A control system based on chaos has made a simulated, multi-legged robot walk successfully.

In chaotic systems, small effects are amplified so rapidly that the systems’ behavior becomes impossible to predict more than a short time ahead.

The robot was able to learn to walk and negotiate obstacles without any conventional programming. And its behavior emerged far more quickly than it would if it had used genetic algorithms.

The researchers suggest that the chaotic approach may have similarities to the way biological systems learn to move.