Research map / Multi-agent systems and games / Human-AI and goal-conditioned
MoDA: Map Style Transfer for Self-Supervised Domain Adaptation of Embodied Agents
Eun Sun Lee, Junho Kim, SangWon Park, Young Min Kim
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 |
|---|---|---|---|
| add flip consistency loss to the pretrained agent (ablation) | NA · Gibson speckle noise + dynamics (generalization), ablation | PointNav success: 0.12 -> 0.20 | Improved |
| add ego style transfer loss on top of flip consistency (ablation) | MoDA · Gibson speckle noise + dynamics (generalization), ablation | ego map MSE: 1.12 -> 1.08 | Improved |
| add temporal consistency loss (ablation) | MoDA · Gibson speckle noise + dynamics (generalization), ablation | PointNav success: 0.20 -> 0.42 | Improved |
| add temporal consistency loss (ablation) | MoDA · Gibson speckle noise + dynamics (generalization), ablation | rotation error (deg): 14.98 -> 3.16 | Improved |
| add global style transfer loss (ablation) | MoDA · Gibson speckle noise + dynamics (generalization), ablation | PointNav success: 0.42 -> 0.56 | Improved |
| add global style transfer loss (ablation) | MoDA · Gibson speckle noise + dynamics (generalization), ablation | rotation error (deg): 3.16 -> 2.61 | Improved |
| apply MoDA adaptation | NA · Gibson speckle noise + dynamics (generalization) | PointNav success: 0.12 -> 0.56 | Improved |
| apply MoDA instead of GMC on Matterport3D | GMC · Matterport3D large scale + dynamics (generalization) | PointNav success: 0.10 -> 0.22 | Improved |
| apply MoDA instead of GMC in specialization | GMC · Gibson speckle noise + dynamics (specialization) | PointNav success: 0.44 -> 0.56 | Improved |
| adapt with PAD (rotation prediction + dynamics) | NA · Gibson speckle noise + dynamics (generalization) | PointNav success: 0.12 -> 0.08 | Hurt |
| adapt with PAD | NA · Gibson speckle noise + dynamics (generalization) | rotation error (deg): 15.02 -> 1.08 | Improved |
| apply domain randomization with ground-truth supervision | NA · Gibson low-lighting + dynamics (generalization) | PointNav success: 0.22 -> 0.20 | No clear effect |
Closest papers by meaning
- C-NAV: Towards Self-Evolving Continual Object Navigation in Open World (NeurIPS 2025 · similarity 0.84)
- Imagine Before Go: Self-Supervised Generative Map for Object Goal Navigation (CVPR 2024 · similarity 0.84)
- World Model Implanting for Test-time Adaptation of Embodied Agents (ICML 2025 · similarity 0.84)
- CityWalker: Learning Embodied Urban Navigation from Web-Scale Videos (CVPR 2025 · similarity 0.84)
- Learning with a Mole: Transferable latent spatial representations for navigation without reconstruction (ICLR 2024 · similarity 0.84)
- Reasoning in Visual Navigation of End-to-end Trained Agents: A Dynamical Systems Approach (CVPR 2025 · similarity 0.83)
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.
