SolarSurfer Continuation Project

@luisgamez Thanks for sharing! Yeah we definitely can’t compete with the capabilities and ruggedness of a Wave Glider, but it is an interesting design to take notes from. Good luck with your expedition!

@rdn11180 The Envirover satellite communications system is up and running with a RPi. GitHub - envirover/SPLRadioRoom: Global satellite telemetry for MAVLink autopilots I have a USB cable plugged into the RPi to the Pixhawk and that works well. Because of that, I had to ditch my WiFi connection for the moment.

@kevink This is great news. Thank you for sharing. I have been busy designing and testing drive components for the boat. I finally worked out the magnetic coupler and now have a working printed prototype. The actual hull has been designed with CG/COB for decent stability. It was a challenge getting this right while also trying to eliminate areas where snags from debris could occur. I’m trying to build it with .080 in. aluminum so I can keep my father involved in the build (he is a welder). We may move to composites in the future if the aluminum doesn’t work. I am now building a ground based rover that will use the rudder mechanism, drive system and autopilot for testing navigation.

For now I will be using ardurover and mission planner. It’s what I am most comfortable with for autonomous ops. Are you still using ardusub for your surface vehicle?


That looks like an outstanding hull and component set-up to me. It looks to incorporate a lot of the lessons learned over the years.

I’ll be interested to see pretty much everything on this, especially the rudder. Keeping rotary joints watertight has always been a challenge.

Yes, I’m still using ArduSub at the moment. I prefer QGC over Mission Planner and I like the Manual Control over Mavlink. Rover looks to have been stuck in development all summer and is moving more towards UGVs with wheel encoders and such. I have a feeling we need a dedicated “ArduBoat” firmware due to the multitude of propulsion options for boats, different sensors, and dynamic positioning.

I’m hoping this hull shape will help with entanglements. It’s probably not the final design but should get us started.

Currently the rudder shaft is supported by a couple of bearings and also a few shaft seals. Like you, I’m a bit worried these may leak. I originally planned on incorporating an axial magnetic coupling but my prototype was really weak. I’ve been super impressed by how strong the radial magnetic coupling is working for the motor drive. I may have to use this for the rudder. I’m still worried that the rudder would lose positioning. It could be a real problem if that happened. Mag coupling scheme below:

Have you been able to get auto missions working with Ardusub? I have a GPS in my ROV so I may have to try auto modes this weekend. I do agree that a dedicated ArduBoat is needed!

Your mag coupling looks like it should do the job.

I’ve only managed to test auto missions on the bench, but when I change the mode to AUTO, the thrusters take off and it wants to try to navigate to the waypoint. Jacob has mentioned that there might be an issue with surface vehicles and waypoints: Way point following issues for surface vehicle - #8 by jwalser I’m at a hard point because I want the manual control and sensors from ArduSub, but the dedicated 2D waypoint and control options (skid steer and conventional) from ArduRover.

Because the Simple-3 frame uses servo mixing in a default skid steer, you would need something like a reverse servo mixer to take the servo outputs and convert them back into a throttle and rudder control for your USV.

@rdn11180 When you get your Iridium modem up and running and connected to your Pixhawk, be sure to try out the tracking feature: https://github.com/envirover/SPLGroundControl/wiki/SPLStream-and-SPLTracks-Web-Services

@kevink I actually just ordered the rockblock MK2 yesterday during Sparkfun’s cyber Monday sale. Perfect timing! The Envirover interface looks really good. When I get all of the components in, I may need to pick your brain for setup help.

Yesterday I was able to manually drive my test rover around. In auto modes it seemed to reach the first waypoint and stop. I haven’t had a chance to dig into it further. It does sound like Randy Mackay is about to roll out the next stable release with a few fixes so i may wait until that time. I am using Rover3.2-rc2 and have noticed documentation/mission planner isn’t up to date with this.

Sure thing! I’ll be here. I’m looking forward to seeing more DIY vehicles out prowling the ocean and the Microtransat looks neat. Unfortunately, nothing here in the Pacific and Hawaii has already been done.

Hmm, yeah Rover documentation is lagging quite a bit. I’m looking forward to seeing your experiments, it looks like you have an excellent platform.

@kevink Have you guys tried both the Arduino 101 and Pi for the SPL RadioRoom sat solution? I am thinking the arduino may be a bit more efficient compared to the Pi. I wonder if the Pi could have more issues if rebooting was required during a low power cycle? Any thoughts?

I tried both the Arduino 101 and the Pi solutions. I could never get the 101 version to work properly even though everything was connected correctly and was able to transmit. For some reason, my high latency Mavlink messages didn’t include the payload portion of the message. Maybe you’ll have better luck, but I gave up.

The Pi version is working fine for me. If power is a concern, you could always try the other Pi versions, like the Model 1 B+ or a Pi Zero W. I think those do better on power consumption than a Pi 3.

Ok thanks for the input. I will give it a try with a Pi 3 first and see how it goes.

My dad and I made some progress on our solar boat. I was able to cut all of the aluminum parts on my cnc in Kentucky and drive them up to Vermont for Christmas vacation. Fitting the parts in the car was a bit of a challenge but we made it work.

We spent a few days assembling the parts and tacking them together. My dad’s welding experience really helped with welding the thin aluminum. It’s nice to finally see the real boat after staring at the cad model for so long!

Looking good!


I would like to introduce the “Mike”; a project I’m working on for the last 3 years, which was initially based on the SolarSurfer .

My goal was from the beginning, to design a solar powered USV, remotely controlled by satellite, with the possibility to steer near shore in undeep waters, making photos and movies on demand. But also the ability to travel long distances, preferably crossing an ocean.

Best thing about steering with 2 thrusters is the lack of a rudder. A constantly moving rudder is a weak point in long sailing DIY USV designs, just as Seacharger’s final days made clear. It will wear out, start leaking, or the multiple moving interactive parts will give way sooner or later.

Second important design feature, is a minimum “plastic bag” and seaweed catching vertical underwater surface. Not only to minimize the chance on a trip-ending collision, but as well to reduce drag.

This made me change the steering/propelling by 2 central placed thrusters, to a T200 at the back for propelling, and a T100 bow thruster for steering.

A self righting weighed deep keel was no option, preventing sailing near the coast for taking cool pictures over there, and because it’s also a drag increasing plastic bag catcher.

Hence the need for a righting system on board, in case the boat is turned upside down.

The system I came up with works fine in theory, as per finite elements floating simulations, but still need to be proven in reality. Only moving parts of that system are a low cost air pump, and some coffee machine solenoid valves.

Sailing towards the coast could be save enough, if depth can be monitored by sonar.

I can’t find any affordable depth/fish finder who can be connected to a Raspberry Pi, to be useful. My solution is to use a cheap hand hold fish finder, put it in glued-Lego made jig with a lens, LED and Raspberry Pi camera pointing at its display, and analyzing the readout to get the depths in Python were I need it. The fish finder is turned on by a small servo, pushing the on/off-button.

This system I have tested in a swimming pool and works fine, but I really would like an alternative…

To prevent the possibility of leakage, which most likely results in death-of-vehicle, I ended up with a concept of glued together fiberglass top and bottom parts, which are after assembly with all inner parts, foamed full with closed cell marine foam.

There will be a front and back part with each a 120W solar panel on top, which can be screwed together on a launching beach, so there is no need to transport the whole boat long (10.8 foot)and heavy (120+ kg) in one piece.

Some potting of electric and electronic connections will have to be done on the beach as well, because water tight connectors are too big, or not reliable if not of industrial price.

The solar energy will be stored in 6 sets of each 4 15Ah LiFePO4 batteries. 1 of these sets is only to be used for the electronics, and so 5 sets left for propelling and steering.

Dedicated Arduinos will monitor voltage of each battery during charging and unloading, and do the needed sets-switchings by relays. The goal is to sail 24h/day, and in theory and by simulations, a speed of up to 1.5 m/s should be full time possible. Steering should take about 4% of energy available.

The whole boat will be controlled by 1 Raspberry Pi, and 4 Arduinos. The Pi should steer against found current, so that a straight course between 2 waypoints is possible. This works fine by simulation. I spend a lot of time making simulation software, because “testing” in real live is way too much work and complicated. It just will have to work.

By Rockblock, the plan is to get every day 8 positions, which will be made into kml files for Google Earth, for all who are interested, to follow. Lots of other information from the boat and its findings and workings will be sent by satellite messages as well, mostly made part of the kml update.

Currently, the positives for the 4 fiber glass parts are being made, so I hope to start assembling from next month. Metal parts are finished, and most of the electrical and electronic parts are in house, and are being tested with a whole lot of software.

I do this all in Vietnam, where I work as an engineering production sourcing consultant. That is great for keeping manufacturing of the metal (including the wire mesh stainless steel welded deflector for T200) and fiberglass parts low cost, but not easy to get your hands on stuff like the needed marine foam resins, and other chemicals to glue and seal the solar panels. I still need a way to ship these from the US to here (must come by sea).

I have not much time to keep a building blog for this project, but I will here give now and then updates, and hope to launch from a Vietnamese beach begin 2019. I will give link to the daily kml files for tracking, and will inform about actions and hopefully progress. The aim is to sail north of the Philippines to the Pacific, and from there to California, making photos and movies during storms (tilt-meter triggered) and during “raids” to island beaches.

All photos and videos will be stored on a USB stick (in a sealed screw-cap bottle), for somebody to collect and upload, so I can add links in the final kml file to them.

But that is still all far away…

Cas




3 Likes

So cool Cas, great design and impressed with how close u are to launching. The electronics r a good combination.
Have u worked out collision avoidance yet? Lidar? Machine vision/learning?

Thanks Kaos,
Next to said electronics, there are also “temporarily” turned-on ones; 2 Arduinos for discharging each individual battery over a resistor during a given time; if needed to level a set off. And 1 Pi for the Depth Meter.
All switching on/off, of these boards, of RockBlock, and even the USB memory stick and the high current power/charging relays, is done by relays (so not by MOSFETs) to save power.

There is no system to prevent collision. Total weight is not much more than a floating log, and not much damage can be caused on both sides.
If GPS coordinates don’t change enough during some time, Mike might be blocked, after which he will sail backwards by force, make a photo, and go back to last waypoint. This, because I worry most about lines and ropes in water of shrimp farms etc. I see all over the SEAsian coast.

Well thought out with power management n understand uneasiness of it getting tangled.
Marine AIS could transmit yr id details to larger ships so they would avoid you, but that would be more power drain. Im sure you don’t need that.

Well if you ever want a full on ocean challenge, send it down my way to New Zealand. Otherwise, keep me in the loop coz would be fun to track the journey of Mike.
Keep up the great work

@CasT Wow, this is an excellent design! I am very impressed that you thought about using a bow thruster as the rudder to correct the heading. What an excellent idea and I wish I would have thought of that!

We’ll have a solution for your depth sounder problem coming out soon (a month or two) that will work with a Raspberry Pi. It will have a Python library for you to work with.

When you’re ready to send it across the Pacific to California, give us a warning and we’ll be ready to pick it up and keep it safe for you.

1 Like

Thanks Kevin,
It’s great to hear you guys will soon have a proper solution for sonar depth finding.
See below current solution; it does work, but as I’m not thinking only do this ones, a device without moving parts and optical risks would be much better.

You might be interested in the low-drag thruster deflector I has made in China. It supposed to fit tight over outside diameter, but they made it a few mm to big. I think that’s not a problem, and I’m planning to fill the distance, and fixate, by hot glue gun. As outside-ring gets smaller at the backside, this should prevent the deflector from ability to be pulled-off to the front. But maybe I can find a better solution with small clips. I had to invest in tooling just to get some samples (luckily I have good contacts in the Shenzhen wire mesh industry), and have a reasonable MOQ.
I will update after start assembling it with the T200.