Artificial intelligence can design autonomous robots in record time, study reveals

Marta Reyes
2 min readOct 14, 2023

--

The future of robotics.

Artificial intelligence (AI) has achieved an astonishing achievement by designing autonomous robots in a record time of just 30 seconds. These robots, while simple, prove to be effective when moving in a straight line.

As Fagen Wasanni explains in a report, although they currently lack the ability to perform more complex tasks, this advance in robot design promises to democratize the creation of autonomous machines, giving access to powerful design tools to smaller entities and individuals.

The design process for these robots uses a version of simulated evolution called “gradient descent.” This approach allows the algorithm to measure the performance of different body designs and iteratively adjust towards the optimal solution.

The researchers tested their algorithm through computer simulations, starting with random shapes and teaching the AI to develop terrestrial locomotion. After just 10 simulations, the resulting robots were able to move at a speed of up to 0.5 body lengths per second, demonstrating the ability to develop functional leg structures.

The team then built physical examples of the top-performing robots using 3D printing, successfully replicating the movements seen in the simulations.

Beyond robotics
Although these robots are relatively simple and have limited functionality, this advance in design has implications that transcend robotics. The ability to design algorithms that can create products has the potential to solve various problems, from the development of sustainable energy solutions to the discovery of new medicines.

This breakthrough paves the way for a future where AI algorithms revolutionize design processes and accelerate progress in a wide range of fields.

Experts suggest that the next step will be to connect the gradient descent method with evolutionary bodies, which will allow the co-evolution of brains and bodies, similar to living organisms. This evolution could mark a milestone in the design of more advanced and sophisticated robots.

--

--