RoboGaze：用结构化视觉-语言分析评估机器人世界模型

Authors:Minh-Loi Nguyen, Nghiem T. Diep, Hung K. Nguyen, Minh Le, Doanh L. Thien, Tuan A. Tran, Hoang H. Tran, Dung D. Le, Vu N. Duong, Khoa D. Doan, Ngan Le, Daniel Sonntag, An T. Le, Vien Anh Ngo, Duy M. H. Nguyen, Tran V. Nhiem

View PDF HTML (experimental)

Abstract:Robot world models now generate synthetic manipulation videos, but evaluating them is hard: realistic-looking outputs often violate physical laws, temporal consistency or task logic, while conventional metrics and monolithic Vision-Language Model (VLM) judges return scores that do not say what failed. We present RoboGaze, a training-free multi-agent VLM evaluator that grounds the task and scene, routes each subtask to dimension-specific specialists and verifies their findings with a critic, producing temporally localized diagnostic reports that state what failed, when, why and how severely, under a schema of 6 dimensions and 30 failure types whose annotation reliability and coverage we measure. To benchmark it, we introduce RoboGazeBench, 914 human-annotated clips from five generators, conditioned on simulated and real-world data and running from 4.9 to 148.9 s. Across eight open-source and proprietary VLM backbones, RoboGaze outperforms every baseline; even a published agentic detector given the same schema and a slightly larger inference budget trails it by 10.1-12.3 description-F1 points, a margin that blinded human raters confirm. On the strongest backbone it closes 83% of the gap between chain-of-thought prompting and human agreement, and its critic curbs the "cry-wolf" false positives of standard VLMs, lifting mean clean-clip accuracy from 17% to 85%. The advantage is largest on held-out generators and long clips, and carries over to selecting generated demonstrations (80% vs. 71% precision).

Submission history

From: Minh-Loi Nguyen [view email]
[v1] Mon, 22 Jun 2026 06:45:09 UTC (37,091 KB)
[v2] Sat, 10 Oct 2026 08:18:36 UTC (11,036 KB)