Detecting Silent Action-Interface Mismatch with Calibrated Relational World Models
cs.RO, cs.LG
Submitted: 2026-11-04
Updated: 2026-11-04
Terminology
Sources
- Unpacking Failure Modes of Generative Policies: Runtime Monitoring of Consistency and Progress
- HYDRA: Hybrid Robot Actions for Imitation Learning
- Finetuning Offline World Models in the Real World
- World Models
- FATE-VLA:Failue-aware test generation for vision-language-action models
- DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset
- LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learning
- Model-Based Runtime Monitoring with Interactive Imitation Learning
- RAPT: Model-Predictive Out-of-Distribution Detection and Failure Diagnosis for Sim-to-Real Humanoid Deployment
- Trojans in Artificial Intelligence (TrojAI) Final Report
- Same Weights, Different Robot: A Deployment Safety View of VLA Policies
- Understanding Impacts of Task Similarity on Backdoor Attack and Detection
- Metamorphic Testing of Vision-Language Action-Enabled Robots
- How to Compress KV Cache in RL Post-Training? Shadow Mask Distillation for Memory-Efficient Alignment
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