Choix

121 views
Skip to first unread message

Francis Roussel

unread,
Aug 9, 2022, 9:53:54 AM8/9/22
to Microtransat
Hello,

Searching the web a bit I found this thread.
https://discuss.ardupilot.org/t/sailboat-support/32060/835

He indicates that the ardupilot project seems to be ready for sailboats
My knowledge of English not being very good I would like a confirmation:
the old configurationd (I have a cc3d) no longer work with the current
versions.
I also understood that the cubepilot configuration was ideal, but to put
it in a boat that has a high risk of never being recovered, it's a bit
too expensive.
Does anyone have any advice on a setup that would work with the sailboat
program but at a low cost.

https://ardupilot.org/copter/docs/common-autopilots.html

Regards

fr

Colin Sauze

unread,
Sep 2, 2022, 9:22:21 AM9/2/22
to microt...@googlegroups.com
Dear Francis,

Sorry for the slow reply to this email. I've been very busy recently and
just returned from holiday to hundreds of emails needing me to reply.

A few teams have used adapted versions of the Ardupilot code in the
past. I think most of these are based on the Ardurover code, but I
wasn't aware of it ever being merged into the main code for Ardupilot.

I remember working with the Ardupilot a few years ago for some UAV work
and it used to run on an Arudino (well technically an ATMega2560) style
board which could attach with a compass, GPS, gyros and accelerometers
and they were often sold as a complete kid together. These cost under
$/£/€ 100 and were called the "Arudpilot Mega". There don't seem to be
many places to buy them now.

I did find one website which still claims to sell them:

Ardupilot mega main controller:
https://www.flyingtech.co.uk/electronics/3d-robotics-ardupilot-mega-apm-26-flight-controller

Ardupilot GPS and compass module:
https://www.flyingtech.co.uk/electronics/ublox-neo-6m-gps-883l-compass-module-apm-fc

I'm sure these can still be found on ebay too and there used to be lots
of companies making cheap clones of them.

These will need to run the older version of the Ardupilot code. From
what I remember that code was a bit complicated, but very well designed
and included support for logging, telemetry, remote control and many
additional sensors. Although they are much simpler than the newer
ardupilots like the Pixhawk and CubePilot, this would still be more than
capable of running a Microtransat boat and much cheaper.

Colin.

On 06/08/2022 16:25, Francis Roussel wrote:
> [RHYBUDD! E-BOST ALLANOL / CAUTION! EXTERNAL E-MAIL]
>
>
> Hello,
>
> Searching the web a bit I found this thread.
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdiscuss.ardupilot.org%2Ft%2Fsailboat-support%2F32060%2F835&data=05%7C01%7C%7C836662803e6746c5bb2e08da7a0e9a06%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637956500374827688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=yhS245Yjx8cyCent%2FqhcGYF1JpoaFFPajpFcvBGbpRg%3D&reserved=0
>
>
> He indicates that the ardupilot project seems to be ready for sailboats
> My knowledge of English not being very good I would like a confirmation:
> the old configurationd (I have a cc3d) no longer work with the current
> versions.
> I also understood that the cubepilot configuration was ideal, but to put
> it in a boat that has a high risk of never being recovered, it's a bit
> too expensive.
> Does anyone have any advice on a setup that would work with the sailboat
> program but at a low cost.
>
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fardupilot.org%2Fcopter%2Fdocs%2Fcommon-autopilots.html&data=05%7C01%7C%7C836662803e6746c5bb2e08da7a0e9a06%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637956500374827688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=U9ISho5n8PZkiH1L%2FbH74771uXyHr0BfpdYWO7wun5E%3D&reserved=0
>
>
> Regards
>
> fr
>
> --
> You received this message because you are subscribed to the Google
> Groups "Microtransat" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to microtransat...@googlegroups.com.
> To view this discussion on the web, visit
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fmicrotransat%2F0dac7e73-1f5e-025c-f266-ae25ed8959ce%2540gmail.com&data=05%7C01%7C%7C836662803e6746c5bb2e08da7a0e9a06%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637956500374827688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=oV1aNt4h75olik1cQYsmR0t9s7LZqkSHH%2Bp5Gqd%2BHVI%3D&reserved=0.


--
Dr. Colin Sauze
Research Software Engineer
Supercomputing Wales Project
Room 2.16, Physical Sciences Building
Aberystwyth University,
Penglais, Aberystwyth
Ceredigion, UK, SY23 3DB

Tel: +44 (0)1970 622774
Webpage: http://users.aber.ac.uk/cos



----------------------------------------------------------------------------------------------------------------------
Y Brifysgol orau yn y DU am Ansawdd ei Dysgu a Phrofiad Myfyrwyr
Best University in the UK for Teaching Quality and Student Experience
(The Times and Sunday Times, Good University Guide 2021)
Rydym yn croesawu gohebiaeth yn Gymraeg a Saesneg. Cewch ateb Cymraeg i bob gohebiaeth Gymraeg ac ateb Saesneg i bob gohebiaeth Saesneg. Ni fydd gohebu yn Gymraeg yn arwain at oedi.
We welcome correspondence in Welsh and English. Correspondence received in Welsh will be answered in Welsh and correspondence in English will be answered in English. Corresponding in Welsh will not involve any delay.

Hernan G Curras

unread,
Sep 2, 2022, 9:48:34 AM9/2/22
to Colin Sauze, microt...@googlegroups.com
@Colin Sauze  @Francis 

Do you think that the next model of sailboats are going to be using the Ardupilot platform, or are we moving to a more-complex ROS like environment (probably using the ardupilot as an IMU/ROS node )? 

I'm seeing both approaches, obviously there is an energy tradeoff of each platform, and also different boat functions related with guidance, navigation and control.  

H.-

..........................

Ing. Hernan Gabriel Curras
Ciudad de Buenos Aires, Argentina


Colin Sauze [cos] (Staff)

unread,
Sep 2, 2022, 11:13:21 AM9/2/22
to li...@microtransat.org
I think there's probably room for both approaches along with custom developed solutions. ROS is great if you want more extensibility, detailed logging, do complex behaviours like swarms/weather routing/automatic collision avoidance or interact with industrial grade sensors like sonars. But it tends to require computers that are more power hungry and expensive. An ardupilot or an even simpler microcontroller based solution should be more than capable of controlling a basic boat and getting it to reach a set of waypoints while costing less, having fewer failure modes and using less power.

Colin.

________________________________________
From: microt...@googlegroups.com [microt...@googlegroups.com] on behalf of Hernan G Curras [hernan...@gmail.com]
Sent: 02 September 2022 14:48
To: Colin Sauze [cos] (Staff)
Cc: microt...@googlegroups.com
Subject: Re: [microtransat] Choix

[RHYBUDD! E-BOST ALLANOL / CAUTION! EXTERNAL E-MAIL]

@Colin Sauze<mailto:c...@aber.ac.uk> @Francis

Do you think that the next model of sailboats are going to be using the Ardupilot platform, or are we moving to a more-complex ROS like environment (probably using the ardupilot as an IMU/ROS node )?

I'm seeing both approaches, obviously there is an energy tradeoff of each platform, and also different boat functions related with guidance, navigation and control.

H.-

..........................
Ing. Hernan Gabriel Curras
Ciudad de Buenos Aires, Argentina


El vie, 2 sept 2022 a las 10:22, 'Colin Sauze' via Microtransat (<microt...@googlegroups.com<mailto:microt...@googlegroups.com>>) escribió:
Dear Francis,

Sorry for the slow reply to this email. I've been very busy recently and
just returned from holiday to hundreds of emails needing me to reply.

A few teams have used adapted versions of the Ardupilot code in the
past. I think most of these are based on the Ardurover code, but I
wasn't aware of it ever being merged into the main code for Ardupilot.

I remember working with the Ardupilot a few years ago for some UAV work
and it used to run on an Arudino (well technically an ATMega2560) style
board which could attach with a compass, GPS, gyros and accelerometers
and they were often sold as a complete kid together. These cost under
$/£/€ 100 and were called the "Arudpilot Mega". There don't seem to be
many places to buy them now.

I did find one website which still claims to sell them:

Ardupilot mega main controller:
https://www.flyingtech.co.uk/electronics/3d-robotics-ardupilot-mega-apm-26-flight-controller<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.flyingtech.co.uk%2Felectronics%2F3d-robotics-ardupilot-mega-apm-26-flight-controller&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173486789%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=3ZiHV%2B6CmPrTJIjEvUzf2k%2BxZveH4GdxbYdnZTF7csY%3D&reserved=0>

Ardupilot GPS and compass module:
https://www.flyingtech.co.uk/electronics/ublox-neo-6m-gps-883l-compass-module-apm-fc<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.flyingtech.co.uk%2Felectronics%2Fublox-neo-6m-gps-883l-compass-module-apm-fc&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173486789%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=N5wSi8153VsHZYaWrAJNEtHPKhsowbLRGjIJGoJJksc%3D&reserved=0>

I'm sure these can still be found on ebay too and there used to be lots
of companies making cheap clones of them.

These will need to run the older version of the Ardupilot code. From
what I remember that code was a bit complicated, but very well designed
and included support for logging, telemetry, remote control and many
additional sensors. Although they are much simpler than the newer
ardupilots like the Pixhawk and CubePilot, this would still be more than
capable of running a Microtransat boat and much cheaper.

Colin.

On 06/08/2022 16:25, Francis Roussel wrote:
> [RHYBUDD! E-BOST ALLANOL / CAUTION! EXTERNAL E-MAIL]
>
>
> Hello,
>
> Searching the web a bit I found this thread.
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdiscuss.ardupilot.org%2Ft%2Fsailboat-support%2F32060%2F835&amp;data=05%7C01%7C%7C836662803e6746c5bb2e08da7a0e9a06%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637956500374827688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=yhS245Yjx8cyCent%2FqhcGYF1JpoaFFPajpFcvBGbpRg%3D&amp;reserved=0<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdiscuss.ardupilot.org%2Ft%2Fsailboat-support%2F32060%2F835&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173486789%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=BRaJclOII3j4X%2FS5G7YeOcTLAYBYyMOUr8yZ5WtlDTk%3D&reserved=0>
>
>
> He indicates that the ardupilot project seems to be ready for sailboats
> My knowledge of English not being very good I would like a confirmation:
> the old configurationd (I have a cc3d) no longer work with the current
> versions.
> I also understood that the cubepilot configuration was ideal, but to put
> it in a boat that has a high risk of never being recovered, it's a bit
> too expensive.
> Does anyone have any advice on a setup that would work with the sailboat
> program but at a low cost.
>
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fardupilot.org%2Fcopter%2Fdocs%2Fcommon-autopilots.html&amp;data=05%7C01%7C%7C836662803e6746c5bb2e08da7a0e9a06%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637956500374827688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=U9ISho5n8PZkiH1L%2FbH74771uXyHr0BfpdYWO7wun5E%3D&amp;reserved=0<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fardupilot.org%2Fcopter%2Fdocs%2Fcommon-autopilots.html&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173486789%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=iqG6%2B4KBMieUdwiSxJzfJel30u19E%2FMGv1foZpKVsrc%3D&reserved=0>
>
>
> Regards
>
> fr
>
> --
> You received this message because you are subscribed to the Google
> Groups "Microtransat" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to microtransat...@googlegroups.com<mailto:microtransat%2Bunsu...@googlegroups.com>.
> To view this discussion on the web, visit
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fmicrotransat%2F0dac7e73-1f5e-025c-f266-ae25ed8959ce%2540gmail.com&amp;data=05%7C01%7C%7C836662803e6746c5bb2e08da7a0e9a06%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637956500374827688%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=oV1aNt4h75olik1cQYsmR0t9s7LZqkSHH%2Bp5Gqd%2BHVI%3D&amp;reserved=0<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fmicrotransat%2F0dac7e73-1f5e-025c-f266-ae25ed8959ce%2540gmail.com&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173486789%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=L8SVTZRVDCduWiRrrWjxSfxxiMs5wTW6%2F%2FOHaimDmX8%3D&reserved=0>.


--
Dr. Colin Sauze
Research Software Engineer
Supercomputing Wales Project
Room 2.16, Physical Sciences Building
Aberystwyth University,
Penglais, Aberystwyth
Ceredigion, UK, SY23 3DB

Tel: +44 (0)1970 622774
Webpage: http://users.aber.ac.uk/cos



----------------------------------------------------------------------------------------------------------------------
Y Brifysgol orau yn y DU am Ansawdd ei Dysgu a Phrofiad Myfyrwyr
Best University in the UK for Teaching Quality and Student Experience
(The Times and Sunday Times, Good University Guide 2021)
Rydym yn croesawu gohebiaeth yn Gymraeg a Saesneg. Cewch ateb Cymraeg i bob gohebiaeth Gymraeg ac ateb Saesneg i bob gohebiaeth Saesneg. Ni fydd gohebu yn Gymraeg yn arwain at oedi.
We welcome correspondence in Welsh and English. Correspondence received in Welsh will be answered in Welsh and correspondence in English will be answered in English. Corresponding in Welsh will not involve any delay.

--
You received this message because you are subscribed to the Google Groups "Microtransat" group.
To unsubscribe from this group and stop receiving emails from it, send an email to microtransat...@googlegroups.com<mailto:microtransat%2Bunsu...@googlegroups.com>.
To view this discussion on the web, visit https://groups.google.com/d/msgid/microtransat/aa560416-480c-6a10-0265-8d5524e70480%40aber.ac.uk<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fmicrotransat%2Faa560416-480c-6a10-0265-8d5524e70480%2540aber.ac.uk&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173643050%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=zf3bRtIuq02YbQfW3WonvQ5vxfHe88EuQETgwkxYUlo%3D&reserved=0>.

--
You received this message because you are subscribed to the Google Groups "Microtransat" group.
To unsubscribe from this group and stop receiving emails from it, send an email to microtransat...@googlegroups.com<mailto:microtransat...@googlegroups.com>.
To view this discussion on the web, visit https://groups.google.com/d/msgid/microtransat/CAMaCG1rJJznT6-oUfDb_mLtnLe8YwF0YwJQXJRKjdn2FAov89w%40mail.gmail.com<https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fmicrotransat%2FCAMaCG1rJJznT6-oUfDb_mLtnLe8YwF0YwJQXJRKjdn2FAov89w%2540mail.gmail.com%3Futm_medium%3Demail%26utm_source%3Dfooter&data=05%7C01%7C%7C8724628414494e61458f08da8ce9d509%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C1%7C637977233173643050%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=q7g5kWitDUqHk01WQagi%2FRECZke75o2HhJ8yxMMpmZk%3D&reserved=0>.

Dermot Tynan

unread,
Sep 16, 2022, 9:43:06 AM9/16/22
to microt...@googlegroups.com
On 02/09/2022 14:48, Hernan G Curras wrote:

>
> Do you think that the next model of sailboats are going to be using
> the Ardupilot platform, or are we moving to a more-complex ROS like
> environment (probably using the ardupilot as an IMU/ROS node )?
>
> I'm seeing both approaches, obviously there is an energy tradeoff of
> each platform, and also different boat functions related
> with guidance, navigation and control.


Sorry, only seeing this, now. My own view, such as it is, is to use a
hybrid model. AVR-based boards like the Arduino are great for real-time
programming and tight PID loops, but not great when it comes to floating
point, or any kind of multi-tasking. Likewise, using something Linux (or
Unix) based gives you the ability to have all kinds of high-level
functions, and floating point is supported by an onboard co-processor,
but real-time performance on a pre-emptive kernel is a challenge. Also,
they have different power consumption models.

By using something like an ATmega to control the rudder and sail trim
purely based on an electronic compass and a wind direction indicator,
means your PID loops are well-controlled and power consumption is
minimal. The low-level board can evaluate power conditions and power up
the main computer as appropriate. Similarly some sort of single-board
computer allows the luxury of reading from a GPS, doing the
great-circle-distance computations, doing the waypoint/track analysis,
and finally, updating the "course to sail" on the low-level board,
before shutting down, seems the way to go. The low-level board can power
up the main computer after a certain period (in time for the next
satellite update), or in the event of an alarm, such as a wind heading
which has shifted by more than ten or twenty degrees.

The upper-level board can also incorporate secondary systems to look at
bigger trends, such as the 8AM battery voltage trend, or satellite
failure rates, or even discrepancies between the GPS course as computed,
and what the compass thinks. The final persuader for me is that I can
simulate the upper-level system using a virtual machine running the
exact same software, and feed it simulated low-level data. It's easier
to come up with a locked-down algorithm for the low-level code, and to
emulate the communications, while debugging the more complex upper-level
components. I've done simulations where the "boat" thought it was
sailing (in real time) around the coastline. Additional work needs to be
done to simulate failures such as a malfunctioning rudder, so that the
upper-level code can detect the error and send an alarm via satellite
communications. The more low-level error conditions I can create, and
then develop upper-level alarms and tests for, the better.

Finally, it's worth pointing out that Volvo Ocean Race boats use a
similar model. The skipper and the navigator aren't part of the watch
system. The two watches just sail the boat as fast as they can - they're
not interested in the GPS, just the hull speed, the AWA and the compass
heading. The navigator then analyses weather patterns, course to
waypoint and boat performance, and adjusts the course sailed accordingly.

            - Der

--
Dermot Tynan
dty...@kalopa.com

Damon McMillan

unread,
Sep 16, 2022, 10:14:39 AM9/16/22
to Dermot Tynan, microt...@googlegroups.com
I wanted to chime in on the topic of autopilots for unmanned boats...

As a mechanical engineer, my problem has always been that I am just not capable enough with software to set up a system that Dermot describes.  A single Arduino is pretty much as much as I can handle.

So perhaps the logical choice for someone like me would be to just buy an off-the-shelf autopilot.  But it seems that off-the-shelf autopilots are lacking some needed functionality, for instance battery voltage monitoring and automatic throttle management for a solar boat, or interfacing with a satellite modem.  To be fair, I have never actually tried to use one; perhaps if I did I would be able to figure out workarounds.  But if an off-the-shelf autopilot requires me to alter the code to get the functionality I need, then what's the point?  And I also wonder about power draw...

So my approach for the two USVs I have built is to use an Arduino Mega running my own autopilot code.  I would never attempt to write my own autopilot code if I were trying to fly a drone, but for a boat that has essentially just one degree of freedom (turn left or turn right), it's really not that difficult.  An Arduino Mega has all the I/O needed to run the GPS, satellite modem, compass, various power monitoring devices, motor, rudder, etc.  And it has no problem doing the great-circle computations and other navigational tasks, plus the various health-monitoring and telemetry tasks.

Obviously image processing or interfacing with complex payloads would require something more than an Arduino.  But if the goal is to just control the boat itself, there is no question that an Arduino with homemade code is more than adequate.  I'm  not saying it's the BEST method, but it certainly is a viable option.

Damon

Damon McMillan
Blue Trail Engineering LLC
1971 Blue Mountain Road
Longmont, CO 80504
720-526-6388 (business)
www.bluetrailengineering.com



--
You received this message because you are subscribed to the Google Groups "Microtransat" group.
To unsubscribe from this group and stop receiving emails from it, send an email to microtransat...@googlegroups.com.

Francis Roussel

unread,
Sep 16, 2022, 2:40:57 PM9/16/22
to microt...@googlegroups.com
Hello,

The time is up to you, I'm leaning towards:
A raspberry zero for calculations and its small size and an esp32 for
wifi to communicate at short distance, it has enough pins, it is small,
has a lot of memory and a well-developed community, can use arduino.
For the servo-controls, I think it is necessary to opt for brushless
motors and magnetic encoders, whether for the wind vane, the positioning
feedback for a sail or rudder winch, because they allow the electronics
to be sent back to watertight areas or even sealed in epoxy. In some
ROVs the engines are directly in water or in oil.

In the European Union, Directive 240 requires ships over 2.5m to have CE
certification (about 1400 euros to have it done by an approved service
provider), you can do without it by declaring a "competition ship".
Procedure in France rather particular.

The challenge should adapt to this maximum length of 2.5m, sailboats
exist in 2.30m in certain IUT university institutes.

It is true that the microtransat is aimed more particularly at
engineering schools, but that is a defect: the project must be the
brainchild of a teacher and the students change every year.

Moving towards an open source project specializing in this challenge at
the electronic control level would make it possible to extend the number
of projects, not everyone practices programming, setting up a hull,
managing the seasons with the winds and currents in the Atlantic. The
pooling of knowledge and know-how would allow more attempts and
therefore more interest

We have nothing more to demonstrate, private companies like Saildrone
have paved the way and demonstrated that it was possible, remains the
pleasure for individuals or learning for schools.

My English is due to google, robots can be forgiven

fr

Julian Blanco

unread,
Sep 16, 2022, 3:00:54 PM9/16/22
to Microtransat
Semi related. Ive been porting a lot of my code from avr microcontrollers to the samd51 recently. The m4 has excellent power performance and is plenty capable of handling floating point math and linear algebra for state-space control techniques and basic kalman filters. Plus adafruit and sparkfun (https://www.sparkfun.com/products/14713)  have nice breakout boards and they are Arduino capable. 

Also hello! I am new to the group and happy to be here. 

Paul Miller

unread,
Sep 16, 2022, 3:01:11 PM9/16/22
to Francis Roussel, microt...@googlegroups.com
Hello Francis,

No reason to apologize! Your Google-English is no doubt better than my Google-French and some of the folks from the Commonwealth may say that my American-English is a crime against humanity!

The length limit imposed by the rules was established to bring as many countries as possible into the competition. While the EU adopted 2.5 meters (metres?) about a decade ago, the US rules have significantly increased restrictions starting at 8-feet (2.42 meters) that were established in 1954. That might be important if a boat in the competition from Europe enters US waters, including the US Virgin Islands, due to Port State Authority.

If anything though, I've been suggesting for many years that the challenge is not going larger, but going smaller, because many crewed boats have crossed the Atlantic fully under auto-pilot for decades, including as short as 1.75 meters (and locally someone is attempting a manned crossing with a 1 meter boat). The real challenge is to make it small enough that it has a real obstacle being faster than adverse currents. Having a 1 meter autonomous boat take off on one side of the Atlantic and land in a targeted area on the other side is a big challenge!

All the best,
Paul


--
You received this message because you are subscribed to the Google Groups "Microtransat" group.
To unsubscribe from this group and stop receiving emails from it, send an email to microtransat...@googlegroups.com.

Pierce Nichols

unread,
Sep 19, 2022, 1:12:57 PM9/19/22
to Francis Roussel, Microtransat
Francis,

I've used a modified version of the Ardupilot sailboat code, and I had weird issues with it losing the plot and wandering off after a short period of time in autonomous mode. I ended up switching to PX4, which doesn't have its own sailing code but does have a much better architecture for writing software and debugging.

-p

--
You received this message because you are subscribed to the Google Groups "Microtransat" group.
To unsubscribe from this group and stop receiving emails from it, send an email to microtransat...@googlegroups.com.

John Lamport

unread,
Sep 19, 2022, 1:30:58 PM9/19/22
to Pierce Nichols, Francis Roussel, Microtransat

The Ardupilot and the Px4 are very versatile although you don’t actually program it as much as configure the operating parameters, with versatility comes complexity, there’s a lot to them. The Arduino controllers are good for peripheral interface of devices, GPS, compass, servos, WiFi module and so forth. Although you can do some limit control such as P&ID, that’s not what they are really made for especially if you’re trying to send an autonomous boat across the Atlantic.  For that you should use a microcomputer such as a Raspberry Pi. I have been working on that kind of a system for a while, actually on and off for a while as I have to many other projects. I have it now that the Arduino reads the GPS and compass and sends that info to the Pi. The pi will store waypoints and calculate the heading based on the info from the Arduino then send a course correction to the Arduino for the servos. There is programming involved with all of these options but I find the Arduino/Pi combo a good middle of the road between the three. I will be upgrading the Arduino to an ESP32 as it has a few other built in options like WiFi that someone had previously mentioned.

 

They don’t use a lot of battery power, you can put them into a sleep mode where the power consumption goes way down. But while it’s a sleep it’s not in control, you can wake it up every 15 minutes or so, check your course, trim the course, send a status message and go back to sleep.

 

John

 


John Lamport

Controls Engineer

Munters Corporation
79 Monroe Street, Amesbury, MA 01913 United States
Tel dir: 978-241-1196,

E-mail: john.l...@munters.com,



FACEBOOKTWITTER | YOUTUBE | LINKEDIN

Munterslogo.gif

This e-mail is solely intended for the addressee(s) and may be of a confidential nature. If received in error, kindly notify the sender and delete the e-mail. E-mail is susceptible to data corruption, interception, unauthorized amendment, tampering and viruses, and we only send and receive e-mails on the basis that we are not liable for any such corruption, interception, amendment, tampering or viruses or any consequences thereof. Any opinions expressed in this message are those of the author and do not necessarily reflect the opinions of Munters.

From: microt...@googlegroups.com <microt...@googlegroups.com> On Behalf Of Pierce Nichols
Sent: Monday, September 19, 2022 1:13 PM
To: Francis Roussel <franc....@gmail.com>
Cc: Microtransat <microt...@googlegroups.com>
Subject: Re: [microtransat] Choix

 

You don't often get email from pierce...@gmail.com. Learn why this is important

CAUTION: This email originated from outside of Munters. Do not click links or open attachments unless you recognize the sender and know the content is safe.

 

Francis Roussel

unread,
Sep 19, 2022, 3:08:50 PM9/19/22
to John Lamport, Microtransat

I like the esp32-s2 but had some issues with it.
I put bme280 sensors all over the place and they refuse to connect to the freebox when they do to a chinese wifi repeater with no problem.

Besides, I just saw to get myself a raspi zero 2w... Not found or at crazy prices. At the same time I saw the prices of a normal raspi 4 when it is available... crazy too.
Is it the consequence of the covid in China?
An explanation ?
fr

Pierce Nichols

unread,
Sep 19, 2022, 3:32:57 PM9/19/22
to John Lamport, Francis Roussel, Microtransat
John,

You can get a long way with Ardupilot or PX4 just by twiddling the parameters... but if you are building something outside the multicopter/general RC space (i.e. just about everyone here), you will more than likely need to write your own modules and/or drivers. For example, there's no sail module in the stock PX4. The one I wrote is too specialized and too hacky for general release at this time. The big advantage they bring is that all of the mathematically fiddly bits (EKF, PID, IMU calibration etc) are already implemented and well-tested. The biggest downside that I can see is that they don't have much power management, and both the IMU and the GPS tend to be somewhat thirsty beasts.

The ESP32 is going to draw more power than any of the 8 bit ATmegas (i.e. most older Arduinos) and more than most STM32s. Not sure about the power consumption of the Atmel ARM chips (most newer Arduinos) -- never had a good reason to investigate. This can be mitigated by doing some work on getting it to sleep as much of the time as possible; not sure what the state of the common RTOSes are wrt that.

The big advantage of a companion computer (i.e. RasPi or equivalent) is the ability to conveniently interface to cellular and satellite communications. Note that RasPi isn't really the best option here because the power management options aren't great -- last I checked (a while ago), there wasn't any sort of real sleep modes available. Beaglebones are better on this, but IIRC the 64 bit version isn't out yet, which means that you are in for some pain if you want to run ROS2.

-p

Francis Roussel

unread,
Sep 19, 2022, 5:15:40 PM9/19/22
to John Lamport, Microtransat


To answer more specifically:
For me, a sailboat must be able to follow a course without constant rudder corrections. A sailboat design job... That's my field.

It has sails set for a pace and must follow the wind. With a device like the time of Moitessier if necessary. So no action on the rudder.

From time to time we check course and position. And so more than putting it to sleep, we can turn everything off.

I don't know the beagleboard, but I ask to know :-)


fr
Le 19/09/2022 à 19:30, John Lamport a écrit :

Pierce Nichols

unread,
Sep 19, 2022, 6:47:01 PM9/19/22
to Francis Roussel, Microtransat
Right, it needs to do its primary course corrections through sail adjustments rather than rudder adjustment.

This is a bit of a tricky problem to solve in parallel with minimizing the power consumption. The control problem is adjusting the sails such that the resulting yaw rate given a neutral rudder position is zero. This is troublesome in practice because the disturbing forces (largely waves and local wind patterns resulting from larger waves) are large and can probably be modeled as pink noise over the frequency response of the system. If you can continually sample your orientation at a high rate relative to the system dynamics, this isn't a horrible problem -- a complementary filter can be used to split the steering output into sail and rudder outputs any way that makes sense for your goals.

However, the primary way to reduce the power consumption of your system is to reduce the frequency at which you sample your orientation. That's because reducing the sample rate reduces the power consumption of your sensors and allows you to put your microcontroller to sleep more of the the time. One common strategy is to connect your IMU's data ready pin to the wakeup pin on the microcontroller. That way, it wakes up each time there's a new sample, processes it, and then promptly goes back to sleep. It should be pretty obvious what the problem is here -- if you reduce the sample rate below twice the rate of the fastest dynamics in your system, you will get aliasing. And that may cause all manner of trouble that's not obvious until you see high winds and big waves.

All that said, your actuator power consumption is likely to completely dominate your sensor and microcontroller power consumption. The fastest way to reduce your controls power consumption is to make all your motion self-locking in some way. However, this has two downsides. First, your actuators will be more vulnerable to damage from large waves because they cannot give way in the face of excessive force unless you add a clutch or compliant mechanism to allow the driven surface to give way under high loading. Second, self-locking mechanisms are much less efficient -- typical self-locking mechanism such as lead screws and worm gears are only self-locking if they are less than 50% efficient. One way to get around both at the cost of extra cost and complexity is to add spring-loaded brakes to each axis. I.e. when power is removed they stop the motion and you apply power to the brakes to release them.

As for the Beaglebones, they're just another popular line of single board computers running Linux.

Francis Roussel

unread,
Sep 20, 2022, 2:41:52 AM9/20/22
to Pierce Nichols, Microtransat
Le 20/09/2022 à 00:46, Pierce Nichols a écrit :
> The fastest way to reduce your controls power consumption is to make
> all your motion self-locking in some way. However, this has two
> downsides. First, your actuators will be more vulnerable to damage
> from large waves because they cannot give way in the face of excessive
> force unless you add a clutch or compliant mechanism to allow the
> driven surface to give way under high loading. Second, self-locking
> mechanisms are much less efficient -- typical self-locking mechanism
> such as lead screws and worm gears are only self-locking if they are
> less than 50% efficient. One way to get around both at the cost of
> extra cost and complexity is to add spring-loaded brakes to each axis.
> I.e. when power is removed they stop the motion and you apply power to
> the brakes to release them.

I think we always come back to the same problem: the boat must be able
to sail with a minimum of controls consuming electricity. It must
therefore be studied much more than what I have seen in the boats of the
microtransat or the world championship.
If the endless screw is an excellent way to make an order irreversible,
my old memories of model maker in the morning of RC cars: to absorb the
shocks on the steering servo there was (perhaps still) a way. On the
axis of the linkage two two tubes ending in a V with a spring to
compress. Normally rotation link between the top tube connected to the
servo and the bottom tube connected to the steering arm. In the event of
a shock, the spring was compressed, allowing the lower axle to rotate
without the upper one moving, then returning to position. Easy to reproduce.
Another advantage of the worm gear reduction is that you quickly have a
huge reduction and therefore the possibility of putting low-power motors.
I loved Model B from https://uk.aliexpress.com/item/1005001636685142.html
the shoulder makes it possible to have a freewheel at the end of the
stroke on the motor side and easy to do on the other side: a very solid
and inexpensive cylinder with encoder for precision, operating directly
in 1S li-ion. Of course it is not fast but is it necessary?
A small sailboat does not have a lot of inertia, so it is able to take
big waves.
For lovers of 3D printers there are also beautiful things on the side of
cycloidal gears
https://www.youtube.com/watch?v=2fDbMWHQbC8


fr

Francis Roussel

unread,
Sep 20, 2022, 7:39:30 AM9/20/22
to Pierce Nichols, Microtransat
Return of a bottle to the sea launched on the maillist of a fablab: the
esp32 (like others) is compatible with ROS2, at least a part.

https://micro.ros.org/docs/overview/hardware/

Still reading even if the sunset is getting earlier and earlier we are
not yet in the long winter evenings

fr

Colin Sauze

unread,
Sep 23, 2022, 5:59:23 AM9/23/22
to microt...@googlegroups.com
The Microtransat was conceived by two university lecturers (one
engineer, one computer scientist) but it wasn't intended as a student
project which would just repeat the same assignment each year. It was
intended as a research project, since when it was conceived in 2005
nobody had done it or done much sailing with any kind of autonomous
surface boats. Obviously since then we've had one successful completion
of the challenge, but that was only in the (possibly) easier (and
definitely shorter) West to East direction, only in the unmanned
division where waypoint updates are permitted and with a commercial boat
near the maximum size. There is still a lot of challenges to make
smaller, simpler, cheaper, faster, more autonomous boats which can
manage the East to West direction.

With regards to a 2.5m limit, do you have a boat design which is between
2.4 and 2.5m long? Are you expecting some major gain in performance from
this increase in length?

I've always tried to encourage people to open source their code and
designs. There is a collection of links to these on the Microtransat
Wiki
(https://sourceforge.net/p/microtransat/wiki/Open%20Source%20Projects/)
which is linked to from the "Links and Resources" section of the
website. If anybody would like to contribute to this list please post
links to your projects or join the Microtransat Sourceforge project and
you can add them yourself (wiki admins willing to move us to a better
wiki platform are also welcome, this one dates back 10+ years now).

To throw in one more open source autopilot, I have written a minimum
viable Arduino based autopilot with PI control of the rudder, a
multiband if statement for sail setting and Rockblock tracking system.
You can find the code for it at https://github.com/colinsauze/morwyn.
This is build for a Honeywell HMC6343 compass and a SIRF3 GPS, with
small adaptations it should work with any common NMEA GPS or I2C
compass. This has had some testing, but it isn't particularly extensive.
I think I've got some PCB diagrams somewhere too if anybody is
interested in using this. I'm currently trying to persuade some students
to use this code and port it to the ESP32, so I hope to have something
better tested and for a more up to date platform later this year or next
year.

Colin.

On 16/09/2022 19:40, Francis Roussel wrote:
> [RHYBUDD! E-BOST ALLANOL / CAUTION! EXTERNAL E-MAIL]
>
>
> --
> You received this message because you are subscribed to the Google
> Groups "Microtransat" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to microtransat...@googlegroups.com.
> To view this discussion on the web, visit
> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fd%2Fmsgid%2Fmicrotransat%2Fc75d2269-062c-80d6-bfc4-4782f552c5d7%2540gmail.com&amp;data=05%7C01%7C%7C9947148b183c4448893808da9812ff1d%7Cd47b090e3f5a4ca084d09f89d269f175%7C0%7C0%7C637989504601539957%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&amp;sdata=mWb1mZ2mFJRvKD5lPwZH9oOd2ZEzEfZRQUjpaL8nudY%3D&amp;reserved=0.

Francis Roussel

unread,
Sep 23, 2022, 7:01:09 AM9/23/22
to microt...@googlegroups.com, Colin Sauze
Thanks Colin for this information.
I browsed a bit on micro ros to see the use of the ESP32, at one point
there was indicated a link with arduino ide but I lost it. I should be
able to find him.

On the other hand, I cannot grasp the interest of ROS.
It allows the use of already proven libraries, to communicate from the
computer to the microcontroller, and what else?

If we have made programs that run well separately on esp32.
- use of GPS (lots of examples on the web)
- use of a magnetic encoder (idem): wind vane
- a servo but with brushless or brushed motor with magnetic positioning
encoder
- use of an mpu9250 IMU and therefore roll, pitch, compass
- power cut module of different circuits with P mosfets depending on the
battery charge on a digispark and ka75330 for the Attiny itself

What's left?
- Putting it all together, not very complicated
- Manage the route to a WayPoint: recording of points, choice of the
best accessible depending on the heading: more complex
- Managing tacking: it would be interesting to see how in ardurover
sailing IamPete programmed it.
- Manage the irridium connection... For the moment without a sponsor I
have no financial solution but if everything else works we can interest
a reseller or other professional.

Extending the length to less than 8ft would accommodate any type of boat
with the most room available but for me that's not (yet) a concern, I
have a very stable old tenrater hull under sails for testing and can
accommodate up to 3 kg of payload, thanks to its very pot-bellied design.

I'll look at the link .../morwyn

Long live esp32!

fr

andy

unread,
Sep 23, 2022, 7:53:46 AM9/23/22
to Microtransat
I intended to use ESP32 because of low power consumption and also wireless programming, but I ended up with an Arduino-like board because of existing libraries to support different sensors. ESP32 had a poor support when compared to Arduino, but that was 5 years ago, maybe things are different now.

Colin Sauze

unread,
Sep 23, 2022, 8:02:53 AM9/23/22
to microt...@googlegroups.com
I thought you could program the ESP32 from the Arduino IDE now, but you have to download an extra package to do this.

Colin Sauze

unread,
Sep 23, 2022, 8:55:42 AM9/23/22
to microt...@googlegroups.com
The Raspberry Pi has been hit badly by the chip shortage. There was a short explanation from them last year, https://www.raspberrypi.com/news/supply-chain-shortages-and-our-first-ever-price-increase/

There's a great website which keeps track of Raspberry Pi availability called rpilocator (https://rpilocator.com/). Unfortunately for the most part it shows they're out of stock everywhere.

Colin.

On 19/09/2022 20:08, Francis Roussel wrote:
[RHYBUDD! E-BOST ALLANOL / CAUTION! EXTERNAL E-MAIL]

Magnus Leijonborg

unread,
Sep 23, 2022, 10:33:37 AM9/23/22
to Colin Sauze, Microtransat
This thread seems to wander all over the place so I add another question:

What is considered the best alternative for a MEMS Compass sensor at a reasonable price (say below 100Eur)?

This is my biggest concern right now in my autopilot tests. 

/Magnus 

Julian Blanco

unread,
Sep 23, 2022, 1:09:26 PM9/23/22
to Magnus Leijonborg, Colin Sauze, Microtransat

D A

unread,
Sep 23, 2022, 2:01:47 PM9/23/22
to microt...@googlegroups.com


On Fri, 23 Sep 2022 at 18:16, D A <dam...@axford.me.uk> wrote:
I’m part of a team developing an entry, we’ve elected to use esp32 as the compute device.  It’s powerful enough todo the real time control of an arduino as well as the higher level navigation planning etc of a RPI.  We have our own software stack and web management system developed, along with custom circuit boards - will release as open source once sea trials are underway.  We’re currently lake testing and building up the endurance.

For sensors, we have a MEMS compass, gps, wind direction (via magnetic absolute sensor), wind speed, IMU, power monitoring (current, voltage) and a selection of radios - medium range mesh, long range point to point, wifi and satellite.

An AIS module is in development to pickup shipping traffic and inform the navigation algorithm for collision avoidance. 

The boat is based on a fibreglass covered cnc cut foam core and a self trimming wing sail.  We built a cnc hot wire cutter to do the foam parts, which also makes prototyping fairly fast/cost effective.  Think we have about 12 test boats in our “fleet” at this point, so gaining lots of experience and having a great time testing/failing! 

Cheers
Damian

Francis Roussel

unread,
Sep 23, 2022, 2:48:47 PM9/23/22
to microt...@googlegroups.com

YES !!!!!!

Quick !!!!



Le 23/09/2022 à 20:01, D A a écrit :

Slava Asipenko

unread,
Sep 23, 2022, 3:31:45 PM9/23/22
to Magnus Leijonborg, Colin Sauze, Microtransat
I tried quite a few compass sensors, and IMO the best affordable one is CMPS14: https://www.robot-electronics.co.uk/cmps14-tilt-compensated-compass.html

Slava

Dermot Tynan

unread,
Sep 25, 2022, 8:55:48 AM9/25/22
to Microtransat on Google
On 23 Sep 2022, at 10:59, 'Colin Sauze' via Microtransat <microt...@googlegroups.com> wrote:
>
> Obviously since then we've had one successful completion
> of the challenge, but that was only in the (possibly) easier (and
> definitely shorter) West to East direction, only in the unmanned
> division where waypoint updates are permitted and with a commercial boat
> near the maximum size. There is still a lot of challenges to make
> smaller, simpler, cheaper, faster, more autonomous boats which can
> manage the East to West direction.

I definitely think the West to East direction is a lot easier, if you look at AIS heat maps for the two routes. A little sailboat wouldn’t have much fun with the typical traffic around Cape Finisterre. And now there are Orcas to deal with. But it’s definitely worth keeping as a goal. Eventually, the race could morph into a round-the-world autonomous race, but there’s a lot left to achieve between now and then.

> I've always tried to encourage people to open source their code and
> designs.

You can find a lot of my code here: https://github.com/kalopa

It all needs a big cleanup as a lot of it dates back from earlier thinking of how to divide up the problem, and there are recent additions I have yet to check in, pending further testing. I’d also like to push up the QEMU-driven emulation stuff because I think it would be interesting to do research such as combining the simulation with live AIS data, and evaluating different responses. There is a strong basis for stating that weather information isn’t useful as the boat is too slow to react. I have a suspicion the same is true of AIS, so it would be good to try it out in a simulated/live environment first.

> To throw in one more open source autopilot, I have written a minimum
> viable Arduino based autopilot with PI control of the rudder, a
> multiband if statement for sail setting and Rockblock tracking system.
> You can find the code for it at https://github.com/colinsauze/morwyn.
> This is build for a Honeywell HMC6343 compass and a SIRF3 GPS, with
> small adaptations it should work with any common NMEA GPS or I2C
> compass. This has had some testing, but it isn't particularly extensive.
> I think I've got some PCB diagrams somewhere too if anybody is
> interested in using this. I'm currently trying to persuade some students
> to use this code and port it to the ESP32, so I hope to have something
> better tested and for a more up to date platform later this year or next
> year.

I’ve done some design work with the ST32F400 series, which are also very powerful. Not sure what they’re like for power consumption though, as that wasn’t part of the remit. There is a youtube channel called “Phils Lab” which has some great introductions to designing with the ST chips, using CAD packages like Eagle and (my favourite) KiCAD, as well as PCB design/leyout, and even how to order fully-stuffed boards from JLCPCB in China. He is more on the Quadcopter side of things, but he does have an interesting series on designing and building a quadcopter controller using the ST32. I think it includes gyros, accelerometers, gps, compass, and so on. Could be a good project for someone to take Colin’s PCB designs and schematics, along with Phil’s tutorials on getting boards made.

One thing I’d recommend over Phil’s designs though is to use more daughter boards. JLCPCB won’t surface-mount any of the very fancy (and very delicate) sensor chips, so he hand-solders those himself. If you use a board from Sparkfun or Adafruit, you can instead put a socket on your own board and plug (or solder) the daughter board in place. The other nice thing about this approach is it gives you more space underneath. I did an AVR design which included an Si4463 radio module, and rather than try to get perfect impedance traces and all the other RF headaches, I used a pre-made daughter board which I soldered to my board. I also packed a bunch of level converters and passive components underneath the daughter board so the board was actually smaller.
- Der

Dermot Tynan
dty...@kalopa.com

Dermot Tynan

unread,
Sep 25, 2022, 8:59:14 AM9/25/22
to Francis Roussel, 'Colin Sauze' via Microtransat, Colin Sauze
On 23 Sep 2022, at 12:01, Francis Roussel <franc....@gmail.com> wrote:
>
> On the other hand, I cannot grasp the interest of ROS.
> It allows the use of already proven libraries, to communicate from the computer to the microcontroller, and what else?

I think it comes into its own if you have various payloads on board. For example, image processing. So you can have processes which manage the operation of the boat, and then leverage off the extensive libraries/processes for more typical robotics applications (such as vision). Probably more useful for the WRSC than the Microtransat, but that’s just my opinion.

Paul Miller

unread,
Sep 25, 2022, 9:15:21 AM9/25/22
to 'Colin Sauze' via Microtransat

Hi All,

 

Way back in 2011 we did a study of the routes and summarized it in a paper at the IRSC. It might be useful in this discussion. Here is a copy of the conclusion,

 

Climatological data provided the means for determining likely times for different routes. Coastal Pilots provided information for specific departure and arrival conditions. The actual selection also included an analysis of shipping traffic. Both the Northern and Southern Routes have significant traffic, although established shipping lanes and local traffic patterns are well-documented and can be avoided. Figure 5, for instance, shows the shipping density around Newfoundland.

Choosing between the Northern and Southern Routes became a question of the probability of success and mitigating potential problems. A first-order approach used the algebra of random variables based on the most likely voyage time. For instance, the reliability of the three prior USNA SailBots during sailing was calculated from testing logs as 0.9985/hr. This would imply a probability of success for the Northern Route of 58% and for the Southern Route, 33%. The uncertainty however puts this in to question as the new boat uses different, and supposedly more reliable, equipment. Additionally, while both routes have similar wind and sea conditions, the Northern Route has some potential for ice and slightly higher traffic, particularly near Ireland. Using a linear probability for ice encounter based on the extent of the berg limit versus the pack ice line and the route through the ice zone, the Northern Route drops to 49%. Ice travels slowly however and short-term forecasting can mitigate that problem. Finally, the Northern Route will provide only 50% probability of sunlight, versus nearly 100% for the Southern Route(3). This may be significant if solar panels are the primary means of generating power.

The predicted time to sail the Southern Route was about 50% longer than the Northern Route, but the Southern Route had a greater probability of following winds, which are much easier to sail for a small boat than head winds.

 

Anecdotally, of the 7 attempts I’ve been involved with leaving from Cape Cod, 5 were “pirated” by fishing vessels. That was not included in the study!

 

Cheers,

Paul

 

From: Dermot Tynan
Sent: Sunday, September 25, 2022 8:55 AM
To: Microtransat on Google
Subject: Re: [microtransat] Choix

 

On 23 Sep 2022, at 10:59, 'Colin Sauze' via Microtransat <microt...@googlegroups.com> wrote:

>

> Obviously since then we've had one successful completion

> of the challenge, but that was only in the (possibly) easier (and

> definitely shorter) West to East direction, only in the unmanned

> division where waypoint updates are permitted and with a commercial boat

> near the maximum size. There is still a lot of challenges to make

> smaller, simpler, cheaper, faster, more autonomous boats which can

> manage the East to West direction.

 

I definitely think the West to East direction is a lot easier, if you look at AIS heat maps for the two routes. A little sailboat wouldn’t have much fun with the typical traffic around Cape Finisterre. And now there are Orcas to deal with. But it’s definitely worth keeping as a goal Eventually, the race could morph into a round-the-world autonomous race, but there’s a lot left to achieve between now and then.

 

> I've always tried to encourage people to open source their code and

> designs.

 

You can find a lot of my code here: https://github.com/kalopa

 

It all needs a big cleanup as a lot of it dates back from earlier thinking of how to divide up the problem, and there are recent additions I have yet to check in, pending further testing. I’d also like to push up the QEMU-driven emulation stuff because I think it would be interesting to do research such as combining the simulation with live AIS data, and evaluating different responses. There is a strong basis for stating that weather information isn’t useful as the boat is too slow to react. I have a suspicion the same is true of AIS, so it would be good to try it out in a simulated/live environment first.

 

> To throw in one more open source autopilot, I have written a minimum

> viable Arduino based autopilot with PI control of the rudder, a

> multiband if statement for sail setting and Rockblock tracking system.

> You can find the code for it at https://github.com/colinsauze/morwyn.

> This is build for a Honeywell HMC6343 compass and a SIRF3 GPS, with

> small adaptations it should work with any common NMEA GPS or I2C

> compass This has had some testing, but it isn't particularly extensive.

> I think I've got some PCB diagrams somewhere too if anybody is

> interested in using this. I'm currently trying to persuade some students

> to use this code and port it to the ESP32, so I hope to have something

> better tested and for a more up to date platform later this year or next

> year.

 

I’ve done some design work with the ST32F400 series, which are also very powerful. Not sure what they’re like for power consumption though, as that wasn’t part of the remit. There is a youtube channel called “Phils Lab” which has some great introductions to designing with the ST chips, using CAD packages like Eagle and (my favourite) KiCAD, as well as PCB design/leyout, and even how to order fully-stuffed boards from JLCPCB in China. He is more on the Quadcopter side of things, but he does have an interesting series on designing and building a quadcopter controller using the ST32. I think it includes gyros, accelerometers, gps, compass, and so on. Could be a good project for someone to take Colin’s PCB designs and schematics, along with Phil’s tutorials on getting boards made.

 

One thing I’d recommend over Phil’s designs though is to use more daughter boards. JLCPCB won’t surface-mount any of the very fancy (and very delicate) sensor chips, so he hand-solders those himself. If you use a board from Sparkfun or Adafruit, you can instead put a socket on your own board and plug (or solder) the daughter board in place. The other nice thing about this approach is it gives you more space underneath. I did an AVR design which included an Si4463 radio module, and rather than try to get perfect impedance traces and all the other RF headaches, I used a pre-made daughter board which I soldered to my board. I also packed a bunch of level converters and passive components underneath the daughter board so the board was actually smaller.

                                                                  - Der

Dermot Tynan

dty...@kalopa.com

 

--

You received this message because you are subscribed to the Google Groups "Microtransat" group.

To unsubscribe from this group and stop receiving emails from it, send an email to microtransat...@googlegroups.com.

Colin Sauze

unread,
Sep 27, 2022, 7:29:50 AM9/27/22
to Microtransat
I'd second the CMPS range, I've not tried the CMPS14 yet but had good results from the CMPS11 and CMPS12. They were probably the best compromise of accuracy and cost that I found, although the power consumption was a little bit higher than my preferred (but very expensive) option of the Honeywell HMC6343.

Colin.

Philip Smith

unread,
Oct 7, 2022, 7:16:56 PM10/7/22
to Microtransat
Some notes on my experience of creating an autopilot using a micro.

My first boat used an Arduino, and I'm now using an STmicro. Basically, the Arduino was sufficient to control everything essential, ie compass, GPS, rudder servo, Iridium (satellite comms). BUT it uses a lot of current if on all the time (about 30mA) which I thought was too much, so I put it into sleep mode most of the time, servicing the rudder PID loop at about 30Hz. GPS and Iridium are every few hours. The problem with sleep is that the standard PWM-out stops working so I added an external PWM generator updated over I2C every PID loop. I could also plug in an LCD (display) for real-time data display, although this wasn't on the boat for its journey. Also for testing I could connect an RC receiver.

The benefit of using an Arduino is that it is pretty simple to program as things like I2C, UART, PWM (in and out) are easy to implement. Also there are available libraries for compass, GPS, Iridium etc so adding these in is easy too. I used a pretty basic one and almost ran out of program space, but just managed to fit everything into 32k. It does floating point (although division is pretty slow) and every other mathematical function you need (sin, cos, sqrt etc). So for a relative software beginner I would strongly recommend an Arduino. 

In order to learn new things, for the second boat I'm working on I'm using a micro made by STmicroelectronics, which has an ARM core (part number is STM32L412RBT6P). The ST micro compared to the Arduino is much, much, more difficult to use. For example, you have to program your own PWM using timers, program your own way of reading the UART (nothing as simple as a stream is provided), and basically program everything yourself from a much more fundamental starting point. I probably took about ten times as long to do anything as it would have taken on the Arduino. Or twenty. The benefit is that it is very low power (1mA) and also has sleep modes for even lower current: I didn't need to use them. Because sleep is not needed the micro can drive the rudder servo PWM directly. There are also timer driven interrupts available which mean you can run the rudder PID, and compass at a predictable rate (of say 25Hz) so slow functions like updating the ePaper display can happen without affecting steering. GPS and Iridium can also happen at a slower rate. The processor is 32bit so very fast at maths, and has good floating point capabilities (I think it has a hardware FP multiplier, so very fast). It also has more I/O pins than the Arduino so you can connect more gadgets to it. Libraries for the Arduino need significant modification to be suitable for the ST. If you enjoy software it is a fun challenge, but I wouldn't recommend it unless you have significant software experience. 

I used a Nucleo board which is ST's ready made board with the micro on, header pins, and a second micro that allows debugging, including real time monitoring of internal variables. This is very useful. Part number is NUCLEO-L412RB-1. The L is for low power which I recommend. This is probably out of date now, but there will be something equivalent, or better. If you disconnect some jumpers and solder links it will run at impressively low current, and voltage.

I have noticed there is something called a blue pill, which I think is an STmicro chip, but you program it like an Arduino. This might give the best of both worlds, but I haven't investigated it, and I might have got the name wrong too.

My compass is also made by ST: LSM303DHLC which is easy to use, but now out of date. Get the latest LSM303 (AGR I think) from Adafruit or similar already mounted on a breakout board, and use their library. Very low current (<<1mA) so not worth turning off.

NOTE that I haven't got anything on the boat that needs significant processing power: no camera, AIS or some AI based system to detect other craft etc. So if you think you want to do that then you may need something more powerful than I've used. However, I did look at a Pi, and found it needed a huge current (500mA or so) so I decided it wasn't the right thing at all for a solar/battery powered device.

I don't see the need for an RTOS as the looping nature of both the Arduino and STmicro simulate parallel processing, and as long as the processor can loop fast enough you get real-time enough to control a small boat. 30Hz is plenty fast enough. You just have to schedule the important things often enough, and don't try to do everything at that rate. Eg the GPS will typically only update at 1Hz, but you might want to read the compass much more often to determine the rotation speed of your craft.

There is one possible reason for using weather information on the boat which is to determine the wind direction at that moment. Ie ask for a forecast for now at the boat's position and you will get the wind direction. This is probably easier than adding a wind vane, or ultrasonic wind direction sensor. I would think that wind direction over the sea is pretty accurate, and getting a new forecast every few hours should be sufficient. I agree though that you won't be able to escape any weather you would prefer to avoid. 

Any other questions please ask. Hope this has been useful to someone.

Phil

John Lamport

unread,
Oct 10, 2022, 8:18:38 AM10/10/22
to Philip Smith, Microtransat

I would like to add that I have been following the Microtransat challenge for a number of years and have always enjoyed the discussions about boats, hardware and different methodologies especially these last few threads about the different controllers and programming structures. I have been working on a Arduino/Raspberry Pi systems for a while, Colin I will check out your autopilot. I am not planning on sending a boat across the Atlantic, my ultimate goal is to enter in the International Robotic Sailing Regatta Open class not the Sailbot class, but that’s a ways away before I’m ready for that. Although I am not going across an ocean and just around a lake, the technology is the same, other than I will have the luxury of changing batteries between challenges.

 

I also belong to a radio control sailboat club so this lets me combine my two hobbies.

 

I agree there is nothing for us to prove other than to ourselves.  I find this a fascinating challenge to myself, plus I’d rather be working on this than watching TV.

 

Thank you all for sharing your knowledge and passion

John

 


John Lamport

Controls Engineer

Munters Corporation
79 Monroe Street, Amesbury, MA 01913 United States
Tel dir: 978-241-1196,

E-mail: john.l...@munters.com,



FACEBOOKTWITTER | YOUTUBE | LINKEDIN

Munterslogo.gif

This e-mail is solely intended for the addressee(s) and may be of a confidential nature. If received in error, kindly notify the sender and delete the e-mail. E-mail is susceptible to data corruption, interception, unauthorized amendment, tampering and viruses, and we only send and receive e-mails on the basis that we are not liable for any such corruption, interception, amendment, tampering or viruses or any consequences thereof. Any opinions expressed in this message are those of the author and do not necessarily reflect the opinions of Munters.

From: microt...@googlegroups.com <microt...@googlegroups.com> On Behalf Of Philip Smith
Sent: Friday, October 7, 2022 7:03 PM
To: Microtransat <microt...@googlegroups.com>
Subject: Re: [microtransat] Compass sensors (was Choix)

 

CAUTION: This email originated from outside of Munters. Do not click links or open attachments unless you recognize the sender and know the content is safe.

 

Some notes on my experience of creating an autopilot using a micro.

Reply all
Reply to author
Forward
0 new messages