This is the Most Aerodynamic Bike, According to AI

By George Dvorsky on at

Competitors at this year’s World Human Powered Speed Challenge are going to have to contend with this — a bullet-shaped bike designed by an artificially intelligent software program.

In 2012, a bicycle screamed across a flat, open road of the Nevada Desert at an astounding 88.13 miles per hour. This record, established by a Dutch team at the annual World Human Powered Speed Challenge, could now be in danger, owing to a new bike designed by researchers at IUT Annecy, with the help of computer scientists at Neural Concept, a Subsidiary of the Swiss Federal Institute of Technology in Lausanne (EPFL).

A prototype of the bike undergoing tests.

To be fair, the IUT Annecy researchers can’t take full credit for the bike’s sleek, aerodynamic shape, nor can any human for that matter. You see, this machine was, in part, designed by another machine — an artificially intelligent program developed by researchers at Neural Concept, who are presenting their findings today in Stockholm, Sweden, at the International Conference on Machine Learning.

The AI utilises a 3D mesh to play around with different configurations.

This AI has a neural net capable of evaluating the aerodynamic properties of various shapes, which are internally represented by polygon meshes, or points that are used to create 3D shapes. The system starts with the most basic shapes and works toward increasing complexity until it finally settles on an optimal, super-aerodynamic shape.

“Our program is based on a learning machine system that learns to develop an intuition about the laws of physics,” said Pierre Baqué, CEO of Neural Concept, in an EPFL News video. “It can then optimise the shape to get the best possible performance.”

To get the AI started on a project, the researchers feed the system with the basics about the bike, like the maximum length and width required to squeeze in the cyclist, the drivetrain, and its four wheels. Armed with these constraints, the AI sorts through an assembly of possible shapes, looking for the best configuration, including, for example, the best location for the bike’s maximum width.

The system understands basic physics and the rudiments of aerodynamics.

The new bicycle scarcely resembles a conventional racing bike. Its bullet-like shape was custom-made to fit a cyclist’s reclined body. “The design objective clearly isn’t cyclist comfort, but making the most out of every inch of the vehicle,” according to an EPFL release. During the next World Human Powered Speed Challenge, scheduled for September 10 to 15, the bike will gradually accelerate over the course of a five-mile lead run, and then have its maximum speed measured along a 650 foot long stretch.

What’s neat about this approach is that it’s free from human biases and preconceived notions of what an aerodynamic vehicle is supposed to look like.

“Our program results in designs that are sometimes 5–20% more aerodynamic than conventional methods,” said Baqué. “The shapes used in training the program can be very different from the standard shapes for a given object. That gives it a great deal of flexibility.” Baqué says the program could also be used to design other machines, such as drones, wind turbines, and aircraft.

The new AI-designed bike looks cool, and it may very well break the new speed record, but we won’t know for sure until the rubber hits the road in September. We’ll be watching. [EPFL]


More Artificial intelligence Posts: