Sheet-to-tiller

[Nee-Wom]

Hi,

This afterthought of an "electronic" sheet-to-tiller system stayed with me. Wouldn't it be a worthwhile idea to effectively explore?

It wont overall compete with a GPS based autopilot, but it would react faster, giving quite an edge in gusty or otherwise unstable conditions. Besides, it would cost a lot less than a windvane and way simpler to install and maintain. 

Hardware seems readily available. Tension load cells are simple enough to read and not that expensive. This one is less than $40 : 2Kkg should do it for starters. Then an amplifier/converter ($15), or this other one with an integrated USB port ($32).

After proper conversion and calibration, quite reliable data could be achieved and sent to Pypilot, where it could be made into a "sheet-to-tiller" mode, or even something always-on to provide a contributing vector for calculating the output to the actuator. And, since the data is there to be used, along with wind, SOW/SOG and heeling data, it could even be made to throw some advice (or warning) on sail trimming.

[seandepagnier]

I also considered sheet to tiller electric autopilot hybrid ideas.

For regular sheet to tiller using elastic I have a few concerns:
1) More chafe on the rigging especially downwind depending on sea state.
2) efficiency, both steering straight and limiting sail combinations that are otherwise faster but not very balanced
3) need to adjust as wind speed changes and change the arrangement completely depending on wind angle.

Now obviously some boats are much more balanced than others, and with crabclaw and other rigs, it has been shown there are arrangements that mitigate adjustments for wind speed changes. On these boats, using a sheet to tiller arrangement the above issues are minimized but never really fully eliminated. Even wind vane suffer from the above compared to a sufficiently strong autopilot, but tiller pilots are generally fairly weak compared to a wind vane in terms of power.

The idea of using an electronic tension meter on the sheet to aid the autopilot is probably a good one, but the measurement from this is overlapping with other measurements already available like wind direction or compass heading. It could potentially provide valuable input but utilizing this in the best way probably needs an automatic tuning or learning algorithm.

One idea for a sheet to tiller hybrid is to have a very small motor with a large gear reduction that adjusts the bungee tension. The idea here is the autopilot would react to changing conditions rather than course changes. It might also react to very large course errors to help the system steer straighter. This would help the situation but is really not enough since you can also adjust where the sheet attaches (near or far from rudder shaft) as well as the strength of the bungee, not just the tension. To have the best setup might require several small motor to adjust these.

The same system could be used on a windvane/autopilot hybrid. Rather than driving the oar with a motor, the motor would rotate the air vane, and another motor could adjust the helm (offset between pendulum oar and main rudder) These motors also could be tiny, something like geared down rc servos.

The filter might need to be a lot lower and the power consumption would be minimal.

[Nee-Wom]

I also considered sheet to tiller electric autopilot hybrid ideas.

For regular sheet to tiller using elastic I have a few concerns:
1) More chafe on the rigging especially downwind depending on sea state.
2) efficiency, both steering straight and limiting sail combinations that are otherwise faster but not very balanced
3) need to adjust as wind speed changes and change the arrangement completely depending on wind angle.

Right, never been a fan of the bungie approach either. All those lines running constitute an additional hazard. But it does make for quite a healthy plan B though. If you know in advance how to readily rig it on your boat. I have toyed with a few setups, and the one I settled on uses the genoa and a pivoted lever, something like this, but only one lever that serves both sides of the boat, having two series of wholes cut in. With the addition of a rail slide under the tiller for bungie attachment, adjustments are kept to a minimum (two anchor points on the lever and slide position for the bungie).

The idea of using an electronic tension meter on the sheet to aid the autopilot is probably a good one, but the measurement from this is overlapping with other measurements already available like wind direction or compass heading. It could potentially provide valuable input but utilizing this in the best way probably needs an automatic tuning or learning algorithm.

There is indeed a possibility for data redundancy, but wind (speed and direction), although being a determining factor, is imperfectly correlated to the actual forces acting on various points of the boat. Windage, sail area, sail shape, sheeting angle, reefing, even rope material for sheets will lead to important force variations for the same wind conditions. We can't reasonably consider measuring the force coming from windage. For a mainsail it isn't an obvious task to effectively measure whatever energy is transferred from the mast to the deck and hull. But the sheet tension is readily accessible, and I suspect it would be quite a good predictor of the overall force (from wind) acting on the boat. And it is instantaneous. A good helmsman will anticipate the force of the gust as soon as it starts and attempts to compensate before going off course. I don’t think that wind data alone can allow for that level of anticipation, but a good model incorporating sheet data could. Probably of the second order, with empirically determined parameters specific to the boat. Maybe sheet data could be used in targeted short windows of time, along with other data, to initiate corrections that would either be added to or momentarily override the regular broader course keeping output to motor. Much like what you are saying about reacting to changing conditions rather than course changes.

The same system could be used on a windvane/autopilot hybrid. Rather than driving the oar with a motor, the motor would rotate the air vane, and another motor could adjust the helm (offset between pendulum oar and main rudder) These motors also could be tiny, something like geared down rc servos.

This is where I'm going. I made a deposit on this windvane and will be installing it in a few months. It has a built-in lever that operates the vane from inside the boat, in front of the sector, for the very purpose of using a tiller pilot (see photo). The wind vane uses the boat's rudder, so only one actuator is needed for steering.

The filter might need to be a lot lower and the power consumption would be minimal.

What filter would that be exactly?

[seandepagnier]

I mean the filter as the control loop for the autopilot would be much less adjustments if you are controlling the angle of the air vane or other offsets compared to a normal autopilot.

[Nee-Wom]

I'm going to investigate further what can be done with sheet data. I ordered the load cell and the board mentioned in my first post.

Also, I was just graciously granted access to a wind databank from the data science department of the University of Bergen. Wind data is taken from an offshore platform with a variety of sensor (LIDAR and ultrasonic) - has been since 2015 if I understand correctly. I'll look into ways of modeling it to serve the purpose of understanding how sheet data could be used to increase the efficiency of an autopilot.