Autonomous vehicles are complicated, undoubtedly, but that didn’t stop us from endeavoring to be the first to launch a successful autonomous navigation of the Atlantic Ocean. What began as a joke between college kids quickly morphed into a challenge, an obsession, and years later, careers in the industry and a startup of our own.
The first iterations of the boat were designed in 2010 with the goal of traversing the Atlantic autonomously. Since it had never been done before and the journey across the vast ocean was largely unknown and mysterious, we called the project Scout. The name was also a shout-out to the Age of Empires video game whose scouts we would send off exploring into dark dangerous worlds during the wee hours of the morning.
The fateful joke that started the project was cracked between Dylan Rodriguez and Max Kramers in Max’s parents’ garage: They’d have to build a robot boat to keep in touch while Max studied in Spain. The project quickly wrapped in more of the neighborhood crew, with Brendan Prior and Dan Flanigan joining the fun. One day I followed Dan, my older brother, into the garage to see what the tinkering was all about.
That garage became the site of a hands-on engineering and maker education for all of us. After our summer jobs we would meet up to sand, solder, and hack away at whatever the current hurdle was. One of my favorite aspects was the frequent pitstops at a whiteboard hanging on the wall, where rapid lessons in basic circuits would come from Dylan, stress and strain from Dan, or lift and hydrodynamics from Max, then after covering the background on why something was being done, we’d jump back to building it. The pace was fast and our target launch date ever approaching. It was full commitment, bordering on over-obsession.
But the project didn’t stop with a few kids in a garage. Building an autonomous boat takes a village, or in this case the town of Tiverton, Rhode Island. The project quickly engulfed parents, friends, neighbors, and much of the local community. Some stopped in for an hour of sanding; others made monumental efforts in key technical areas. Tom Schindler knew Dylan from FIRST Robotics and was psyched to add his software skills to the team when we were closing in on launch day. For Tom it was a no-brainer: “These kids were up to something cool, and as soon as I heard about the project, I wanted in.” Mike Mills, owner of the local marine and woodworking supply Jamestown Distributors, offered to sponsor the project with all the composite materials we needed.
Along with the moral support and helping hands, there were quite a few chuckles at the endless late nights and the sheer scope of the ambitious endeavor. And with no successful Atlantic crossings to date, despite dozens of attempts by teams on either side of the ocean, ambitious it was.
Building a Trans-Atlantic ASV
The initial design for Scout contained just the basic requisite components: a motor from an R/C airplane, an Arduino backbone, a rudder servo, and a Tupperware to hold it all.
But as the project evolved, so did the systems needed for a true autonomous surface vehicle (ASV). While neighbor Greg “Jonesy” Jones was debugging signal noise issues, our young team was busy implementing current sensing, data telemetry, and a hacked GoPro intended to take a few minutes of video each day. The original 3-foot concept craft sank in a pond after an hour, but from that soggy grave it returned and, through six generations, grew to a craft measuring 12 feet.
Much of the complexity involved communication between various technologies. Some components, like the INA219 current sensor breakouts from Adafruit, EM-406 GPS from SparkFun, and LSM303 compass from Pololu, had sample Arduino code that made testing easy. Others, like the Quake Q9612 Iridium transceiver, were more difficult to integrate. At Worcester Polytechnic Institute, Dylan met Ryan Muller who joined the team and spent weekends writing code in the garage for the boat to communicate with the Iridium module. Nowadays, options like the RockBlock module make adding satellite communications to embedded projects much easier.
A landmark day for our team was Scout’s first successful autonomous mission navigation, which was supervised over lunch at a nearby sandwich shop. “That was one of our proudest days,” recalls Dylan. “One of the few that didn’t end in paddling after a runaway craft, debugging the IMU, or falling asleep at the bench.” For the team, it was the first day of true autonomy, and a taste of encouraging reward that helped make up for skipped hangouts and late nights.
The original “two-week timeframe” stretched into three years and Scout accumulated 2,500 followers on Facebook. Our local paper The Sakonnet Times ran several articles on the project including one titled “A Slow Boat to Spain.” And finally, on August 24, 2013, after two failed excursions and countless other tests, Scout took to the high seas on its third, record-setting attempt. (Make: reporter Andrew Terranova wrote about the launch.)
A local legend by this point, Scout’s public tracking page amassed 750,000 views during its final 5-week journey. It powered through days and most nights, occasionally falling asleep on cloudy days. The page was updated with the latest data every 20 minutes. Friends and followers jokingly complained that it was hard to do their work while excitedly refreshing the page hoping for mission updates.
Sadly, after more than 1,200 miles, Scout encountered unknown difficulties and fell off the map, so to speak. Our team of excited young engineers, young adults by this point, hung up our hats and parted ways. “We were proud, and devastated, by what we had accomplished,” says Dylan, “but we decided it was time to move on to bigger and better things — that’s what we thought, at least.”
Dylan got recruited to an autonomous boat company based in San Diego. Max began working for high-performance boat shop Guck Inc., Dan for Reichel-Pugh Yacht Design, Mike for DMC Engineering, and Brendan for Bristol Marine. Tom continued his work with NAVSEA and the team all stayed in touch, one way or another.
But the story didn’t end there. Over the next seven years, the notion of returning for another attempt continued to bounce around. And as it did, we were gaining industry skills and experience that would be invaluable for our
As we worked at companies and tracked the progress of the growing ASV scene, we had a common feeling that the costs and complexities of ASV products were too high. Also, it was clear that the companies building these ASVs were ignoring relevant open-source projects that could leapfrog autonomous technology.
Sketches began to get thrown around again and before long a prototype emerged. “We’ve long felt that we have something to add to the ASV space,” says Dan Flanigan. “Scout formed the ashes from which Seasats will rise.”
Armed with professional experience, this attempt was a bit different. The Seasats project kicked off in 2020 as a side project of Kroova LLC, the engineering consulting company that Dan, Max, and Dylan had launched three years earlier. Kroova gave the team a proper shop and a slim budget with which to work. The cycle of building, testing, and failing resumed, with the occasional success to nudge things onward. By the end of 2020, the team had built a craft that was less expensive and more effective than some of the million-dollar products currently being sold to the US Navy.
That’s when we started to understand the opportunity we had. Over the years we had unknowingly accumulated a niche skillset to build just what the market was missing: long-range, autonomous surface vehicles designed for months at sea, but easy enough to launch in minutes, and at a price point that could let the advantages of autonomy reach a wider community.
Today Seasats is in active development. Last winter we ran missions for the nearby Scripps Institute of Oceanography, providing water samples regularly and autonomously. Now we’re working with customers in whale tracking, offshore wind power, and the U.S. Navy.
It’d be great to say we came up with an ethos, determined the company vision, and then stuck to that with a laser focus, but the truth is that we started as makers and tinkerers. And honestly today, we’re still making and tinkering. But we’re no longer in a garage, and we’re no longer quite so clueless.
Trans-Pacific Is Next
Still, the challenge of crossing an ocean lingers in all of our minds. Now located in San Diego we see the Pacific Ocean every day and imagine the challenges and possibilities. Later this year, we will attempt to cross that ocean with one of our Seasats. Let’s call it Scout 2.0.
“Victory isn’t guaranteed, and in many ways the fear of failure is stronger now,” says Dylan, “It’s easy to be the underdogs, a couple of kids who no one expects to succeed. Now that we look the part, and know enough to be leaders in our field, there’s some expectation that we should be able to do this.”
But the ocean is a dangerous place with no shortage of obstacles. Stormy seas, overcast days, corrosive salt water, garbage patches, and solar-deck-pooping seagulls all stand in our way. But just like last time, we’ll forge forward undaunted. “Once you accept uncertainty as the only certainty,” Dylan says, “you begin to realize that it’s possible.”
At its most basic form an autonomous boat is a great DIY project that you can tackle with basic hobby parts and relatively few components. The description here outlines a high-level approach to making a boat that could work well in bays or ponds.
Why high-level? Because part of the fun of being a maker is being a believer in guidelines, not rules! So if you don’t have a boogie board, that’s fine, maybe use a couple soda bottles! Or if you don’t want to buy a motor for the propulsion, but have an old battery powered fan, great, make it an airboat!
Start with the following core components, indicated in the illustration above.
1. Motor and propeller Propels the boat forward. There are lots of options and mounting configurations, but some of the simplest require either a long shaft to keep the motor dry while the propeller is in the water or else a waterproof motor.
- Simple brushed motor with shaft — Banggood #1367490, usa.banggood.com
- Blue Robotics thruster T200, bluerobotics.com
2. Steering servo Directs the flow of water in order to steer the boat. There are two common methods for steering. Method 1 uses a fixed motor and a separate rudder (a piece of wood or plastic can work well). Method 2 cuts out the need for a rudder by simply aiming the output of the motor; if you have a waterproof motor this can be a great option. Either configuration could use a servomotor like this one:
- Micro continuous rotation servo Feetech FS90R, Pololu 2820, pololu.com
3. GPS This is a sensor that triangulates its position from overhead satellites and can tell the boat where it is. A great thing about Adafruit’s GPS is that they provide fantastic starter code and wiring.
- Ultimate GPS Breakout module — Adafruit 746, adafruit.com
4. Compass This sensor measures Earth’s magnetic fields to tell the boat which direction it’s pointing. A critical thing to know when trying to navigate at sea! Once again, Adafruit’s part has excellent starter code and wiring.
- Triple-axis accelerometer / magnetometer Adafruit 1120
TIP: Don’t mount the compass anywhere near the motor, motor cables, or big pieces of metal! They can cause interfering magnetic fields that mess up your boats navigation.
5. Controller A brain of sorts is needed to read the input sensors and, based on commands programmed into its software, direct output signals to the motor and steering servo. Arduinos of all kinds are an excellent option for a low-cost, easy-to-use, controller.
- Arduino microcontroller boards arduino.cc
6. Battery You’ll want to let your boat go free and wander about in the water, so you’ll need a battery to power it. For convenience, choose something with a similar voltage to your controller, motor, and servo to cut down on the number of voltage regulators you’ll need.
7. Hull You need your boat to float! This can be a wonderful place for creativity and fun. Use whatever you can get your hands on: a boogie board, a plastic box, soda bottles, your choice.
OK, so with those core components you have all the necessary parts for your very own DIY autonomous boat. Maybe you’ll add LEDs (always an awesome decision) or a solar panel to charge your battery. But aside from those fun bits, how do you get your boat to actually start doing things on its own? That’s where software comes in.
Above, shows a basic loop with the steps needed for an autonomous boat to drive to a waypoint. Seems simple? Well, there are steps left out, but a helpful lesson for software is to keep things in small, testable chunks. If you follow the general steps below and work on making each line work on its own with the necessary sensors and motors then you’ll be well on your way to a working boat, minimizing the amount of frustration driven by heaps of code that doesn’t work. For software the time-old maker rule is still king: KISS — Keep It Simple, Stupid!
That’s one way to get to a functional autonomous boat. Another way we would highly recommend is to look into the fantastic open-source community project ArduPilot and, even better, its ArduBoat section. The community has detailed instructions, components, and software to get you up and
running err floating, fast!
Autonomous Boat Do’s and Don’ts:
Some Lessons We Learned the Hard Way
- Budget ¼ of your timeline for design, ¼ for build, and ½ for testing. At least ½.
- Include a bilge pump — regardless of your confidence in your boat’s watertightness.
- Test core components like the motor and rudder servo — for months — on the bench.
- Select important components like solar panels and batteries early in the design phase — before building your boat.
- Compile and use a testing checklist. It’s too easy to forget to bring the rudder, the custom-made Allen key, or the spare battery when you head out for testing at 2am.
- Set a launch date ASAP. Announce it loudly and confidently.
- Glue all hatches shut just prior to launch, so no one can tamper with your boat.
- Pot all connections in epoxy, test the system, and then rework as needed.
- Find the limit of your telemetry range by driving the boat away from you.
- Keep the keel bulb as light as possible to save on weight.
Ro, Ro, Roboats: More USV Projects We Love
SEA CHARGER – Website
Damon McMillan’s solar-powered boat autonomously navigated 2,413 miles from California to Hawaii in 2016 (see “The Little Boat that Could,” Make: Volume 55). Then he sent it along to New Zealand for the heck of
it, and nearly made it; after 6,480 nautical miles the rudder failed, but still! Amazing. Now, Damon says, “I am actively working on a version 2.0 that I hope to be able to sell as a kit for others who want to do this type of thing.”
LOCARB – Website
Inspired by McMillan’s feat, Adrian Li built a Low-Cost Autonomous Robot Boat for less than $2,000, based on an R/C motor and ESC, a 110W Sunpower solar panel, Arduino Mega microcontroller, PixHawk 1 autopilot, and RockBlock Iridium satellite modem. After two shakedown attempts last year, he launched it May 30 from California to Hawaii and it made 615 miles before succumbing to spring winds and leaks in the battery compartment. Try, try again. Adrian has shared his designs, code, and tips.
OPEN OCEAN ROBOTICS – Website
Husband-and-wife explorers Julie and Colin Angus spent years doing water expeditions for National Geographic, then launched this startup in Tofino, British Columbia, building robust, surf-worthy USVs to gather data on the environment and sea life on Canada’s seemingly endless coastlines.