Bayesian Active Learning for Intent Disambiguation in Interactive Robot Planning
Bayesian Active Learning for Intent Disambiguation in Interactive Robot Planning
This research introduces a Bayesian framework to address ambiguity in natural language instructions for interactive robot planning. The method utilizes large language models (LLMs) to generate candidate formal specifications and translate informative contrasts into clarification questions, while employing Bayesian optimization to estimate uncertainty over user intent and select queries that maximize information gain. After convergence, the inferred Signal Temporal Logic (STL) specification is passed to a formal planner to synthesize a verifiable robot trajectory. The approach was evaluated across four simulated and real-world task domains, demonstrating higher task satisfaction and fewer clarification rounds compared to LLM baselines, while also helping smaller models close the performance gap against larger reasoning models.
The paper proposes a Bayesian active learning framework for intent disambiguation in interactive robot planning, using LLMs to initialize formal specifications and Bayesian optimization to select clarification queries that maximize information gain over grounded Signal Temporal Logic task specifications.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
- code: not_reported
- weights: not_reported
- dataset: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
ain. After convergence, the inferred STL specification is passed to a formal planner to synthesize a verifiable robot trajectory. Across four simulated and real-world task domains, our approach generally achieves higher task satisfaction and requires fewer clarification rounds than LLM baselines, while helping smaller models close the performance gap against larger reasoning models.
Open source S4
Bayesian Active Learning for Intent Disambiguation in Interactive Robot Planning
Open source S1
arXiv:2609.34270v2 Announce Type: replace Abstract: Interactive robot planning requires robots to infer and execute human intentions from natural language instructions that are often ambiguous, incomplete, or underspecified. Although large language models (LLMs) provide a powerful interface for clarification, relying on the generative model to drive an multi-turn conversation can introduce systematic failures. We pro
Open source S2
pose a Bayesian framework that treats clarification as an active learning problem over grounded Signal Temporal Logic (STL) task specifications. Our method uses LLMs to initialize candidate formal specifications and translate informative contrasts into natural-language clarification questions, while Bayesian optimization maintains uncertainty estimation over user intent and selects queries that maximize information g
Open source S3
Source：arXiv Robotics — research abstracts · arxiv.org