What Can a VLA Learn from the Agent That Calls It? Task-Level Experience Distillation for Embodied Navigation
cs.RO, cs.LG
Submitted: 2026-11-12
Updated: 2026-11-17
Terminology
Sources
- On Evaluation of Embodied Navigation Agents
- Understanding the Benefits of Scaffolds through Computational Separations
- Beyond On-Policy vs. Off-Policy: At What Level Should LLM Agent Distillation Follow the Student?
- TinyAstral: Transferring Self-Evolving Harnesses to Tiny Generalist Manipulation Agents
- Nipping the Drift in the Bud: Retrospective Rectification for Robust Vision-Language Navigation
- Beyond Policy Patches: Distilling RL Specialists into Vision-Language-Action Policies
- X-Fly: Learning to Reach and Stop in Full-Chunk Aerial Vision-Language Navigation
- Scaffolding Foundation Models into Physical-World Agents Pushes the Frontier of Long-Horizon Navigation
- Learning Goal-Oriented Vision-and-Language Navigation with Self-Improving Demonstrations at Scale
- Guava: Distilling Frontier VLMs into a Compact Agent through a Robotic Manipulation Harness
- Recova: Agent-Guided Failure Recovery for Autonomous Robotic Manipulation
- StopWatch: Goal Grounding for Object-goal Navigation
- TILDE: Teacher-in-the-Loop Distillation from Embodied Foundation Models
- Asymmetric On-Policy Distillation Learns In-Context Exploration
- From Region Arrival to Instance-Level Grounding in Vision-and-Language Navigation
- Externalise the Memory: A Pre-Registered Study of Where an Embodied Agent's Experience Should Live
- AdaRSI: On-Demand Recursive Self-Improvement for Embodied Agents
- OmniBrain-0: A Unified Agentic Model for Embodied Intelligence
- Harness-Zero: Harness Distillation via Agent-as-Harness
- MATCH: Execution-Aware Hierarchical Reinforcement Learning for Vision-and-Language Navigation in Continuous Environment
Related papers
- Rethinking Actions: Continuity-Aware Priors for World-Model Planning
- MOSAIC: Modular Obstacle and Spatial-Aware Framework for Manipulation In Cluttered Scenes
- SAGAS: Semantic-Aware Graph-Assisted Stitching for Offline Temporal Logic Planning
- DexSeed: Expanding Human Hand-Object Interactions into Robot Demonstrations for Dexterous Policy Learning
- LaST0.5: A Generalist World Action Model with Interaction-Centric Latent Reasoning
- CIALLO: Learning Shared Intent Representations for Cross-Embodiment Manipulation