PECAN PY

[Gareth JonesParticipant @garethjones79649]

The inspiration for this project came from John Redearth's thread 'Scuffy and the compass' which appeared on this site about a month ago. When I first read his post I found it interesting but unlikely to have any direct application to my models. However Bob Abell pointed out that Kim Holland and myself (and my wife Elizabeth) are occasionally to be found trying to get vintage pond yachts to sail from one end of the Llandudno boating lake to the other and this might be a modern alternative to mechanical self steering gears. This sparked my imagination and since most of the other reaction to John Redearth's post was negative I took it as a challenge to have a go and make the project work.

Firstly a bit of background to vintage pond yacht racing. This originated long before the advent of radio control and the most common practice is to race the yachts in pairs from one end of the pond to the other. If the wind is across or behind the yacht's course the aim is to try and sail as fast as possible directly to the opposite end of the pond (reaching or running) If the wind is against the direction of sailing, the yachts must tack across the pond from side to side in a series of zig zags (beating). The winner scores 2 points for winning a reach or run and 3 points for winning a beat. Normally every yacht in the competition will be raced against every other yacht in a series of heats.

My wife and I are both members of the Vintage Model Yacht Group. Elizabeth has a large collection of models, ranging from a 20 inch free sailing 'Star' yacht to a 78 inch 10 Rater that can either be sailed under radio control or with vane steering, A couple of these models can be sailed with 'Braine gear' self steering, invented by George Braine in the 1920's. This method connects the main sail to the rudder as the means of self steering. We also have a number of yachts including a 36R, Marblehead and 10 Rater that can be vane steered. In this method a light balsawood feather on the stern is connected to the rudder by a variable gearing mechanism. The vane gear can be self tacking in that when the yacht is being sailed on a beat and reaches one side of the pond, the yacht can be turned by means of a pole and the vane will automatically take up the correct setting for the opposite tack, without the need to stop the yacht and adjust the setting of the vane or sails. Generally a well set up yacht will hold its course, relative to the wind direction quite well on a beat, without any assistance from the steering gear. It is on the reach and run where the assistance is necessary as the centre of effort of the sails starts to try and turn the yacht as well as drive it forwards.

The thought for the Pecan Py project was to try and build an electronic self steering mechanism that I could race against a modern vane steered design. within the normal vane racing conventions. The Model Yachting Association free sailing rules state 'The use of radio control or any device not activated by the force of the wind, including timing devices for the operation of a tacking guy is prohibited.' Pecan Py will therefore not be a legal competitor in any formal racing competition, it's just a project to experiment with, learn from and perhaps develop some knowledge that can be applied back to traditional vane steered yachts.

The next instalment of this thread will be a guide to the steps we plan to take in developing a working system, with the help of Kim Holland, the well known Manxman, who knows far more about model yachts and Arduino processors than I do.

[Gareth JonesParticipant @garethjones79649]

Prototype Electronically Controlled Autonomously Navigating Pond Yacht

[Dodgy Geezer 1Participant @dodgygeezer1]

I can imagine a yacht race over a triangular course, with gates which must be passed through. The yachts would be autonomously controlled, using onboard computers which took data from a variety of sensors (speed, wind, heel, etc) and made course decisions, including tacking strategies, accordingly. The gates would be infra-red beams which are detected by the boat, or possible GPS….

[Gareth JonesParticipant @garethjones79649]

Dodgy,

You are way ahead of me there but that's the general direction I am heading. I will be satisfied if I can get the yacht to sail from one end of the pond to the other. A very simple system schematic is shown in the diagram below.

The system will need some means to input the desired sailing direction. I don't know what this is going to be at the moment but I am thinking along the lines of having a push button switch, you point the yacht in the direction you want it to go and press the button, it reads that heading and tries to hold it. Its clearly impractical to have to connect up an external device every time you reach the pond edge and want to change direction when beating.

The system will need to know wind direction, relative to the model, compass heading and global position. It will also need feedback of the rudder and sail position. The processor will have to drive the rudder and eventually the sail servos. All these requirements are essentially as on the RoboSail website which was linked on the Scuffy thread.

I have learned from 40 odd years of systems engineering experience that it is vital you understand exactly what is going on in the system while it is working. One thing I have noticed from expert vane sailors is that they can walk alongside the model and know exactly what the vane and sails are trying to do. All I can manage is to see if its heading in the right general direction. Consequently some form of recording system will be required so that each mission can be replayed afterwards to see how the system performed.

The overall plan is to progress in a series of small steps, applying the knowledge gained from the predecessors.

1. Fit the system but operate it passively as a recording system for a conventional vane steered pond yacht. You can then get some idea of how the mechanical vane system performs and look at the effect of different gear ratios between vane and rudder.

2. Remove the mechanical vane steering mechanism and connect the processor to the rudder servo. Programme the processor to hold a particular course relative to the wind direction. In this mode the system is mimicking the mechanical vane steering system.

3. Programme the processor to hold a compass heading by controlling the rudder servo. As several people pointed out on the Scuffy thread, this will not counteract the tendency of the model to drift sideways in a crosswind

4. Programme the processor to hold a course to a fixed point in the distance, or series of way-points in a straight line derived from GPS data

5. Programme the processor to adjust the sail angle for variations in wind direction

At the end of these steps we should have a reasonable idea of how to optimise the straight line performance of the model and then move on to the much more challenging objective of getting it to change direction at the optimum point on the pond. This could lead on to pre-programmed self tacking and sailing around a triangular course.

The next step is to come up with a more detailed system schematic and decide which model is going to become the test bed.

Gareth

[Steffan BowenParticipant @steffanbowen96329]

Gareth

Have you thought of adding a gyroscope in order to to give you the desired sailing direction. This could be started with your push button when a boat is pointing in that direction.

Staffan

[Dodgy Geezer 1Participant @dodgygeezer1]

The basic services needed for this application have already been worked out.

About 10 years ago a Canadian team of modellers was experimenting with a project of sending a model glider on a hydrogen weather balloon up to the edge of space – around 80-100,000 ft up. The balloon would then burst, and the glider would fly itself back to its starting point. This required an on-board autonomous autopilot (working in three dimensions and capable of remembering things like windspeeds in the lower layers of the atmosphere) Their web site is here (rather old now, so links are going to be poor):

**LINK**

Since then several team have done this sort of thing, and there are now complete autopilot kits which can be bought to fly a model plane autonomously – here's the first one I googled:

**LINK**

Aircraft sets are obviously more complex than a boat set, with the need for altitude measurement, levelling and stall/spin recognition – but they are essentially the same concept. The sets also include data transmission between the fixed and moving sites, and provision for an on-board camers…

If you want to make your own with a Raspberry or similar, I suggest that you study the Canadian site – perhaps even get in touch with them – they should be able to explain their system block diagrams and design philosophy…

[Gareth JonesParticipant @garethjones79649]

Thanks for the information Staffan and Dodgy. I am planning to use a compass module which incorporates a three axis gyro, accelerometer and magnetometer so it should be able to provide a potential steering reference. There is lots of useful information on the link that John Redearth gave in his original Scuffy and the compass post, which takes you to a robotic model yacht site.

Unfortunately at the moment Pecan Py is having to play second fiddle to another project I am involved in which is to build a 1:3 scale flying model of a Blackburn SP seaplane. This was the first aircraft to fly from Brough in November 1916 and the model is being built to celebrate the centenary of what is now the BAE Systems site in East Yorkshire. Brough is believed to be the oldest site in the world to have been continuously involved in aircraft design and manufacturing. The model will be powered by a pair of twin cylinder petrol engines and have a wingspan of just over 7 metres. However, unlike the original it will be making its maiden flight from a runway, not the River Humber. My part in the project is the radio control system for the flight and engine controls. The system is defined and we are just awaiting delivery of a few more items before we can assemble it and carry out some testing.

Gareth

[Kimosubby ShipyardsParticipant @kimosubbyshipyards]

Hi,

just want to confirm Gareth's ideas and mine. We know that this has been done, after all there's cars, planes, rockets etc all doing it right now. There is even a specific website for robot yachts! What we are going to try is start from scratch, because that way we will learn from our mistakes and start to understand the skills needed. Our first endeavour, and as far as we are concerned, only endeavour, is to get a yacht to sail from A to B using what we have cobbled together to guide it, rather than use what was invented in the early 1900's using a piece of elastic and later a wind feather (vane).

Sounds so simple, so we are going to become that elastic band and think like it to steer/guide our yacht along. Only by doing it ourselves will we get anything from the task. Once we've worked out how the band can be replaced by electronics, we'll move up to the vane. Once that's done we can then think in 2 dimensions and gybe the yacht, but till then, steering a straight course based on the wind direction and our sail setting is the aim. We will still be poling the turn till then. After all, the most important feature of this part of the yachting is getting the yacht to sail as best it can by the sails as set. That cannot be replaced by a computer! And for those that still don't know, the rudder only operates on the RUN, that's down wind, so the rudder is keeping the yacht on the wind, not letting it fall off either by luffing up or gybing. On the beat (upwind) the rudder is locked dead ahead!

Another project would be to use the onboard electronics to sail into the wind (the beat), that is enabling the yacht's electronics to decide when a tack is required to make the target range decrease. A much harder task, but doable all the same.

We're using Arduino cos they're there and its what we started with. It also mixes with all the other boards and communicates with any and almost all. There are loads of sensor attachments that can be included, we will just plod along till we get it to happen.

Yes, there are several websites where we can download sketches already written, and purchase equipment from lists already collated and combine them all according to someone else drawings and plans. We could also use modern yachts with sail winches. Thats not for us. We are starting as newbies and will plot out own course (obviously we'll look and see what others have down) but we will do it in our own time, our own pace and our own assimilation of the relevant coding requirements. Me, I can blink LEDS to required patterns, switch stuff on and off, and other very simple bits, so this is new ground. Luckily I did use a BBC and a Commodore 64, so that bit of simple basic has helped so far with the sought of logic required.

If Gareth can fly a float plane then I'm sure he can come to terms with programmable chips. He's the systems engineer, and knows how to plan the test stages. Me, I'm along to sail and add the possibles and what ifs. Also, I live right next door to a very large expanse of usable water (16 acres) so have the test tank and safety boats on hand.

Aye, Kim

Aye, Kim

[CookieOldParticipant @cookieold]

Clever Boys,

Dave

[Author]

Posts