The Factory Must Grow: Automation in Factorio (2102.04871v1)

Abstract: Efficient optimization of resources is paramount to success in many problems faced today. In the field of operational research the efficient scheduling of employees; packing of vans; routing of vehicles; logistics of airlines and transport of materials can be the difference between emission reduction or excess, profits or losses and feasibility or unworkable solutions. The video game Factorio, by Wube Software, has a myriad of problems which are analogous to such real-world problems, and is a useful simulator for developing solutions for these problems. In this paper we define the logistic transport belt problem and define mathematical integer programming model of it. We developed an interface to allow optimizers in any programming language to interact with Factorio, and we provide an initial benchmark of logistic transport belt problems. We present results for Simulated Annealing, quick Genetic Programming and Evolutionary Reinforcement Learning, three different meta-heuristic techniques to optimize this novel problem.

• The paper introduces a novel logistic transport belt problem formulated as an integer programming model to optimize resource flow in Factorio.
• It develops the Factorio Optimizer Interface (FOI) that allows external meta-heuristic methods like SA, qGP, and ERL to interact with the game engine.
• Numerical tests show qGP excels in obstacle-free scenarios while ERL generalizes robustly across varied configurations despite computational challenges.

Automation in Factorio

The paper "The Factory Must Grow: Automation in Factorio" addresses optimization challenges within the context of the video game Factorio, exploiting its potential as an operational research simulator. This paper introduces a novel problem referred to as the logistic transport belt problem, drawing parallels to real-world logistical challenges such as those encountered in material transport, vehicle routing, and resource scheduling.

The paper details the development of a mathematical model for the logistic transport belt problem alongside the creation of an interface enabling external optimizers to interact seamlessly with Factorio. The optimization framework is tested using several meta-heuristic methods: Simulated Annealing (SA), quick Genetic Programming (qGP), and Evolutionary Reinforcement Learning (ERL).

Core Contributions

1. Definition and Model Formulation: The authors formulate the logistic transport belt problem as a mathematical integer programming problem. The model captures constraints and objectives relevant to the optimization of transport belts in Factorio, which are critical for maximizing outputs and ensuring efficient resource allocation.
2. Factorio Optimizer Interface (FOI): An interface, compatible with multiple programming languages, was developed to facilitate communication between external optimizers and the Factorio game engine. The FOI translates problem instances into their corresponding Lua commands for execution in the game, receiving feedback in real-time during the optimization process.
3. Meta-Heuristic Techniques Application:
   • Simulated Annealing (SA): A classical optimization method modified for parallel execution, enabling simultaneous solution evaluation and adaptation based on stochastic sampling.

• quick Genetic Programming (qGP): A newly presented variant of Genetic Programming tailored for the logistic belt problem, leveraging domain-specific operators and operands to optimize the belt layout efficiently.
• Evolutionary Reinforcement Learning (ERL): Integrates Genetic Programming with reinforcement learning paradigms to evolve generalist agents capable of solving multiple instances of the belt problem by direct interaction and adaptation.

Numerical Results and Observations

The performance of the three optimization strategies across varying grid sizes (3x3, 6x6, and 12x12) and obstacle configurations was analyzed. qGP demonstrated a superior capability in handling problem instances without obstacles, as it utilized specialized knowledge embedded in the operators. Meanwhile, ERL showed significant robustness, particularly in generalizing solutions across different problem configurations; however, it faced challenges with computational efficiency on larger grid sizes. The parallel SA method was efficient but limited in scaling to higher complexities inherent in larger grid layouts.

Implications and Future Directions

The implications of this research extend beyond the scope of video games, offering insights into the automated resolution of complex logistical challenges using meta-heuristic approaches. The interaction between optimization algorithms and simulation environments such as Factorio suggests pathways for solving real-world logistics and supply chain problems more effectively, enhancing adaptability and resilience.

Future developments might focus on complexifying the logistic belt problem by integrating additional in-game variables such as splitter dynamics and throughput constraints. Furthermore, exploring hybridization strategies or co-evolution techniques may unveil new optimization frontiers, enhancing the diversity and efficiency of solutions obtained through automated game interaction models.

This paper thus provides a foundational exploration into the dynamics of optimization problems within virtual environments, illustrating their utility as benchmarks and testbeds for developing cutting-edge computational techniques applicable to real-world operational challenges.