Constrained Decoding for Safe Robot Navigation Foundation Models
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866914477977370624 |
|---|---|
| author | Kapoor, Parv Ganlath, Akila Clifford, Michael Liu, Changliu Scherer, Sebastian Kang, Eunsuk |
| author_facet | Kapoor, Parv Ganlath, Akila Clifford, Michael Liu, Changliu Scherer, Sebastian Kang, Eunsuk |
| contents | Recent advances in the development of robotic foundation models have led to promising end-to-end and general-purpose capabilities in robotic systems. Trained on vast datasets of simulated and real-world trajectories, these policies map multimodal observations directly to action sequences for physical execution. Despite promising real-world capabilities, these models are still data-driven and, therefore, lack explicit notions of behavioral correctness. We address this gap by introducing SafeDec, a constrained decoding framework for autoregressive, transformer-based robot navigation foundation models that enforces safety specifications expressed as Signal Temporal Logic (STL) formulas. Our method ensures that generated actions provably satisfy STL specifications under assumed dynamics at runtime without retraining while remaining agnostic of the underlying policy. We evaluate SafeDec on tasks from the CHORES benchmark for state-of-the-art embodied navigation policies across hundreds of procedurally generated environments and show that our decoding-time interventions are useful not only for filtering unsafe actions but also for conditional action generation. Videos are available at constrained-robot-fms.github.io |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_01728 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Constrained Decoding for Safe Robot Navigation Foundation Models Kapoor, Parv Ganlath, Akila Clifford, Michael Liu, Changliu Scherer, Sebastian Kang, Eunsuk Robotics Machine Learning Logic in Computer Science Recent advances in the development of robotic foundation models have led to promising end-to-end and general-purpose capabilities in robotic systems. Trained on vast datasets of simulated and real-world trajectories, these policies map multimodal observations directly to action sequences for physical execution. Despite promising real-world capabilities, these models are still data-driven and, therefore, lack explicit notions of behavioral correctness. We address this gap by introducing SafeDec, a constrained decoding framework for autoregressive, transformer-based robot navigation foundation models that enforces safety specifications expressed as Signal Temporal Logic (STL) formulas. Our method ensures that generated actions provably satisfy STL specifications under assumed dynamics at runtime without retraining while remaining agnostic of the underlying policy. We evaluate SafeDec on tasks from the CHORES benchmark for state-of-the-art embodied navigation policies across hundreds of procedurally generated environments and show that our decoding-time interventions are useful not only for filtering unsafe actions but also for conditional action generation. Videos are available at constrained-robot-fms.github.io |
| title | Constrained Decoding for Safe Robot Navigation Foundation Models |
| topic | Robotics Machine Learning Logic in Computer Science |
| url | https://arxiv.org/abs/2509.01728 |