Craftium: Bridging Flexibility and Efficiency for Rich 3D Single- and Multi-Agent Environments
This provides a more efficient tool for researchers and developers working on 3D single- and multi-agent environments, though it is incremental as it builds on existing environment platforms.
The paper tackles the problem of creating rich 3D environments for autonomous agents by introducing Craftium, a customizable platform that reduces computational costs, achieving over 2,000 more steps per second compared to Minecraft-based frameworks.
Advances in large models, reinforcement learning, and open-endedness have accelerated progress toward autonomous agents that can learn and interact in the real world. To achieve this, flexible tools are needed to create rich, yet computationally efficient, environments. While scalable 2D environments fail to address key real-world challenges like 3D navigation and spatial reasoning, more complex 3D environments are computationally expensive and lack features like customizability and multi-agent support. This paper introduces Craftium, a highly customizable and easy-to-use platform for building rich 3D single- and multi-agent environments. We showcase environments of different complexity and nature: from single- and multi-agent tasks to vast worlds with many creatures and biomes, and customizable procedural task generators. Benchmarking shows that Craftium significantly reduces the computational cost of alternatives of similar richness, achieving +2K steps per second more than Minecraft-based frameworks.