The Microcontroller Mariners: UBC’s Sailbot Team Takes a New Tack

On a calm August morning off the coast of Newfoundland, members of the University of British Columbia’s Sailbot team said goodbye to “Ada” (see Fig.1). Ada was a custom-built sailing vessel designed to become the first fully autonomous, robotic sailboat to cross the Atlantic Ocean. The team had spent years building Ada and testing her subsystems. They spent 10 days driving across the continent and a month doing shakedown cruises from their temporary home in St. John’s. The navigation algorithms were working flawlessly, the infrared object avoidance routines had been checked, the solar panels were peaking and the batteries were charged. The last line was cast off, and Ada was on her own.

Figure 1. The University of British Columbia’s robotic sailboat, “Ada,” seen here during testing off Newfoundland. Ada would go on to set the record for greatest distance achieved in a Trans-Atlantic ocean crossing by a fully autonomous sailboat. (Image courtesy of the UBC Sailbot team.)

The initial results were outstanding. Ada’s communication system reported her progress to the crew in Newfoundland and the team at home base in Vancouver. In her first day at sea, she cleared 180km and successfully detected and avoided other ships in the area. By her second day, she had sailed 300km, setting a new record for autonomous Atlantic crossings.

As she crossed into international waters, she caught a stiff breeze, setting her surfing atop rollers and exceeding her calculated hull speed. After three days at sea, she was 20 percent of the way to her finish line off the coast of Ireland. On the fourth day, however, 800km from shore and cruising at over 12 knots, she made an unexpected turn to the south (see Fig.2).

No one knows exactly what happened, but Ada had lost control. Analysis of the new path and sensor data suggested a problem with the rudder, and by studying their SOLIDWORKS files and photographs, the student team identified a potential weak spot in a rudder control linkage. It was one of the few critical components on the boat that did not have redundancy. Without rudder control, Ada was now at the mercy of the sea, unable to avoid storms or set her bow into the waves.

From September through December, as the rough Atlantic winter rolled in, Ada was pummelled by five major storms. She drifted towards Portugal and looped back around the Azores, covering almost 8,000km. As she neared the Azores, there was hope that she may be rescued, but in late November, the data transmissions stopped. In December, the Sailbot team announced that Ada had been lost at sea.

Figure 2. After rapidly covering 800 km, covering 20 percent of the distance to Ireland, Ada likely suffered a rudder failure, leaving her at the mercy of the winds and waves. She continued to travel almost 8,000 km over the next three months before being lost at sea south of the Azores. (Image courtesy of the UBC Sailbot team.)

Team co-captains Serena Ramley and Oliver Terry took the news in stride. Ada had set a new distance record and taught the team much about their design. The software worked well, their electronics stayed dry and functional for months and the hull took everything the North Atlantic could throw at her. One of the keys to that success was Ada’s stability.

“Ada sailed for four months through five storms and always returned to vertical,” said Ramley. “We got the keel right(see Fig.3).” She explained how keel design is a trade-off:The mass of the lead bulb at the bottom of the keel is necessary to right the boat, but too much weight in the keel will increase the boat’s displacement, slowing it down. To minimize weight, but still achieve a good righting moment, the bulb can be placed at the end of a long keel, but that increases bending moment in the keel and the stress where the keel attaches to the hull.

Figure 3. Ada was tested extensively in her home waters on the Pacific coast before attempting a Trans-Atlantic crossing. This photo shows Madie Melcer, mechanical team co-lead, hoisting Ada and the long, red keel that kept her upright through four months of North Atlantic storms. “We got the keel right … Satellite transmissions require a clear view of the sky, so the boat needs to be sufficiently upright,” said team co-captain Serena Ramley. (Image courtesy of the UBC Sailbot team.)

“Ada’s keel bulb weighed 72 kg,” Ramley explained, “about half the weight of the boat.” Connecting the bulb to the hull was a custom-designed wood and carbon fiber keel that had been designed and simulated in SOLIDWORKS (see Fig.4).

Figure 4. The UBC Sailbot team used SOLIDWORKS simulation tools to design their keel. Team members Michael Schnetzler, Greg Wong and Adrian Granchelli measured Young’s modulus for a sample of the raw material and then created simulations to determine the final design. “We have to get it built right the first time … There is no option for mid-course repairs,” said team co-captain Serena Ramley. (Image courtesy of the UBC Sailbot team.)

“We got a material sample of the keel and took a rectangular cross-section to find Young’s modulus by measuring the deflection from a cantilever beam setup,” Ramley said.“We then used that for the model. We had to get it built right the first time due to time and budget constraints, and the software helps us achieve that. There is no option for mid course repairs (see Fig. 5).”

Figure 5. Alumni co-captain KristofferVik Hansen checking Ada’s keel deflection in a cantilever beam setup after construction. (Image courtesy of the UBC Sailbot team.)

Three-time champions of the International Robotic Sailing Regatta, the team has been designing robotic sailboats in SOLIDWORKS for almost a decade. “One of the things we like about SOLIDWORKS is that it is so easy to learn,” said Ramley, referring to the support available for beginning users. “We used to have a team tradition in which senior members would teach SOLIDWORKS to the junior members on a CAD orientation day. Now UBC teaches SOLIDWORKS to all engineering students during the first year. It’s really good.”

Ramley added that some of the initial hull design work is done by mechanical team co-lead Oscar Janzen in a hydrostatic-specific package called Orca3D, but is then immediately ported to SOLIDWORKS for the detail work. “Sub-teams work on components individually and then join them to the hull. We know what the boat will look like because we are working on it every day, but we want to share our vision with the people who are following our project. SOLIDWORKS helps with that (see Fig. 6).”

Figure 6. An early 3D render of the boat made in SOLIDWORKS by Mechanical Team Co-Lead Alex von Schulmann. (Image courtesy of the UBC Sailbot team.)

While Ada may be lost at sea, the team is already hard at work on their next challenge. This time they will be sailing closer to home as they design a sailbot to compete in the 2018 Vic-Maui yacht race, a biennial classic that challenges some of the finest racing yachts on the Pacific coast.

“We’re going with a dual-rudder design,” Ramley said, referring to their analysis of Ada’s journey. This will be just one of the new vessel’s upgrades. “Ada had a windsurfing rig, because we wanted to keep it simple and they can berated to handle storms, but the Vic Maui will feature more upwind sailing, so we are going with a sloop rig… a main sail and jib… with furling so it won’t get knocked about.”

The new boat will also be a bit “beamier” (wider) and able to carry more solar panels. Electrical team co-lead Melika Salehi also made use of CAD modeling to help plan the deck and hull hardware arrangements. This will allow the team to accommodate heavier and higher power electronics where needed. Ada’s Hemisphere GPS and LCJ Capteurs Ultrasonic Wind Sensor will be making a comeback inAda 2.0 because they had already provided very high performance with low power consumption. While Ada used thermal infrared to track obstacles, the team is now experimenting with light detecting and ranging (LIDAR) technology.

“LIDAR is a key area of research. It is heavy and uses a lot of power, but unlike infrared, you can more easily determine the distance to a target. Our software team lead, Arek Sredzki, hopes to make similar breakthroughs to what we previously did with infrared,” Salehi explained.

They are also upgrading their processing system from a Raspberry Pi to an NVIDIA Jetson TX1 and their onboard sensor and control network to the CAN architecture used in automobiles. “It’s a very robust communication protocol—it can transmit messages over the entire length and height of the boat without getting corrupted. CAN-bus architecture isn’t typically taught in undergraduate engineering programs, but it’s currently being spearheaded by our control team lead, Alan Fisher, and it’s a great learning opportunity for everyone on the team,” she said.

UBC Sailbot team co-captain Serena Ramley (center), with members of the Sailbot electrical sub-team. The team’s next venture will be to compete in the 2018 Vic-Maui yacht race. (Image courtesy of the UBC Sailbot team.)

It has been 20 years since a computer defeated Gary Kasparov at chess. It has been five years since a computer won at Jeopardy! Self-driving cars are proving to be safer and more efficient than human drivers in many real-world trials. Will it be long before robotic sailors rule the waves? The potential is huge… from transportation solutions to research vessels, and from intelligent autopilots to assistance for recreational cruisers. The UBC Sailbot team isn’t predicting victory in the 2018 Vic-Maui yacht race, but the old salts had best beware that there is some fresh silicon in the race, and the sailbots will just keep getting better.

To learn more about SOLIDWORKS education programs, follow this link. If you are a researcher looking for access to SOLIDWORKS, click here.

SOLIDWORKS has sponsored this post. It has provided no editorial input. For more information, go to www.solidworks.com.