Hybrid self steering

S

Slycat

Well-Known Member
Joined
15 May 2009
Messages
253
Visit site
Hi all,

Been using my Raspberry pi a lot recently and impressed with what it can do and had a thought..

For self steering , I'm assuming one of the most difficult elements is accurately translating change in apparent wind to steering input. Rather than use a wind vane I was wondering if a hybrid solution would be possible.

It would work something like this....

Raspberry pi uses GPS to monitor boat heading
As the heading veers it powers a small servo to adjust a mini rudder on the back of the boat
This rudder both moves left/right and rotates (like a regular wind vane setup) so the small left/right movement is translated into a stronger rotational force
The stronger rotational force move the tiller (again like a regular wind vane).


I'd always thought the sticking point in a pi auto steer was translating the course correction to the tiller as large servos would be needed, in this way the force is amplified for us by the mini rudder being forced up/down.
 
I wouldn't have thought that GPS derived heading data was accurate enough or responsive enough to change in direction.... you might get something like this to work though with a gyro compass and depacking the NMEA data and using that....it in effect is what Hydovane do when they stick a tiller pilot on their wind vane (albeit using a servo pendulum, rather than connection back to the main rudder)

Interesting idea though!
 
Last edited:
Just a thought, but isn't the point of a wind-vane system that you steer to the wind direction, not a compass course? For steering a compass course, I'd have thought that the investment in the machinery you're suggesting would be considerably greater than the cost of a tiller-pilot, which will do the job efficiently and with a degree of "intelligence" regarding response to waves etc.
 
AntarcticPilot said:
Just a thought, but isn't the point of a wind-vane system that you steer to the wind direction, not a compass course? For steering a compass course, I'd have thought that the investment in the machinery you're suggesting would be considerably greater than the cost of a tiller-pilot, which will do the job efficiently and with a degree of "intelligence" regarding response to waves etc.
Click to expand...
This might have a lot more power than a tiller pilot though....
 
There are plenty of compass sensor IC's sold for connecting to Arduino boards. I assume you could use a Pi.

The problem with all these things is the delay time in response often causes the feedback loop to be unstable.
I think you need to be clear what you want from a system in terms of response time, and do the sums for delay in the loop.
There are many good books on control theory (and some very dull ones..) weighing down bookshelves, including mine.

If you just want to hold a course within ten degrees while motoring in flat water, it's easy.
IF you want to adapt to waves and so forth, using a lot of rudder, it gets very technical...
It would be nice to have direct control of the loop parameters sometimes.
 
lw395 said:
There are plenty of compass sensor IC's sold for connecting to Arduino boards. I assume you could use a Pi.

The problem with all these things is the delay time in response often causes the feedback loop to be unstable.
I think you need to be clear what you want from a system in terms of response time, and do the sums for delay in the loop.
There are many good books on control theory (and some very dull ones..) weighing down bookshelves, including mine.

If you just want to hold a course within ten degrees while motoring in flat water, it's easy.
IF you want to adapt to waves and so forth, using a lot of rudder, it gets very technical...
It would be nice to have direct control of the loop parameters sometimes.
Click to expand...

Agree entirely; feedback loops for control systems are NOT simple. Commercial systems have solutions built in; it would take a lot of trial and error to get it right in a home-brew. Not impossible; I'm sure there are people on here with the right knowledge and experience, but before starting you need to know it isn't as simple as just moving the tiller to react to a change of direction.

Also, there are intrinsic issues in steering to a compass direction for long distances. The wind direction and strength will almost certainly change, requiring changes of sail-trim and/or changes of course. A system steering to the wind direction will keep the boat sailing without attention, but of course will require attention to the course steered. Point is that the latter is probably more desirable to a long-distance sailor, while the former is more useful to coastal sailors like me.
 
Last edited:
Yes good idea , agree compass heading better than GPS, I'd assume the PI could get that data from somewhere.

I think the problem with tiller pilots is that they draw a large amount of power over long distances? The pi draws very little and the water would be doing the heavy lifting.

Now I just need find a 'not working for parts' wind vane listing on ebay... ;)
 
AntarcticPilot said:
Agree entirely; feedback loops for control systems are NOT simple. Commercial systems have solutions built in; it would take a lot of trial and error to get it right in a home-brew. Not impossible; I'm sure there are people on here with the right knowledge and experience, but before starting you need to know it isn't as simple as just moving the tiller to react to a change of direction...
Click to expand...

You're absolutely spot on there! Back in about 1995 a colleague and friend made from scratch, getting stainless steel parts fabricated to his design a hybrid self-steering system. It was a bit like an Aries in that it had a pendulum servo with ropes to the tiller, with a small electric servo motor turning the 'rudder' of the pendulum servo. The control electronics used a PID (Proportional - Integral - Differential) algorithm implemented in software, and a flux-gate compass rather than GPS. We fitted it to my boat's transom one foggy day, and set out from Fowey towards Falmouth for sea trials. This was before I had GPS or any electronic nav aids btw.

Below 4 knots all was well and it worked brilliantly, but above 4 knots it broke into oscillation, and at 5 kts I managed to get it to make a complete figure of 8!

We played with the PID gains, but finding the best parameters by fiddling around wasn't possible - I concluded that we needed a water speed input. By now we were out of sight of land, disoriented, and with rather a large circle of uncertainty! I enjoyed the subsequent intellectual exercise of navigating safely home (we 'aimed off' to find the coast near Chapel Point than aiming direct for Dodman).
 
Slycat said:
I think the problem with tiller pilots is that they draw a large amount of power over long distances? The pi draws very little and the water would be doing the heavy lifting.
Click to expand...

So use a tillerpilot to drive the servo part of a windvane. Loads of people do that already.

Pete
 
Interesting, hadn't thought of tiller pilot to windvane servo.. very nice plan.

But a 'cheap' tiller pilot is 350+ quid? And a pi + cheap servo is like 50 quid?
 
Slycat said:
Interesting, hadn't thought of tiller pilot to windvane servo.. very nice plan.

But a 'cheap' tiller pilot is 350+ quid? And a pi + cheap servo is like 50 quid?
Click to expand...
Yep....I guess the other £300 is the cost of working out how to develop an acceptable closed loop response system!...

It would be an interesting project regardless... even if its just a bit of fun.
 
You must log in or register to reply here.

Share this page

Top