AI Paper Atlas

Research map / Vision-language and multimodal / Robot action and navigation

Robobench: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models as Embodied Brain

Yulin Luo, Chun-Kai Fan, Menghang Dong, Jiayu Shi, Xiangju Mi, Mengdi Zhao, Bo-Wen Zhang, Cheng Chi, Jiaming Liu, Gaole Dai, Rongyu Zhang, Ruichuan An and 10 more

ECCV 2026 · Vision-language and multimodal › Robot action and navigation · approach: Embodied multimodal LLM agents · 30 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
remove image input (text-only GPT-5.4) (ablation)GPT-5.4 · RoboBench Perception Reasoningavg accuracy: 56.20 -> 27.12Hurt
remove image input (text-only GPT-5.4) (ablation)GPT-5.4 · RoboBench Affordance Predictionavg accuracy: 46.43 -> 25.61Hurt
implicit instead of explicit instructionClaude-Sonnet-4.6 · RoboBench Instruction Comprehensionaccuracy: 79.94 -> 61.38Hurt
add chain-of-thought intent scaffold to implicit instructions (ablation)GPT-5 · Implicit plan itemsworld-simulator composite: 43.6 -> 45.2No clear effect
add chain-of-thought intent scaffold to implicit instructions (ablation)Claude-Opus-4.7 · Implicit plan itemsworld-simulator composite: 34.2 -> 34.2No clear effect
add chain-of-thought intent scaffold to implicit instructions (ablation)Gemini-3.1-Pro · Implicit plan itemsworld-simulator composite: 39.5 -> 36.4Hurt
add chain-of-thought scaffold, step verification (ablation)GPT-5 · Implicit step-verification itemsaccuracy: 53.8 -> 59.0Improved
add chain-of-thought scaffold, step verification (ablation)Claude-Opus-4.7 · Implicit step-verification itemsaccuracy: 64.1 -> 59.0Hurt
add chain-of-thought scaffold, step verification (ablation)Gemini-3.1-Pro · Implicit step-verification itemsaccuracy: 56.4 -> 61.5Improved
use DAG-rollout prompt v3 vs basic judge v1 (ablation)MLLM-as-world-simulator · Q1 judge consistency (200 items)node ICC(2,k): 0.935 -> 0.948Improved
MLLM-as-world-simulator vs baseline LLM scoringevaluation framework · human alignmentPearson r: 0.73 -> 0.83Improved
multi-view input when front view occludedMLLMs · Cross-View Planningplanning scoreImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy