Towards self-organized control: Using neural cellular automata to robustly control a cart-pole agent

TL;DR

This paper demonstrates that neural cellular automata can be trained to robustly control a cart-pole system, exhibiting life-like behaviors such as regeneration, stability, and robustness through deep-Q learning.

Contribution

It introduces a novel approach using neural CA for control tasks, showing emergent stable behaviors and resilience in a dynamic environment.

Findings

Neural CA maintained computing abilities over hundreds of thousands of iterations.

The system exhibited developmental phases and regeneration after damage.

Demonstrated robustness to noise and input disruptions.

Abstract

Neural cellular automata (Neural CA) are a recent framework used to model biological phenomena emerging from multicellular organisms. In these systems, artificial neural networks are used as update rules for cellular automata. Neural CA are end-to-end differentiable systems where the parameters of the neural network can be learned to achieve a particular task. In this work, we used neural CA to control a cart-pole agent. The observations of the environment are transmitted in input cells, while the values of output cells are used as a readout of the system. We trained the model using deep-Q learning, where the states of the output cells were used as the Q-value estimates to be optimized. We found that the computing abilities of the cellular automata were maintained over several hundreds of thousands of iterations, producing an emergent stable behavior in the environment it controls for thousands of steps. Moreover, the system demonstrated life-like phenomena such as a developmental phase, regeneration after damage, stability despite a noisy environment, and robustness to unseen disruption such as input deletion.