RoboGene: Boosting VLA Pre-training via Diversity-Driven Agentic Framework for Real-World Task Generation

TL;DR

RoboGene is an agentic framework that automates the generation of diverse, feasible robotic manipulation tasks to enhance pre-training data quality, leading to improved robot learning and generalization in real-world scenarios.

Contribution

The paper introduces RoboGene, a novel framework combining diversity sampling, self-reflection, and human-in-the-loop to generate high-quality, diverse robotic tasks for pre-training.

Findings

RoboGene outperforms existing foundation models in task quality and diversity.

Pre-training with RoboGene-generated data improves robot success rates.

Large-scale real-world experiments validate the effectiveness of RoboGene.

Abstract

The pursuit of general-purpose robotic manipulation is hindered by the scarcity of diverse, real-world interaction data. Unlike data collection from web in vision or language, robotic data collection is an active process incurring prohibitive physical costs. Consequently, automated task curation to maximize data value remains a critical yet under-explored challenge. Existing manual methods are unscalable and biased toward common tasks, while off-the-shelf foundation models often hallucinate physically infeasible instructions. To address this, we introduce RoboGene, an agentic framework designed to automate the generation of diverse, physically plausible manipulation tasks across single-arm, dual-arm, and mobile robots. RoboGene integrates three core components: diversity-driven sampling for broad task coverage, self-reflection mechanisms to enforce physical constraints, and human-in-the-loop refinement for continuous improvement. We conduct extensive quantitative analysis and large-scale real-world experiments, collecting datasets of 18k trajectories and introducing novel metrics to assess task quality, feasibility, and diversity. Results demonstrate that RoboGene significantly outperforms state-of-the-art foundation models (e.g., GPT-4o, Gemini 2.5 Pro). Furthermore, real-world experiments show that VLA models pre-trained with RoboGene achieve higher success rates and superior generalization, underscoring the importance of high-quality task generation. Our project is available at https://robogene-boost-vla.github.io.