Research map / Multi-agent systems and games / Human-AI and goal-conditioned
UnrealZoo: Enriching Photo-realistic Virtual Worlds for Embodied AI
Fangwei Zhong, Kui Wu, Churan Wang, Hao Chen, Hai Ci, Zhoujun Li, Yizhou Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| parallel rendering, batch commands and IPC sockets (UnrealCV+) | UnrealCV · object mask capture | FPS: 70 -> 154 | Improved |
| UnrealCV+ optimizations | UnrealCV · color image capture | FPS: 74 -> 83 | Improved |
| UnrealCV+ batch commands | UnrealCV · 10-agent interaction | FPS: 8 -> 16 | Improved |
| train on 8 environments instead of 1 (ablation) | offline RL tracker · 16 unseen environments, Wilds | success rate: 0.42 -> 0.70 | Improved |
| train on 8 environments instead of 1 (ablation) | offline RL tracker · TerrainDemo | success rate: 0.56 -> 0.56 | No clear effect |
| increase training diversity from 1 to 8 environments (ablation) | offline RL tracker · ModularGothicNight | success rate: 0.52 -> 0.76 | Improved |
| control frequency from 3 FPS to 30 FPS (ablation) | RL agent 1 Env. · FlexibleRoom | success rate: 0.34 -> 0.92 | Improved |
| add distractors from 0D to 10D (ablation) | RL 8 Envs. · social tracking | success rate: 0.76 -> 0.56 | Hurt |
| add distractors from 0D to 10D (ablation) | OpenVLA · social tracking | success rate: 0.10 -> 0.00 | Hurt |
| transfer from human to robot dog body | RL 8 Envs. · 4D distractors | success rate: 0.68 -> 0.48 | Hurt |
| transfer from human to robot dog body | RL 8 Envs. · 10D distractors | success rate: 0.56 -> 0.40 | Hurt |
| transfer from human to robot dog body | PID · 0D distractors | success rate: 0.34 -> 0.30 | Hurt |
Closest papers by meaning
- VirtualEnv: A Platform for Embodied AI Research (AAAI 2026 · similarity 0.90)
- Learning Interactive Real-World Simulators (ICLR 2024 · similarity 0.89)
- Habitat Synthetic Scenes Dataset (HSSD-200): An Analysis of 3D Scene Scale and Realism Tradeoffs for ObjectGoal Navigation (CVPR 2024 · similarity 0.89)
- Habitat 3.0: A Co-Habitat for Humans, Avatars, and Robots (ICLR 2024 · similarity 0.88)
- Synthesizing Photorealistic and Dynamic Urban Environments for Multimodal Robot Navigation and Collaboration (NeurIPS 2025 · similarity 0.88)
- SimWorld Studio: Automatic Environment Generation with Evolving Coding Agent for Embodied Agent Learning (NeurIPS 2026 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
