America Cup: an AI to test hydrofoil designs faster than sailors

Team New Zealand won the prestigious America’s Cup against Italians Luna Rossa. The competition which saw flying monohulls face off is always an opportunity to experiment with countless technological innovations and for this 36th edition artificial intelligence was naturally part of the game.

As the competition rules allow for certain modifications to the design of hydrofoils, the ability to simulate and test them quickly is a major advantage, explains McKinsey, who worked with the New Zealand team. Their idea: to use artificial intelligence to test designs at high speed, so sailors can train on the water and spend less time in the simulator.

Accelerated learning

Specialists first taught intelligent bots to navigate like seasoned sailors by combining deep learning and reinforcement learning techniques (deep reinforcement learning). “Initially, the AI ​​agent doesn’t know anything. It learns by trial and error using countless variables – wind speed, direction, adjustments to 14 different sail and boat controls – and it gets fine tuned over and over again, ”says Nic Hohn, Data Scientist at QuantumBlack, the McKinsey cell specializing in analytics.

The specialists also created a network allowing the thousands of bots learning at the same time to exchange information. Once the basic navigation techniques have been learned, AI agents begin to learn more sophisticated maneuvers.

Better designs, better sailors

After which the bots trained in virtual navigation having reached the performance of the sailors in the simulator, they were deployed to test the variations of hydrofoils, with the classic advantages of algorithms: speed and scale.

Besides the accelerated exploration of multiple designs, the employment of intelligent agents has also proved useful for sailors, who have been able to improve their own practice for a given design. “It’s by being able to compare what you’re doing with the bot that you see performance gains,” explains Peter Burling, coxswain for Team New Zealand.

Swiss industry is also starting to use AI for innovation and product development:

Syngenta will use deep learning to develop phytosanitary products

Firmenich creates a grilled beef aroma with an AI aftertaste

–