 

yes, specially following a pre programmed course. on the pilot a had done on the say blinker arm there were 3 speed options
recall using the quik turn mode was constantly adjusting and dangerous at speed tho
motor was in the interface between dash and brought the wheel nicely some inches up. added gps nmea gmdss (or whats the name)
and navsoft that even kept a log. had serial ports wired to the dash and inside for the laptop (can nowadays be bluetooth)
never understood why both drives in the bilge had to hooked back tho. think they stopped this model but it worked well

 

Like I said in post #9, it is on my list but not on top.
A new system will be very different from the 1th because the power steering is history. There will also be no need for small correction: with the stern drives the boat refused to stay on course, but now she runs like on rails.

For this winter the planning is to lift one engine and overhaul it, plus some minor corrections around the tunnels to get a bit more planing surface. When that is ready it will be spring again....

 

I did some initial design for a tiller pilot for a saiboat. If you have good skills in programming and a bit in electronics, I think that would be doable.

I would have definitely bought the drive unit. Here is a very good candidate: http://www.linak.com/corporate/openpdf.aspx?pdffile=/corporate/pdf/english/la36_eng.pdf
It costs ~400$ with a potentiometer for posiotion and tolerates salt water. Another option is to use an old commercial tiller pilot.

For steering reference I would have used a rate gyro (these can be as cheep as 10$) and COG from GPS. I would think a fluxgate is not necessary, but I don't know any commercial autopilot without it.

For "brains" I would have used a microcontroller (ARV or ARM) or a small Linux platform (e.g. slug = nslu2). These can be programmed using C, which is the most famialar one to me. Costs is about 100$. Additionally you need some rather simple electronics to control the motor of the drive unit.

At the moment the project is on hold, since I noticed that the newest version from Raymarine already has a gyro and is quite reasonably priced (~1000$): http://www.raymarine.com/ProductDetail.aspx?SITE=1&SECTION=2&PAGE=87&PRODUCT=4173

Here is an interesting article of the demands for a good autopilot: http://www.marinea.fi/downloads/Ocean_Racing_12_2006_Autopilot.pdf

I have recenty bought a Raymarine ST2000+ and I'm very unsatisfied with it's performance. You can get it working for one conditions by adjusting gain and damping, but then it won't work in the next condition. Either it just keeps doing very small back and fort movements, when a much bigger adjustment would be needed or then it oscillates the course +-20 degrees. All of these even without any waves and bot on engine and while sailing.

 

I have almost finished a board with a compass chip and GPS on board. The intent is to use it for auto steering as well as instumentation. I will probably have it made some time next year, the PCB manufacturere are closing down on Monday. Can you believe it, the year is on it's way out !

Just to brag a bit. If all goes well a friend and myself may be going fishing in Moz... I'm getting extraction symptoms just thinking about it

 

Not knowing the inards of the current crop of autopilots can't really comment on them, but i believe the DIY's out there are looking at the problem incorrectly. Coming from a Aviation background, the levels of autopilots start from very simple system at heading/altitude/speed hold, to next level with stabilty arugmentation to fully fled hands off take off and landing systems. It is quite amazing watch a helo hover just off the ground in 45 knots gusts and not moving more than a foot and watching the rotorblades move all over the place compensating.

This is done through a stabilty argumentation system, which i believe is the first step in designing an autopilot for a boat. Reason for this is a boat has to sail off course before the effect of the wave is felt. If you use a stabilty argumentation system, the sensors feel the wave motion and adjust for this even before the boat sails off course.

When the a particular militry jet (sorry don't want to name it due to security laws) first came out, it didn't have a GPS system fitted, it was totally independant to outside input for its nav solution. The pilot would enter the Lat/Long on startup. A advance autopilot system on a boat wouldn't use a GPS as the only navigation solution, but use it to compare its computed possition with what the GPS is stating, that way it has the ability to sence an error and alert the crew.

 

In aviation things are quite different Dave. The autopilot in a plane has to maintain speed, altitude and course and must do all that with a fast response time. It has at least three sets of rudders and throttles to control, but it also can get input from systems like ILS.
In a boat there is just one rudder, the only complicating factor is the unpredictable effect from the waves.

I think that for recreational navigation a simple system with course input from a compass and position from a GPS will perform satisfactorily. On long trips it makes life a bit easier for the helmsman, like cruise control does for a driver; expectations beyond that are unrealistic.

 

Also the changes in wind speed and direction has an unpredictable effect on steering. Especially on most modern boats wind gusts have a violent effect on steering.

At least on my boat Raymarine ST2000+ works far from satisfactorily. On the other hand the system pilots with rate gyro on my friends boats work extremely well.

 

I think you should qualify this statement. From my admittedly very little experience I find this to be a huge problem on monohulls requiring simultaneous action on both rudder and mainsheet while it is not a problem at all on my catamaran installed with semibalanced rudders.

 

CDK: True, aircraft require a faster responce compaired to a boat.

ALL: Listening to people complain about their installed systems suggest to me that the designers haven't look at the problem properly and only provided a bandaid solution. Joakim stated in his comment that his raymarine is not the best, yet the other system with a rategyro works better, the key here is that that system uses something other than a fluxvalve and a GPS, so it has a stabilty argumentation system built in.

I would use accelorometers, put one in the front, one in the rear. If the wave hits the bow and pushes it, the front accl would fell this while the back one doesn't and the autopilot would compsenate. For a current/wind drift to the side, both accl's would feel the same and the autopilot would then command the boat to crab to compensate. This is all without the use of a compass and gps, your boat already becomes more course stable. If you where in a tight channel with a strong cross wind, the SAS would keep you from drift out of the channel and running aground.

 

I wouldn't be sure about the faster response need of an aircraft. I have never steered one, but watching it doesn't seem like very fast responses are needed. Steering some sailboats do need quite fast responses in hard conditions in order just to keep the boat going even somewhat straight.

There are quite different level of autopilots even from one brand. The cheepest one suitable for my boat would costs ~200$ and the best ones ~5000$. Certainly these are different. The one I have (~500$) seems to be a quite old design, but it probably works just fine for many boats.

I don't know of any model that uses accelerometers. The standard seems to be a fluxgate compass in all models and additionally a rate gyro in the top models. The top models also have input for speed of the boat and sense the rudder position.

Rate gyro measures the angular acceleration (I think all autopilots use just one axis: yaw). The fluxgate output needs a lot of damping due to accelerations and thus rapid changes can not be detected with it. Rate gyro will give accurate short term heading, but it needs to be corrected due to cumulative errors in heading integration.

 

You have a very weak arguing position! I flew anything from a single prop Piper to military jets, believe me it's quite different from a sailing boat.

 

I am currently talking to the developers of YBW OpenPilot (which doesn't have anything to do with my OpenPilot project) about inclusion within the OpenPilot framework. Their system looks to be pretty well developed, so it might be worth looking at it.

http://sourceforge.net/projects/ybwopenpilot/

Cheers,

Tim B.

 

Get onto youtube and watch an F-18 land on a carrier, whatch the back stab's, they will move up and down quite alot for very little aircraft change, alot of that is gain control so there is some pilot input. But a good fraction of that is the stabilty arugmentation system.

Rate gyros measure angular change around an axis, while accelorometers measure movement along an axis. So without any hull dynamics calulated into the solution a rategyro can only tell you how fair you have twisted off your heading, but with accelorometers it can tell you how far you have moved off your track, but using distance measure, and time taken to get to that distance you have your rate of twist. So with accelorometers not only have you got the same infomation that a rategyro would give you, but also the amount of error required to get back on track

Tim b: if you want any help understand autopilots feel free to pm me

Dave