# From Failures to Learnable Worlds: Policy-Conditioned Environment Generation for Navigation Agents

arXiv: https://arxiv.org/abs/2610.19420 · HTML: https://mlsanity.com/paper/2610.19420/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Reinforcement learning for LLM agents, one of 16 directions in LLM agents and tool use.
- Builds on 51 of its 54 references that are in the map.
- First posted on arXiv 14 Oct 2026.

## Placed in

- LLM agents and tool use / [Reinforcement learning for LLM agents](https://mlsanity.com/agents/reinforcement-learning-for-llm-agents/)

## Builds on (most cited first, 30 of 51)

- [Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities](https://mlsanity.com/paper/2507.06261/) (2025)
- [Matterport3D: Learning from RGB-D Data in Indoor Environments](https://mlsanity.com/paper/1709.06158/) (2017)
- [Habitat: A Platform for Embodied AI Research](https://mlsanity.com/paper/1904.01201/) (2019)
- [Habitat-Matterport 3D Dataset (HM3D): 1000 Large-scale 3D Environments for Embodied AI](https://mlsanity.com/paper/2109.08238/) (2021)
- [ProcTHOR: Large-Scale Embodied AI Using Procedural Generation](https://mlsanity.com/paper/2206.06994/) (2022)
- [DD-PPO: Learning Near-Perfect PointGoal Navigators from 2.5 Billion Frames](https://mlsanity.com/paper/1911.00357/) (2019)
- [LayoutGPT: Compositional Visual Planning and Generation with Large Language Models](https://mlsanity.com/paper/2305.15393/) (2023)
- [Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design](https://mlsanity.com/paper/2012.02096/) (2020)
- [Holodeck: Language Guided Generation of 3D Embodied AI Environments](https://mlsanity.com/paper/2312.09067/) (2023)
- [RoboGen: Towards Unleashing Infinite Data for Automated Robot Learning via Generative Simulation](https://mlsanity.com/paper/2311.01455/) (2023)
- [ATISS: Autoregressive Transformers for Indoor Scene Synthesis](https://mlsanity.com/paper/2110.03675/) (2021)
- [Prioritized Level Replay](https://mlsanity.com/paper/2010.03934/) (2020)
- [RoboTHOR: An Open Simulation-to-Real Embodied AI Platform](https://mlsanity.com/paper/2004.06799/) (2020)
- [Evolving Curricula with Regret-Based Environment Design](https://mlsanity.com/paper/2203.01302/) (2022)
- [Habitat Synthetic Scenes Dataset (HSSD-200): An Analysis of 3D Scene Scale and Realism Tradeoffs for ObjectGoal Navigation](https://mlsanity.com/paper/2306.11290/) (2023)
- [DiffuScene: Denoising Diffusion Models for Generative Indoor Scene Synthesis](https://mlsanity.com/paper/2303.14207/) (2023)
- [Uni-NaVid: A Video-based Vision-Language-Action Model for Unifying Embodied Navigation Tasks](https://mlsanity.com/paper/2412.06224/) (2024)
- [iGibson 1.0: a Simulation Environment for Interactive Tasks in Large Realistic Scenes](https://mlsanity.com/paper/2012.02924/) (2020)
- [Infinigen Indoors: Photorealistic Indoor Scenes using Procedural Generation](https://mlsanity.com/paper/2406.11824/) (2024)
- [SceneCraft: An LLM Agent for Synthesizing 3D Scene as Blender Code](https://mlsanity.com/paper/2403.01248/) (2024)
- [PhyScene: Physically Interactable 3D Scene Synthesis for Embodied AI](https://mlsanity.com/paper/2404.09465/) (2024)
- [Scaling Data Generation in Vision-and-Language Navigation](https://mlsanity.com/paper/2307.15644/) (2023)
- [Enhanced POET: Open-Ended Reinforcement Learning through Unbounded Invention of Learning Challenges and their Solutions](https://mlsanity.com/paper/2003.08536/) (2020)
- [LayoutVLM: Differentiable Optimization of 3D Layout via Vision-Language Models](https://mlsanity.com/paper/2412.02193/) (2024)
- [AnyHome: Open-Vocabulary Generation of Structured and Textured 3D Homes](https://mlsanity.com/paper/2312.06644/) (2023)
- [Offline Visual Representation Learning for Embodied Navigation](https://mlsanity.com/paper/2204.13226/) (2022)
- [OMNI-EPIC: Open-endedness via Models of human Notions of Interestingness with Environments Programmed in Code](https://mlsanity.com/paper/2405.15568/) (2024)
- [PoliFormer: Scaling On-Policy RL with Transformers Results in Masterful Navigators](https://mlsanity.com/paper/2406.20083/) (2024)
- [SceneWeaver: All-in-One 3D Scene Synthesis with an Extensible and Self-Reflective Agent](https://mlsanity.com/paper/2509.20414/) (2025)
- [Auxiliary Tasks Speed Up Learning PointGoal Navigation](https://mlsanity.com/paper/2007.04561/) (2020)

## Abstract (arXiv)

Simulated environments enable scalable training data generation for embodied agents, but more data does not necessarily provide what a particular agent needs to learn. Existing environment-generation pipelines largely optimize generic properties such as realism, diversity, or task coverage, while different pretrained policies can fail for different environmental reasons. We therefore introduce WEDGE, a framework for weakness-guided navigation learning in which the failure modes of a pretrained agent guide the generation of training environments at each adaptation round. Given a pretrained policy, WEDGE uses an LLM to propose executable weakness features from contrasting successful and failed rollouts, automatically validates them on the diagnostic set, and selects the strongest validated weakness based on its success-rate gap. The validated weaknesses are then mapped to the environmental factors that control them. WEDGE varies these factors to construct environments in the difficulty regimes where the policy struggles, producing targeted training experience for adaptation. Across three navigation agents with different adaptation mechanisms, we find that policies exhibit distinct weakness profiles and benefit from different training environments. Weakness-conditioned adaptation consistently outperforms matched random and hard-example alternatives on ProcTHOR, while the resulting gains for Uni-NaVid and Gemini also extend to HM3D after ProcTHOR adaptation. Iterative re-diagnosis further shows that dominant weaknesses can persist or shift as the policy improves, providing updated guidance for subsequent environment construction. Together, these results show that policy failures are not only diagnostic signals, but actionable guidance for constructing training environments that target a policy's current weaknesses.