Skip to main content

A little red boat and the game of drones

On a recent trip to Europe, I found myself on the trail of autonomous sailing vessels, a technology that has been under development for a while and is now getting more attention with the advent of package delivery by drone and driverless cars. No longer is getting things to drive themselves a hobby for remote-control geeks; it’s a business on the verge of becoming big, built on the promise of — what else? — eliminating fallible and expensive humans from dull jobs.

Based on a Norwegian keelboat design, the self-sailing Roboat won the World Robotic Sailing Championship four times.

In a naked office in the concrete jungle of Vienna’s northern outskirts, I met Roland Stelzer, a jovial, soft-spoken man who earned his Ph.D. from the University of Leicester in the U.K. with a research project that became known as Roboat, an autonomous sailboat (aka sailing drone) that won the World Robotic Sailing Championship four times. Its charm is in its look — so cute, so familiar, not the least bit intimidating. It is based on a 12-foot production keelboat designed by Norwegian Jan Herman Linge, the father of the Soling and Yngling. Linge called his keelboat the Laerling (Norwegian for “apprentice”) because it was intended for youth training. Although Stelzer’s Roboat uses the same hull, it is skippered by a computer and operated by sensors and actuators under the command of artificial intelligence and algorithms.

Campaigned from 2007 to 2012 by Stelzer and fellow computer programmers with a knack for sailing, Roboat’s development was sponsored by grants and donations, not by venture capital. Yet the Ferrari-red sailboat was the “bot” to beat in competitions that included station-keeping in a confined area, collision avoidance and fleet racing, with all boats going at it simultaneously in a regatta.

You can call what Roboat does zero-handed sailing, but, in essence, it is a circus act similar to a driverless Uber whisking you to dinner across town, or Amazon dumping packages on your doorstep from a quadcopter. There are no traffic lights, bicyclists, pedestrians or speeding drivers to deal with on the ocean, but sailing drones face other challenges. To be of value for scientific or intelligence projects, for which these drones are primarily designed, they have to sail efficiently. The rig has to be fool- and bomb-proof because a muck-up such as a snagged sheet or a twisted block would throw a monkey wrench into the works, rendering the vehicle inoperable. And because there aren’t many floating plug-in stations, these boats must run on renewable energy that powers the on-board systems and myriad sensors.

“Look, that’s the brain of the operation,” Stelzer says as he opens a garage door in a yard near his office. He pulls out a yellow waterproof Pelican case that contains a spaghetti mess of wires, processors, memory cards, switches and more. “Instead of putting all that stuff on the boat where it is difficult to access, we stuffed it into a suitcase that we could disconnect and take with us to work on it.”

Programming is a game that never ends because computers don’t know squat, except how to sort zeroes and ones. Teaching them to sail a small boat in choppy waters and shifting winds is a bit like herding cats. “You need fundamental rules, but you also have to allow for complexities created by puffs and waves,” Stelzer says. “Putting down the helm to tack and waiting for the wind sensor to signal breeze from the other side is not enough. The algorithms must be able to cope with a botched tack when the boat is in irons.”

Roboat recognizes that and turns the rudder in the opposite direction when drifting backward until the bow is through the wind and can bear off on the new tack. And then there’s navigation and tactics; those are entirely up to the on-board computer, which has to make heads or tails of the course to the next waypoint and the conditions the sensors report.

Saildrone’s autonomous vessels have covered more than 60,000 miles, collecting data and monitoring fish stocks and environmental conditions.

Roboat’s cockpit is crammed with sailing instruments, actuators, sensors, batteries, the controller case, solar panels capable of providing 285 watts of power and a fuel cell as a backup energy source. The sheet winch is powered by an electric motor and a bicycle chain — a brilliant hack that works. The aluminum mast supports Dacron sails, but isn’t that a bit retro in this day and age? “It’s the sail plan that was designed for this boat,” Stelzer says, adding that it depowers as it heels in a puff. “So we don’t use as much energy for the sail controls.” A wing mast, he says, was beyond the team’s collective expertise.

They swapped out the rudder for a prebalanced one that further reduces power consumption. Stelzer’s energy budget called for about 40 percent of Roboat’s power for the on-board computer, 32 percent for the weather sensors, 28 percent for the sail gear and 0.2 percent for operating the custom rudder. Energy efficiency, Stelzer says, is a key part of the game, and the area where significant progress has been made in the past couple of years. Computers are more powerful, and they use less juice.

So Roboat won regattas, but what did it accomplish in practical terms? “During a test in the Adriatic, my small inflatable chase boat ran out of gas, and nobody answered my calls for assistance on the VHF,” Stelzer says. “So I reprogrammed Roboat to come alongside and tow me back in.”

The boat’s biggest moment came in 2012 on the storm-tossed Baltic Sea, towing a hydrophone to locate porpoises while sailing autonomously. She did fine, Stelzer says, but when conditions worsened, the boat was reprogrammed to run for shelter. When a hefty puff disabled the sail controls, Roboat had to be towed to safety by the mother ship that followed her. It was a brave showing, but if there is a future to that concept, more development work and more capital are needed.

Meanwhile, in Alameda, California, Saildrone is developing autonomous sailing vessels at breakneck speed. The firm is led by Richard Jenkins, a Briton who holds the land speed record for sail-powered vehicles, at 126.2 mph. Saildrone’s base — a hangar at the former Naval Air Station across the bay from San Francisco — provides a stark contrast to Stelzer’s humble office in Vienna. In the large production hall, an armada of little orange trimarans are lined up, ready to be stuffed with sensors and powered by an innovative wing sail.

Saildrones, the company says, have covered more than 60,000 miles of open water, collecting oceanographic and climate data, and monitoring fish stocks and environmental conditions. One crossed the Pacific from San Francisco to Hawaii a couple of years ago, and another got halfway across the Atlantic before tangling with the remnants of a hurricane, which caused some damage and forced a detour to the Caribbean, where the boat was collected and repaired.

It’s obvious why and how Roboat and Saildrone went off on different tangents. First, there’s the money. Stelzer operated on a shoestring budget; Saildrone recently raised $14 million in a round of financing and enjoys the backing of venture capital firms and Silicon Valley high-rollers, including the Schmidt Family Foundation (as in Eric Schmidt, chairman of Alphabet, Google’s parent company). With money in your pockets and an ocean at your doorstep, you have a few advantages.

Then there’s making the move from a testing platform to a product. Jenkins is not just fiddling with drones; he intends to make a splash in traditional sailing, too. “In the Vendée Globe the systems are prehistoric in terms of power consumption,” he says. “Sailors have to carry 200 pounds of diesel and generators to support electronics. There needs to be a revolution of autopilot systems — they need to be low-friction, low-power. That’s entirely possible with technology we developed.”

Another difference between Roboat and Saildrone is a willingness to roll the dice, which is habitual in Silicon Valley, though not necessarily ingrained in your thinking if you’re brought up in a small country in central Europe. “The legal situation of autonomous sailing vehicles was not sufficiently clarified, which made it a risky bet [for investors],” Stelzer says, explaining why he ditched the game of sailing drones. Like Jenkins, he intended to market some of the technology that was developed for Roboat. However, after weighing his options, he stayed home and founded HappyLab, a company that offers high-tech tools such as 3-D printers and CNC routers to entrepreneurs who want to manufacture prototypes.

What if Stelzer had taken Roboat to Silicon Valley? What if he had found a million or five to develop this boat with the help of a billionaire and a gang of talented nerds who dream of changing the world with disruptive technology? These are moot points, of course, because Stelzer played his cards differently. He parked Roboat in a cramped garage, convinced that he’d taken it as far as he could with what he had. “I believe she has an Atlantic crossing in her,” he says wistfully, adding that he’d lend his support to the next owner if the boat were to be readied for the Microtransat Challenge, an autonomous crossing of the pond, considered the Holy Grail of robotic sailing.

Perhaps it’s a pipe dream, but it would be a lot of fun to see another chapter added to the story of the little red boat that could.

This article originally appeared in the January 2017 issue.