arXiv cs.AI· Maitree Meher, Siming He, S. Shankar Sastry, Jitendra Malik, Claire Tomlin, Jonas Frey·· 4 小时前AI 评分38
预算约束下的目标搜索:研究如何教会 VLN 模型在时间预算内探索
Racing the Clock: Teaching VLNs Budget-Conditioned Object Search
AI 导读
arXiv 论文研究视觉语言导航(VLN)模型的预算条件目标搜索任务:给定目标类别和动作预算,智能体需在预算耗尽前找到尽可能多目标实例。
正文 · 原文
Abstract:A rescue robot entering a building with thirty seconds on the clock should sprint toward the single most promising room; given ten minutes, it should instead explore a broader tour. Yet vision-language-navigation (VLN) models are typically evaluated under a fixed episode limit that the policy never observes, even though time-critical missions require behavior to adapt to the available budget. We study budget-conditioned object search: given a target category and an explicit action budget, an agent must find as many distinct target instances as possible before the budget expires. We investigate how to supervise VLNs for this task by comparing a privileged TSP expert, which plans near-optimal routes using the full map and true target locations, with a room-level Bellman expert that maximizes expected return under partial observability using high-level cross-scene semantic priors. Despite the TSP expert achieving substantially higher expert performance, its demonstrations are less effective for learning exploration: on held-out HM3D scenes, the Bellman-supervised VLN consistently outperforms the TSP-supervised VLN, with the advantage becoming more pronounced at longer budgets. At 2000 steps (500m), the Bellman-supervised VLN finds 71.4% of reachable target instances, compared with 65.5% for the TSP-supervised VLN and 44.8% for the pretrained VLN. These results highlight a distinction between optimal routing and optimal exploration: effective exploration supervision should optimize decisions under partial observability rather than imitate routes that are optimal only once the environment is fully known.
| Comments: | 12 pages, 4 figures, 3 tables |
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2611.17442 [cs.RO] |
| (or arXiv:2611.17442v1 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2611.17442 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Maitree Meher [view email]
[v1]
Thu, 12 Nov 2026 19:13:20 UTC (5,081 KB)
来源:arXiv cs.AI · arxiv.org