Autopilot response to waves

[rgleason]

With Flir or Lidar, Nvidia + software is now quite capable of fitting a mesh to reality, in real time. It would not take too much detail to give an autopilot some better sense for what is coming. Consider where autonomous vehicles are now.

It could become one of the next new fancy gadgets that that some people have to have on their boat.

[Bill O]

Give any of those to Sean to develop software. He has had this idea for years.

[rgleason]

Yes, I now recall the discussion, it seems much more possible to me now, but there would be more gear involved hanging of the mast. I wonder if it would provide a significant advantage, but I expect it would be used by large freighters to get better performance first.

[Bill O]

It would be a significant break through in AP steering. Heck w/the large freighters, I would like it for our boats!

It would be great if anyone wanted to donate one of them to Sean to facilitate progress towards a fully autonomous AP.

[RaymondR]

I've spent a good deal of time pondering on the problem of poor autopilot performance in following, quartering, seas and have decided that in order to hold a straight course two things are required, anticipation and the very rapid application of counter helm such as that possible with tiller but seldom with wheel steering.

My experience with course keeping when manually steering with a tiller is that one feels the degree of acceleration from the passing wave and applies the appropriate counter with the tiller. Consequently it appears that the use of the cheap accelerometers mow available this scheme of compensation might be applied successfully.

The rapid application of helm problem is highly dependent on vessel design. Both the vessels I have owned have been long keeled cruising boats without the balance a spade rudder can be designed to have. This makes the steering fairly heavy, particularly at speed in a following sea.

My autopilot is of the hydraulic drive type with reversing DC motor which would require a much larger motor in order to provide the speed required to effectively apply counter helm being mindful that the amount of counter helm which to some degree probably depends upon the speed of application. A continuous running motor with an accumulator and directional control valve system would probably prove more effective.

[Rucksta]

There are APs that perform well in a following or quartering sea.
Some have AI with predictive algorithms, auto variable response rates,
gust recovery and apparent wind compensation due to mast head movement.

Try a B&G H5000 system.

I can helm the boat better - for about 30 min.

[RaymondR]

Originally Posted by Rucksta

There are APs that perform well in a following or quartering sea.
Some have AI with predictive algorithms, auto variable response rates,
gust recovery and apparent wind compensation due to mast head movement.

Try a B&G H5000 system.

I can helm the boat better - for about 30 min.

The fact that you can helm the boat better indicates that further development is required.

[Bill O]

Originally Posted by Rucksta

There are APs that perform well in a following or quartering sea.
Some have AI with predictive algorithms, auto variable response rates,
gust recovery and apparent wind compensation due to mast head movement.

Try a B&G H5000 system.

I can helm the boat better - for about 30 min.

Which ones have true AI (artificial intelligence)? I'm not aware of any commercially available APs w/true AI. The B&G definitely doesn't have it.

[wingssail]

Originally Posted by RaymondR

I've spent a good deal of time pondering on the problem of poor autopilot performance in following, quartering, seas and have decided that in order to hold a straight course two things are required, anticipation and the very rapid application of counter helm such as that possible with tiller but seldom with wheel steering.

My experience with course keeping when manually steering with a tiller is that one feels the degree of acceleration from the passing wave and applies the appropriate counter with the tiller. Consequently it appears that the use of the cheap accelerometers mow available this scheme of compensation might be applied successfully.

The rapid application of helm problem is highly dependent on vessel design. Both the vessels I have owned have been long keeled cruising boats without the balance a spade rudder can be designed to have. This makes the steering fairly heavy, particularly at speed in a following sea.

My autopilot is of the hydraulic drive type with reversing DC motor which would require a much larger motor in order to provide the speed required to effectively apply counter helm being mindful that the amount of counter helm which to some degree probably depends upon the speed of application. A continuous running motor with an accumulator and directional control valve system would probably prove more effective.

In the first place I wonder if an exactly straight course is required. If the boat yaws on a wave then yaws back, who cares?

Secondly, I think a better steering solution is a good wind vane and a boat with a responsive spade rudder. If there are big following seas that usually (not always) means enough wind to sail. If there is enough wind to sail then there is enough wind to use the windvane. The windvane corrects quickly and pulls the tiller (with a short lever) and the rudder snaps the boat back on course. I steer with a six foot tiller. The windvane effectively steers with a two foot foot tiller. This puts a hellava load on the windvane structure but I've reinforced my monitor and it does fine. So, if we accept some yawing, our course with the windvane is very satisfactory.

None of this is by accident. We knew we wanted wind vane performance from the get go. Of course the fin keel and spade rudder were always in the plan. And when we put the boat together for cruising we made sure nothing was installed on the back of the boat to interfere with the windvane. Of course an arch and all the crap that goes with that were never planned but the windvane is further insistence that we never install such devices or structures back there in its way.

All if this goes way back to the original planning for a sailboat. Sailing performance is primary. Safety is of course number one, but the boat must sail superbly. So that is the way we put it together. Nor did we disregard comfort. In our view our boat is the most comfortable possible with a smooth ride and easy motion. So, we have it all, including the wind vane.

Our autopilot sees little use. It does not need AI.

Photo: I've posted this before, but if it can steer in this, off Cabo de Vela, it can steer in anything.

[seandepagnier]

Originally Posted by RaymondR

My autopilot is of the hydraulic drive type with reversing DC motor which would require a much larger motor in order to provide the speed required to effectively apply counter helm being mindful that the amount of counter helm which to some degree probably depends upon the speed of application. A continuous running motor with an accumulator and directional control valve system would probably prove more effective.

A continuous running motor will consume more energy, especially in lighter conditions when it can spend a lot of the time not moving.

Originally Posted by wingssail

In the first place I wonder if an exactly straight course is required. If the boat yaws on a wave then yaws back, who cares?

Well, some people care, some don't. You may care some times depending on the point of sail. It really depends.

Secondly, I think a better steering solution is a good wind vane and a boat with a responsive spade rudder. If there are big following seas that usually (not always) means enough wind to sail. If there is enough wind to sail then there is enough wind to use the windvane. The windvane corrects quickly and pulls the tiller (with a short lever) and the rudder snaps the boat back on course. I steer with a six foot tiller. The windvane effectively steers with a two foot foot tiller. This puts a hellava load on the windvane structure but I've reinforced my monitor and it does fine. So, if we accept some yawing, our course with the windvane is very satisfactory.

So you are yawing. This translates into reduced overall speed because of added drag as well as less straight course. Some faster boats cannot have an acceptable course from a wind vane on many courses. Some boats can only pole out the headsail if the course is kept tight enough.

It is often good to have both for redundancy and to complement each other if the autopilot is not very fast and powerful.

All if this goes way back to the original planning for a sailboat. Sailing performance is primary. Safety is of course number one, but the boat must sail superbly. So that is the way we put it together. Nor did we disregard comfort. In our view our boat is the most comfortable possible with a smooth ride and easy motion. So, we have it all, including the wind vane.

Our autopilot sees little use. It does not need AI.

Sure, you do not need it, but you also do not need fiberglass or a fin keel. You do not need dacron sails, or absolutely anything plastic or electronic.

An autopilot is much cheaper and lighter than the wind vane. It cost maybe 10% as much and weighs also about 10%. So you could buy and carry 10 for the cost of a wind vane.

An autopilot can be tweaked much more easily to adjust the parameters. Few (if any) wind vanes can change the gain etc, and even if they do, this is one of many parameters such as vane, size, foil size, and all different angles as well.

There is no (known) windvane design that can steer perfectly in changing conditions and you would have to adjust many parameters at once which is not practical. An autopilot even without AI is much easier to adjust parameters.

Autopilot has the potential to anticipate and react faster, it can steer based on gps/compass/wind or even by a camera. It can use the compass and gps to help it steer better to the wind. It can use gyroscopes to avoid overcompensating from a wave motion. None of these sensors are typically utilized by a wind vane.

[wingssail]

Sean, you have some errors in your response:

Originally Posted by seandepagnier

...So you are yawing. This translates into reduced overall speed because of added drag as well as less straight course.

Please recall that the discussion was about a more powerful or responsive autopilot, possibly incorporating AI in its programming. So it is very likly that such an autopilot will be making powerful corrective actions in order to keep the boat on a perfectly straight line. Seems to me that this will add more drag than allowing a boat to follow the waves a bit more and yaw a little.
...
Some boats can only pole out the headsail if the course is kept tight enough...

So I guess your point here is that some boats could not successfully carry a poled out headsail unless they had the ability to keep a perfectly straight course? That is not my experience. We sail with poled out headsail, using the windvane, quite a lot and don't find it impossible, and that is with an old IOR race boat which is not known for its straight running in heavy winds. I think if we can almost all boat could do it, and we certainly can.

An autopilot is much cheaper and lighter than the wind vane. It cost maybe 10% as much and weighs also about 10%. So you could buy and carry 10 for the cost of a wind vane.

I think this is incorrect Sean, you do get carried away at times, don't you Sean? To get an autopilot which can handle big following seas in a 20,000lb boat would cost significantly more than 1/10 the price of a good windvane. I think the prices might be closer to equivilent.

An autopilot can be tweaked much more easily to adjust the parameters. Few (if any) wind vanes can change the gain etc, and even if they do, this is one of many parameters such as vane, size, foil size, and all different angles as well.

There is no (known) windvane design that can steer perfectly in changing conditions and you would have to adjust many parameters at once which is not practical. An autopilot even without AI is much easier to adjust parameters.

Autopilot has the potential to anticipate and react faster, it can steer based on gps/compass/wind or even by a camera. It can use the compass and gps to help it steer better to the wind. It can use gyroscopes to avoid overcompensating from a wave motion. None of these sensors are typically utilized by a wind vane.

Well, Sean, we're sailors and we are accustomed to making adjustments as conditions change.

Modern autopilots are improving dramatically as is seen in the major ocean racing classes. I think many sailors would love to have such equipment on their boats (me too) but those autopilots are not cheap or simple and more importantly, they won't make a full keel heavy cruiser going 5 knots sail (or motor) in a perfectly straight line which is where we started this conversation.

My point was, (and still is) a well designed boat with a good responsive rudder, using a wind vane, is a better solution then trying to develop an autopilot that could avoid the yawing the OP was objecting to. Perhaps you disagree?

Photo: sailing downwind across the southern Indian Ocean using the Monitor and poled out jib. Winds up to 25-30 true, Log bookpages: Log Book Pages: Sumatra to Mauritius

[seandepagnier]

Originally Posted by wingssail

Sean, you have some errors in your response:
Please recall that the discussion was about a more powerful or responsive autopilot, possibly incorporating AI in its programming. So it is very likly that such an autopilot will be making powerful corrective actions in order to keep the boat on a perfectly straight line. Seems to me that this will add more drag than allowing a boat to follow the waves a bit more and yaw a little.

I was responding to your use of windvane. An AI autopilot can potentially do many useful things including minimizing power consumption and wear and maximizing surfing by steering with the waves.

So I guess your point here is that some boats could not successfully carry a poled out headsail unless they had the ability to keep a perfectly straight course?

Not my point.

That is not my experience. We sail with poled out headsail, using the windvane, quite a lot and don't find it impossible, and that is with an old IOR race boat which is not known for its straight running in heavy winds. I think if we can almost all boat could do it, and we certainly can.

To clarify, with an autopilot the sail can stay poled out much farther off the wind than with a wind vane which can only do it on a smaller angle. It also greatly depends on the sea state. The autopilot gives more options, but I certainly sailed many thousands of miles on wind vane with poled out headsail.

I think this is incorrect Sean, you do get carried away at times, don't you Sean? To get an autopilot which can handle big following seas in a 20,000lb boat would cost significantly more than 1/10 the price of a good windvane. I think the prices might be closer to equivilent.

Do you know about the autopilot I develop? Several people have sailed long passages across pacific from uk to new zealand, from japan to philippines and many many passages. Several claimed that my pilot works better than b&g autopilot.

I did not say anything about 20,000lb boat, but in fact, I know someone who used a power steering motor from a car on 42ft boat $60 from junkyard and sailed in southern ocean so total cost of $240 for complete autopilot system but you can get this figure down to about $100 in total if you build it yourself from the free designs and load the free software as many have.

The fact remains that material cost for a wind vane exceeds the cost of an autopilot, unless you make it from say galvanized steel or plywood and glass. It is possible to home make a wind vane for the same cost as an autopilot, but all the builds I have seen did not have the performance of even the typical monitor.. and even if you achieved it with improved tolerances it is a considerable amount of work to build one. My friend will soon be offering kits however, so the cost then becomes few hundred, but this is plywood and glass and generally useful for inshore and coastal.

the autopilots that cost thousands really only cost $50 for the electronics: they are ripping everyone off with a marine industry scam.

Well, Sean, we're sailors and we are accustomed to making adjustments as conditions change.

It is just impossible to make many of the adjustments the autopilot provides on a wind vane. Most wind vanes only have a few adjustments, vane size, vane angle, and helm offset. You can add in adjustments for vane to pendulum gain, pendulum to rudder gain, and adjust the vane stops, but you end up with a convoluted system that requires changing several things at once. Everyone just accepts the S course that wind vanes steer on broad reach in following seas, and it's not exactly optimal. It doesn't work well at higher speeds either.

Modern autopilots are improving dramatically as is seen in the major ocean racing classes. I think many sailors would love to have such equipment on their boats (me too) but those autopilots are not cheap or simple and more importantly, they won't make a full keel heavy cruiser going 5 knots sail (or motor) in a perfectly straight line which is where we started this conversation.

Nothing is "perfectly straight" but the gyro input allows correction before deviating off the wind. The wind vane does not react to the wind until there is already an error. It cannot utilize the compass and gps to improve sailing to the wind etc.

My point was, (and still is) a well designed boat with a good responsive rudder, using a wind vane, is a better solution then trying to develop an autopilot that could avoid the yawing the OP was objecting to. Perhaps you disagree?

I agree with a good responsive rudder, but disagree that a wind vane is a better solution in many cases. It depends on the particular sailor, boat and intentions. It is like wind vs solar power. Mostly solar panels make more sense and wind generators can be nice to have and in theory you could have only a wind generator but generally it's not the best solution.

Photo: sailing downwind across the southern Indian Ocean using the Monitor and poled out jib. Winds up to 25-30 true, Log bookpages: Log Book Pages: Sumatra to Mauritius

I made also this passage doing the same, but most of the time I had too much wind to need more than 1 sail.

[goboatingnow]

No wind wave can steer as well as a properly instrumented AP. A wind wane has one control input and AP can have multiple including yaw information , roll information , 3axis acceleration , absolute heading ,speed over the ground and through the water as well as true and apparent wind strength and angle.

High end APs easily out perform wind wanes The fact that an el-cheapo AP doesn’t is simply a marketing decision.

[rgleason]

Raymond wrote:

have decided that in order to hold a straight course two things are required, anticipation and the very rapid application of counter helm such as that possible with tiller but seldom with wheel steering.

I agree with Wingsail, the optimal course may not necessarily be straight. There are many parameters, safety, comfort, speed, power usage, all limited by the equipment and what is possible. So the "Optimal course" is a subjective setting.


One factor to consider is the considerable reduction of speed when using a self steering rig. Yes it is a good backup, and does not use electrical power, but the effective use of a autopilot can shorten a long trip by quite a lot.

[rgleason]

As the OP, the discussion is about possible a way to improve Autopilots with the improved technology from autonomous vehicles which have (it's happening now!) much better sensing devices (lidar & infrared flir), faster brains (nvidia graphics) to create a mesh of reality (waves) and AI algorithms to respond.
It is not a discussion about the pros and cons of self-steering, so please keep this in mind. As a reminder, here is the original post.

With Flir or Lidar, Nvidia + software is now quite capable of fitting a mesh to reality, in real time. It would not take too much detail to give an autopilot some better sense for what is coming. Consider where autonomous vehicles are now.

It could become one of the next new fancy gadgets that that some people have to have on their boat.

Also the conversation absolutely did not start out here

they won't make a full keel heavy cruiser going 5 knots sail (or motor) in a perfectly straight line which is where we started this conversation.

This is just an erroneous statement.