Robo2VLM: Visual Question Answering from Large-Scale In-the-Wild Robot Manipulation Datasets

TL;DR

Robo2VLM introduces a large-scale VQA dataset derived from real robot manipulation data, enabling the evaluation and enhancement of vision-language models in understanding complex robotic scenes and tasks.

Contribution

The paper presents Robo2VLM, a novel framework for generating VQA datasets from robot trajectories, bridging robotic manipulation data with vision-language model benchmarking.

Findings

Robo2VLM-1 contains 684,710 questions across 463 scenes.

The dataset covers 3,396 manipulation tasks from 176k trajectories.

Results show improved VLM benchmarking for spatial and interaction reasoning.

Abstract

Vision-Language Models (VLMs) acquire real-world knowledge and general reasoning ability through Internet-scale image-text corpora. They can augment robotic systems with scene understanding and task planning, and assist visuomotor policies that are trained on robot trajectory data. We explore the reverse paradigm - using rich, real, multi-modal robot trajectory data to enhance and evaluate VLMs. In this paper, we present Robo2VLM, a Visual Question Answering (VQA) dataset generation framework for VLMs. Given a human tele-operated robot trajectory, Robo2VLM derives ground-truth from non-visual and non-descriptive sensory modalities, such as end-effector pose, gripper aperture, and force sensing. Based on these modalities, it segments the robot trajectory into a sequence of manipulation phases. At each phase, Robo2VLM uses scene and interaction understanding to identify 3D properties of the robot, task goal, and the target object. The properties are used to generate representative VQA queries - images with textural multiple-choice questions - based on spatial, goal-conditioned, and interaction reasoning question templates. We curate Robo2VLM-1, a large-scale in-the-wild dataset with 684,710 questions covering 463 distinct scenes and 3,396 robotic manipulation tasks from 176k real robot trajectories. Results suggest that Robo2VLM-1 can benchmark and improve VLM capabilities in spatial and interaction reasoning.