AI systems have taken massive strides in designing prose, artwork, gameplay, software, and proteins but have yet to master the design of complex physical machines. A team led by Northwestern University introduced an automatic optimization method that can design self-moving robots—from scratch by tracing failures in their behavior to errors or inefficiencies in particular parts of their physical structure. The researchers call this 'instant evolution'.

Because this method improves the robot in this way, it can optimize the interdependent parts of the robot much more rapidly than the current approach, in which the designer tries different robot designs in a trial-and-error fashion. This opens the way toward bespoke AI-driven design of robots for a wide range of tasks, rapidly and on demand.

Robots are notoriously difficult to design because of the complex interdependencies between their physical structure, sensory and motor layouts, and behavior. Despite this, almost every detail of every robot built to date has been manually determined by a human designer after several months or years of iterative ideation, prototyping, and testing. Inspired by evolutionary design in nature, the automated design of robots using evolutionary algorithms has been attempted for two decades. However, it remains inefficient: days of supercomputing are required to design robots in simulation that, when manufactured, exhibit the desired behavior. 

Through this research, the researchers show de novo optimization of a robot's structure to exhibit a desired behavior within seconds on a single consumer-grade computer and the manufactured robot's retention of that behavior. Unlike other gradient-based robot design methods, this algorithm does not presuppose any particular anatomical form; starting instead from a randomly generated apodous body plan, it consistently discovers legged locomotion, the most efficient known form of terrestrial movement. If combined with automated fabrication and scaled up to more challenging tasks, this advance promises near-instantaneous design, manufacture, and deployment of unique and useful machines for medical, environmental, vehicular, and space-based tasks.

Zero to walking

While the AI program can start with any prompt, the team began with a simple request to design a physical machine capable of walking on land. That's where the research input ended, and the AI tool was over.

The computer started with a block about the size of a bar of soap. It could jiggle but not walk. AI quickly iterated on the design because it had not yet achieved its goal. With each iteration, the AI assessed its design, identified flaws and whittled away at the simulated block to update its structure. Eventually, the simulated robot could eventually bounce in place, then hop forward and shuffle.

Different design

On its own, AI surprisingly came up with the same solution for walking- legs. But unlike nature's decidedly symmetrical designs, the robot has three legs, fins along its back, a flat face and is riddled with holes.

According to the researchers, it's interesting as they did not tell the AI that a robot should have legs. It rediscovered that legs are a good way to move around on land. Legged locomotion is, in fact, the most efficient form of terrestrial movement.

While the evolution of legs makes sense, the holes are a curious addition. AI punched holes throughout the robot's boy in seemingly random places. Three researchers do not yet know what the holes do. However, if they remove the holes, the robot can't walk anymore. 

When humans design robots, we tend to design them to look like familiar objects. But AI can create new possibilities and paths that humans have never considered. It could help us think and dream differently. And this might help us solve some of the most difficult problems we face.

Northwestern University