Biomaker CA: a Biome Maker project using Cellular Automata

Abstract: We introduce Biomaker CA: a Biome Maker project using Cellular Automata (CA). In Biomaker CA, morphogenesis is a first class citizen and small seeds need to grow into plant-like organisms to survive in a nutrient starved environment and eventually reproduce with variation so that a biome survives for long timelines. We simulate complex biomes by means of CA rules in 2D grids and parallelize all of its computation on GPUs through the Python JAX framework. We show how this project allows for several different kinds of environments and laws of 'physics', alongside different model architectures and mutation strategies. We further analyze some configurations to show how plant agents can grow, survive, reproduce, and evolve, forming stable and unstable biomes. We then demonstrate how one can meta-evolve models to survive in a harsh environment either through end-to-end meta-evolution or by a more surgical and efficient approach, called Petri dish meta-evolution. Finally, we show how to perform interactive evolution, where the user decides how to evolve a plant model interactively and then deploys it in a larger environment. We open source Biomaker CA at: https://tinyurl.com/2x8yu34s .

Overview of "Biomaker CA: a Biome Maker project using Cellular Automata"

This paper presents Biomaker CA, a framework leveraging Cellular Automata (CA) for simulating dynamic and evolving biomes in virtual environments. A notable emphasis of Biomaker CA is on morphogenesis, where plant-like organisms grow from initial seeds and evolve within nutrient-scarce, two-dimensional environments governed by CA rules. Computations are efficiently managed using JAX on GPUs, allowing for parallel processing and large-scale experimentation.

Framework Design and Capabilities

Biomaker CA provides a broad platform for exploring areas such as artificial life, complexification, and open-ended evolution. The framework's design allows for substantial customization, offering varied environmental settings and mutable laws of 'physics' that dictate the interactions of diverse model architectures within. This adaptability makes Biomaker CA a potential tool for studying complex ecological and evolvable systems.

Environment and Agent Dynamics

Biomaker CA's environments are grid-based, built around the concept of falling-sand games where different cell types (agents, air, earth, etc.) follow specific CA rules. The agents, represented as plant cells, need to execute survival strategies by absorbing nutrients, growing, and eventually reproducing. The system design is inherently survival-based, posing a challenge for organisms due to the dynamic and competitive nature of resources and the inherent need for both air and earth nutrients.

Methodology and Experiments

The authors outline extensive experimentation possibilities, highlighting configurations such as 'persistence', 'collaboration', and 'pestilence' to demonstrate the adaptability of Biomaker CA. The environments are differentiated by the distribution of resources and the rate of organism aging or nutrient dissipation, each posing unique evolutionary challenges.

Notably, Biomaker CA supports multiple evolutionary strategies:
- In-environment Evolution: This occurs naturally within the simulation as organisms replicate and mutate over time.
- End-to-End Meta-evolution: Employs evolutionary algorithms to optimize organism parameters over multiple simulation runs to prevent extinction or promote specific adaptations.
- Petri Dish Meta-evolution: Involves evolving agents within isolated, controlled environments to ensure fitness before reintroducing them into larger simulations.
- Interactive Evolution: Allows human operators to guide evolution based on their assessment of the organisms' exhibited traits and behaviors.

Results and Analysis

The paper presents numerical evaluations illustrating how different setups and evolutionary strategies influence organism success rates and overall biome stability. For instance, end-to-end meta-evolution showed substantial improvements in resisting extinction under 'pestilence' conditions—highlighting the efficacy of systematic parameter optimization.

Implications and Future Work

The framework's open-endedness serves as both an asset and a challenge. While it offers extensive experimentation possibilities, it also emphasizes the complexity of achieving stable, evolving ecosystems. The implications of this work extend to artificial intelligence, particularly in understanding how evolving systems can adapt and grow in complexity.

Future research could focus on refining evolutionary algorithms, enhancing organism interactions, and addressing limitations related to agent mobility and higher-dimensional CA frameworks. Biomaker CA provides a versatile foundation for exploring ecological dynamics and evolution in simplified models, potentially informing broader understandings of life and complex systems.

In summary, "Biomaker CA" presents an innovative approach to simulating artificial biomes with CA, offering significant insights into evolving ecosystems and complex systems modeling. Its flexible design encourages further exploration into the mechanisms of evolution and complexification, fostering advancements in artificial intelligence and related fields.