Pypilot - Raspberry Pi autopilot. Any thoughts?

[NickRobinson]

Just been reading about the 9 axis, multifunction, helm from your phone/from home PC(!)/TV remote control on board.
Rudder reference/over current/end stop/clutch all available.
Autopilot Computers
All for $100 with $60 for the power control box to run a dead wheel or tiller pilot
A Raymarine (no drive) EV100 is multiples of this price.
Too good to be true? Supermarket toy drones steer and balance them selves competently for pocket money prices...
Anyone tried one?

 

[vas]

IIRC someone on this forum has made one.
If I'm mistaken, there's a project and drawings etc online either in arduino or somewhere else I forget.

 

[rogerthebodger]

The forum member who knows all about then and other Raspberry Pi and its various applications has left the building due to some of the recent negativity on the forum.

 

[TernVI]

I think getting the sensor data into the Pi (or maybe arduino?) is not hard.
Controlling a motor is not hard.

The hard bit is catching up with 40 years of development and real life experience when it comes to the algorithm to make the motor respond to the sensor inputs.
A drone is easy, they only have a bit of wind to worry about, while a yacht has to work in waves.

 

[Jonny A]

The electronic engineers at work all like Raspberry Pi and are happy to suggest them as low cost hardware for a variety of projects that need good reliability, for what it's worth.

 

[TernVI]

The electronic engineers at work all like Raspberry Pi and are happy to suggest them as low cost hardware for a variety of projects that need good reliability, for what it's worth.

They are great low cost hardware.
The operating system is sound.
I use one as an everyday browsing/office/media computer.
The only thing it doesn't do is keep my study warm like the PC used to!

However, I don't think the platform is the issue with making a 'good' autopilot. I think the issue is control theory, algorithms, real life knowledge of what you want the machine to be doing, in each and every combination of circumstances.

If you think of the 'brain' part of the autopilot as a 'black box' with inputs like compass, gyro, acceleration, speed, rudder angle etc, and outputs of motor drive, implementing the black box is the easy bit IMHO. The hard part is defining what precisely it does for every combination of inputs.
It's a control loop, I've done lots of those using only 'dumb' analogue circuits, and a few using all sorts of digits.
When you get down to it, some 'gyros' were pretty sophisticated using mechical analogue computing.

Whether you use a PC, or a Pi or an Arduino or other micro to do the sums will affect the speed and power consumption, but the big thing is doing the right sums as far as I can see?

I'm always suspicious of electronics design where the guts of the hardware is chosen before the problem is properly understood.

 

[Hoolie]

There's a lot of discussion and info on cruisersforum.com particularly on the opencpn forum. The developer, Sean Depargnier, has put a lot of effort into integrating the autopilot with the nav system - more so than available commercial devices - and has steered his own boat long distances in confined waters. All the hardware and software data is in the public domain but using it is not for the faint hearted!

 

[ottow]

This summer I converted an old Autohelm 2000, ripping out the innards of the control box and replacing it with a TinyPilot + motor controller.
Here is the thread on the Pypilot forum: Pi and motor controller in same box?
And some pictures here: Index of /maxi999/pypilot

I were only able to do a few sails with this setup last summer, but everything is working as expected and I am pretty happy with the sysytem.
I was very keen on getting the steer-to-wind functions working, but Sean did not release the updated firmware supporting Tinypilot as a NMEA Wifi client before I went onto the hard for the year.