Search papers, labs, and topics across Lattice.
The authors introduce World2Minecraft, a system for converting real-world scenes into structured Minecraft environments using 3D semantic occupancy prediction, enabling flexible embodied AI research. They address the challenge of inaccurate occupancy prediction by creating MinecraftOcc, a large-scale dataset of 100K+ images from 156 indoor scenes, using a novel low-cost data acquisition pipeline. Experiments demonstrate that MinecraftOcc significantly improves occupancy prediction and provides a challenging benchmark for state-of-the-art methods.
Escape the limitations of existing embodied AI simulation environments with World2Minecraft, a system that converts real-world scenes into customizable Minecraft worlds.
Embodied intelligence requires high-fidelity simulation environments to support perception and decision-making, yet existing platforms often suffer from data contamination and limited flexibility. To mitigate this, we propose World2Minecraft to convert real-world scenes into structured Minecraft environments based on 3D semantic occupancy prediction. In the reconstructed scenes, we can effortlessly perform downstream tasks such as Vision-Language Navigation(VLN). However, we observe that reconstruction quality heavily depends on accurate occupancy prediction, which remains limited by data scarcity and poor generalization in existing models. We introduce a low-cost, automated, and scalable data acquisition pipeline for creating customized occupancy datasets, and demonstrate its effectiveness through MinecraftOcc, a large-scale dataset featuring 100,165 images from 156 richly detailed indoor scenes. Extensive experiments show that our dataset provides a critical complement to existing datasets and poses a significant challenge to current SOTA methods. These findings contribute to improving occupancy prediction and highlight the value of World2Minecraft in providing a customizable and editable platform for personalized embodied AI research. Project page:https://world2minecraft.github.io/.
