RoboCerebra: A Large-scale Benchmark for Long-horizon Robotic Manipulation Evaluation

TL;DR

RoboCerebra introduces a comprehensive benchmark for evaluating long-horizon, high-level reasoning in robotic manipulation, combining a large simulation dataset, hierarchical planning framework, and evaluation of vision-language models' reasoning abilities.

Contribution

It presents RoboCerebra, a novel benchmark with extended tasks, hierarchical VLM-based planning, and structured evaluation protocols for robotic reasoning.

Findings

State-of-the-art VLMs show limited performance in long-horizon planning.

The benchmark reveals challenges in memory and reflection in robotic systems.

Longer, more complex tasks expose gaps in current VLM capabilities.

Abstract

Recent advances in vision-language models (VLMs) have enabled instruction-conditioned robotic systems with improved generalization. However, most existing work focuses on reactive System 1 policies, underutilizing VLMs' strengths in semantic reasoning and long-horizon planning. These System 2 capabilities-characterized by deliberative, goal-directed thinking-remain under explored due to the limited temporal scale and structural complexity of current benchmarks. To address this gap, we introduce RoboCerebra, a benchmark for evaluating high-level reasoning in long-horizon robotic manipulation. RoboCerebra includes: (1) a large-scale simulation dataset with extended task horizons and diverse subtask sequences in household environments; (2) a hierarchical framework combining a high-level VLM planner with a low-level vision-language-action (VLA) controller; and (3) an evaluation protocol targeting planning, reflection, and memory through structured System 1-System 2 interaction. The dataset is constructed via a top-down pipeline, where GPT generates task instructions and decomposes them into subtask sequences. Human operators execute the subtasks in simulation, yielding high-quality trajectories with dynamic object variations. Compared to prior benchmarks, RoboCerebra features significantly longer action sequences and denser annotations. We further benchmark state-of-the-art VLMs as System 2 modules and analyze their performance across key cognitive dimensions, advancing the development of more capable and generalizable robotic planners.