Autopilot - Open Source

[rgleason]

Sean has developed his own autopilot,
https://github.com/seandepagnier?tab=stars

MIT Technology Sept/Oct wrote about Lorenz Meier who has developed PX4, an open source Autopilot for Drones which adapted the affordable motion sensors in smartphones.
https://github.com/PX4/Firmware

PX4 Website: Lorenz Meier, Author at PX4 Pro Open Source Autopilot
Blog: Lorenz Meier's Page - DIY Drones
Youtube Presentation:

https://www.youtube.com/watch?v=Ov65AbtIT8I

We know a good marine autopilot is very much involved in 3D, only the fluid is less forgiving than air, but there are huge similarities. There may be some opportunities for adoption of code both ways.

[rgleason]

https://www.inf.ethz.ch/personal/lomeier/


Mentions boat under airframe.
http://www.proficnc.com/


This one is also fascinating - using Smartphones as 3d scanners
http://cvg.ethz.ch/
Useful for architecture.

[rgleason]

Sean about his efficient autopilot.

I think the most efficient and responsive combination would be the steering portion of a cape horn or moniter, or self made steerer combined with an electronic stick type autopilot like Sean's autopilot with wind sensors at the masthead. The disadvantage of this is the loss of speed due to the extra steering foil. So the best choice really is to have an efficient solar powered autopilot attached to your wheel or tiller without all the added weight of ss wind steering gear at the end of the boat. However in that case you must be sure your rudder bearings and gear is good.

http://www.cruisersforum.com/forums/...d.php?t=191438

[Moo]

Originally Posted by rgleason

Sean about his efficient autopilot.

I think the most efficient and responsive combination would be the steering portion of a cape horn or moniter, or self made steerer combined with an electronic stick type autopilot like Sean's autopilot with wind sensors at the masthead. The disadvantage of this is the loss of speed due to the extra steering foil. So the best choice really is to have an efficient solar powered autopilot attached to your wheel or tiller without all the added weight of ss wind steering gear at the end of the boat. However in that case you must be sure your rudder bearings and gear is good.

CapeHorn compared to Windpilot - Cruisers & Sailing Forums

In the case of windvanes; the most efficient and responsive way to use it, for ocean passage making, is to use it as intended. Properly set up a wind vane steers almost perfectly, if correctly balanced it compensates for the heave of the swell too. My Aries steers in everything from about 5 knots to 50 knots apparent.

Granted in very calm conditions an electronic pilot is great and one attached to the windvane gives it the added power of the servo's lever arm if conditions are sloppy. Inshore an electronic pilot is arguably safer as it can steer a course independent on the wind and currents, i.e. a true course over ground.

Not sure about your comments on the rudder's bearings as unless your wind gear uses a secondary rudder (most don't) the rudder bearings will get the same workout whether being exercised by wind or electricity.

[GILow]

Originally Posted by rgleason

Sean about his efficient autopilot.

I think the most efficient and responsive combination would be the steering portion of a cape horn or moniter, or self made steerer combined with an electronic stick type autopilot like Sean's autopilot with wind sensors at the masthead. The disadvantage of this is the loss of speed due to the extra steering foil. So the best choice really is to have an efficient solar powered autopilot attached to your wheel or tiller without all the added weight of ss wind steering gear at the end of the boat. However in that case you must be sure your rudder bearings and gear is good.

CapeHorn compared to Windpilot - Cruisers & Sailing Forums

I accept that a focus on efficiency is relevant to racers, but for cruisers I would say the focus should be on reliability/redundancy/safety.

In that context an extra foil is a positive due to the redundancy it introduces. (Auxiliary rudder configuration)

Otherwise, I will be interested to see how some of the quasi A.I. challenges have been addressed.

[barnakiel]

I agree on one thing: great effort, we need them, may be an opportunity to use the code both ways.

I will disagree on a couple of things like drag and redundancy of an extra foil (one used for a separate AP foil). None an issue of this thread though.

Also, a boat AP is possibly NOWHERE as complex (computationally nor mechanically) as a plane AP, esp a drone AP. So this code translation and adoption seems more likely in the drone to boat direction.

I think, in a sailingboat, proper implementation of wind data AND availability of cheap, reliable (or at least field-repairable) wind sensor is one of the limiters now. I think I mentioned elsewhere building one. But this happens later, on the Caribbean side, provided I get there. Unless someone builds something sooner.

Great effort. I will re-read and chime in if anything I do or have contributes.

Cheers and my best regards to all developers that SHARE,

barnakiel

[seandepagnier]

Originally Posted by Moo

In the case of windvanes; the most efficient and responsive way to use it, for ocean passage making, is to use it as intended. Properly set up a wind vane steers almost perfectly, if correctly balanced it compensates for the heave of the swell too. My Aries steers in everything from about 5 knots to 50 knots apparent.

Almost perfectly? Is 5 knots low enough? Not for me.

I am not aware of any wind vane that can steer around breaking waves in any sort of intelligent way.

My monitor works also down to 3 knots or less apparent wind, but in swells, it does not handle cross swell very well at all. With significant secondary (or teritary) swell, by this I mean 5 meter or more (obviously primary swell is even more than this) it can be enough to throw the wind vane off completely, even in 15 knots of wind.

Also the wind vane always steers an S course. This is a fact of all wind vanes, and even electric autopilots to some extent. This limitation limits it's windard performance compared to an intelligent autopilot or human steering by a few degrees, more if there is swell.

Granted in very calm conditions an electronic pilot is great and one attached to the windvane gives it the added power of the servo's lever arm if conditions are sloppy. Inshore an electronic pilot is arguably safer as it can steer a course independent on the wind and currents, i.e. a true course over ground.

Not sure about your comments on the rudder's bearings as unless your wind gear uses a secondary rudder (most don't) the rudder bearings will get the same workout whether being exercised by wind or electricity.

Not exactly true. Since the electric autopilot can steer a straighter course, and use less rudder movements, it may wear the rudder bearing less. Most rudder bearings can last a very long time though so it should not normally matter much.

Originally Posted by GILow

I accept that a focus on efficiency is relevant to racers, but for cruisers I would say the focus should be on reliability/redundancy/safety.

In that context an extra foil is a positive due to the redundancy it introduces. (Auxiliary rudder configuration)

Otherwise, I will be interested to see how some of the quasi A.I. challenges have been addressed.

Yes, well for redudancy sure, I would not want to sail without a vane, and a bungee (sheet to tiller) for backup. I also currently have 4 complete autopilots.

Last night, I was in drifting conditions in the gulf of maine, and my autopilot was able to keep the boat much closer to the wind. It was difficult to feel the wind at all, in fact, the sails were slack a lot of the time.

I was moving 0.5 knots through water. The wind was technically zero, but because of the tidal current, I felt a wind. The power consumption of the motor average only 0.2 watts to do this. The wind vane cannot do it, there is not enough water speed to move the paddle, and not enough wind to move the air blade (the anenometer cups didn't even rotate)

Using the wind vane in this situation as I have many times before, relies on the natural balance of the boat more than any corrections from the vane. I can usually balance but not as close to the wind, but also often the boat will spin around eventually leading me to give up and drop sails until more wind appears.

Originally Posted by barnakiel

I agree on one thing: great effort, we need them, may be an opportunity to use the code both ways.

I will disagree on a couple of things like drag and redundancy of an extra foil (one used for a separate AP foil). None an issue of this thread though.

Also, a boat AP is possibly NOWHERE as complex (computationally nor mechanically) as a plane AP, esp a drone AP. So this code translation and adoption seems more likely in the drone to boat direction.

I think you are wrong about this. It is a different complexity. The plane AP does not deal with sea state or waves. In flat water, the boat autopilot is simpler as it is essentially only one or two control loops where the plane has many.

With waves, would be like a plane AP that can deal with turbulence to it's advantage, can perform dynamic soaring (like albatross) and so far I am not aware of any drone AP gliders that can cross oceans this way, but in theory they could.

I think, in a sailingboat, proper implementation of wind data AND availability of cheap, reliable (or at least field-repairable) wind sensor is one of the limiters now. I think I mentioned elsewhere building one. But this happens later, on the Caribbean side, provided I get there. Unless someone builds something sooner.

I have developed a semi cheap wind sensor:

weather sensors - pypilot store

It is working, but below a few knots of wind, not great, I steer to compass in this case.

What sensor are you building?

[rgleason]

Awesome Sean! weather sensors - pypilot store
I hope this is the first of many innovative sailing tools!
-- May I make a reference & description in the Manual Hardware Section?

--Here https://opencpn.org/wiki/dokuwiki/do...re_temperature

[Moo]

Originally Posted by boat_alexandra

Almost perfectly? Is 5 knots low enough? Not for me.

I am not aware of any wind vane that can steer around breaking waves in any sort of intelligent way.

5 knots is not low enough hence my comments about an electric autopilot if low wind conditions. I agree that an electric pilot generally steers a straighter course, but does the amplitude of a windvane's course matter? Sure it will add distance and time to the voyage but how much? The course changes are small so the effect on VMG is small.

I am not aware of any electronic autopilot that can sense and steer around breaking waves either. Are you just throwing this in to stir the pot, making a flippant comment, a joke; or are you making this claim for your pilot? I'm not being funny - I am trying to find out where you are coming from with that comment. A pilot that can sense, or anticipate, and then steer around breaking waves would be quite something.

Cheers

[seandepagnier]

Originally Posted by Moo

5 knots is not low enough hence my comments about an electric autopilot if low wind conditions. I agree that an electric pilot generally steers a straighter course, but does the amplitude of a windvane's course matter? Sure it will add distance and time to the voyage but how much? The course changes are small so the effect on VMG is small.

Mostly true, but at one point, I measured a loss of 0.2 knots boat speed made good compared to the autopilot at the same conditions. This was because of the very large S steered by the vane in trade winds when sailing around 5 knots wing and wing. So only losing 5 miles a day, but this is significant.

Also, it is possible to poll out the head sail at a wider range of wind angles, so in this particular course you will gain a lot more

I can improve this loss by tuning the vane, mostly by limiting the angle the air paddle can move, in this case it would only be 1 or 2 miles lost. This difference would vary considerably depending on the boat, vane, wind and waves.

I am not aware of any electronic autopilot that can sense and steer around breaking waves either. Are you just throwing this in to stir the pot, making a flippant comment, a joke; or are you making this claim for your pilot? I'm not being funny - I am trying to find out where you are coming from with that comment. A pilot that can sense, or anticipate, and then steer around breaking waves would be quite something.

Cheers

It is not yet at this point able to "see" waves with a video camera, but the theory is possible. Just steering straighter makes the boat much less vulnerable to breaking waves.

My bad experiences with the vane, the boat gets knocked from one wave, and the next breaker hits the boat from the side causing damage.

[Moo]

Originally Posted by boat_alexandra

It is not yet at this point able to "see" waves with a video camera, but the theory is possible. Just steering straighter makes the boat much less vulnerable to breaking waves.

My bad experiences with the vane, the boat gets knocked from one wave, and the next breaker hits the boat from the side causing damage.

It would need 360 degrees of vision. I am guessing that the camera would look for white water on the face/crest of the wave? Or would you envisage it visually analysing wave shape/size/pattern? Combined with the inputs from the motion sensors of course.

I agree with you latter point. This can also happen with a human who is not fully awake I have steered(relied) on autopilot, on a couple of boats in winds up to 70knots and provided the pilot is powerful enough they can do a good job. Not much more to do than tweak the gain settings on the older pilots the newer ones sort themselves out.

I haven't yet had to rely on windvane in winds much over 40, gusting around 50 perhaps, but cocking about changing wind sails etc.. definitely looses out pushing a button in the heavier conditions.

One area where an autopilot scores over a vane is when making sail changes as messing around with the sails upsets the balance sometimes. With the pilot just set it to steer whatever wind angle you need for the duration of the sail change and get on with it.

[rgleason]

Wouldn't motion sensors be able to determine a change in wave pattern?
Wouldn't motion sensors be able to determine primary, secondary and tertiary wave patterns (because they each have their own signatures)?

We need to look into the math behind waves. There must be someone who has studied this and tried to represent it mathematically.

There may be a way to determine the basic characteristics of each of those wave patterns (perhaps not the isolated giant, or the variation, at first)

Much like electrical monitoring programs can figure out cycling loads and user loads and with enough data be able to determine which loads are from the refrigerator, air conditioning, lighting, etc.

Admittedly, crew movement and wind would be other factors to consider.

[seandepagnier]

Inertial sensors can feel sea state, but cannot see waves before they reach you.

Try sailing with your eyes closed compared to open.

[Moo]

Originally Posted by rgleason

Wouldn't motion sensors be able to determine a change in wave pattern?
Wouldn't motion sensors be able to determine primary, secondary and tertiary wave patterns (because they each have their own signatures)?

We need to look into the math behind waves. There must be someone who has studied this and tried to represent it mathematically.

There may be a way to determine the basic characteristics of each of those wave patterns (perhaps not the isolated giant, or the variation, at first)

Motion sensors are already integrated into autopilots and they monitor wave amplitude and the boats attitude using this data to dynamically adjust rudder input and gain settings - I have not seen the code, this is by observation. (I have spent thousands of hours observing autopilots!!!)

Modern off the shelf autopilots have a lot less controls and require a lot less calibration, then those of a few generations ago; which frustrates some. They do learn and adapt to the current conditions and do it quite well. In terms of the rate, amplitude and duration of rudder input as the boat lifts to a wave. (they also automatically work out things like counter rudder etc..)

No amount of analysis is going to help an autopilot to steer in real sea conditions so I don't think there is much need to worry about mega complex analysis of wave theory. I would not use an autopilot that relied on theory.

The pilot can't disregard secondary, tertiary or giant isolated waves because that is the real world - hence cameras to enable it to see what is coming.

[rgleason]

Ok. Ok. You guys just jump on me. Seeing before it arrives is certainly better, but what DSP is going to analyze the magnitude, speed and direction of just a single wave, let alone a bunch of them?
Maybe Raymarine and Flir are working on that right now?

Sean, doesn't your AP have sensors? What does your program do about sensing waves?