An Embodied Simulation Platform, Benchmark, and Data-Efficient Augmentation Framework for Wet-Lab Robotics

2026-06-11

AI Summary

This work addresses the limited scalability of wet-lab robotic learning, which suffers from a lack of customizable simulation environments, open experimental assets, and efficient data generation mechanisms. To overcome these challenges, the authors introduce Pipette, an integrated platform comprising a simulation environment, benchmark suite, and data augmentation framework. Pipette enables high-quality training data generation from minimal human demonstrations—only 30 per task—and supports natural language–specified task definitions. The platform releases, for the first time, 43 editable wet-lab assets and introduces multi-dimensional simulation perturbations (e.g., lighting, camera pose, speed, and action noise) alongside automated success detection for data augmentation. A comprehensive benchmark covering 11 categories of wet-lab operations is established. Experiments demonstrate that simulation-based augmentation significantly improves performance: SmolVLA’s success rate rises from 44.1% to 74.7%, π0 from 40.4% to 46.5%, and ACT achieves an average success rate of 65.5%.

Abstract

Wet-lab robots can improve the reproducibility, throughput, and safety of biomedical experiments, but scaling their learning requires customizable simulators for safe and reproducible task generation, open editable laboratory assets, and efficient pipelines that turn limited demonstrations into usable training data. We present Pipette, an embodied simulation platform, benchmark, and data-efficient augmentation framework for wet-lab robot learning. Pipette releases over 43 open-source and re-editable wet-lab assets, together with an extensible asset-building pipeline. A key component of Pipette is its simulation-based data augmentation pipeline, replaying human demonstrations in simulation, applies lighting, camera, speed, and action perturbations, and filters generated episodes with automatic task success checks, rapidly expanding usable training data from limited manual demonstrations. We further introduce an 11-task wet-lab embodied benchmark covering sample handling, culture-ware manipulation, device operation, and precision placement. With only 30 demonstrations per task, ACT achieves 65.5% average success rate, while simulation augmentation improves SmolVLA from 44.1% to 74.7% and π0 from 40.4% to 46.5%, validating the effectiveness of Pipette for data-efficient VLA training and evaluation. Pipette also supports natural-language-driven scene construction and task registration, lowering the barrier for non-expert users to define new wet-lab robotic tasks.

Problem

• wet-lab robotics
• simulation platform
• data-efficient learning
• embodied benchmark
• training data augmentation

Innovation

• embodied simulation
• data-efficient augmentation
• wet-lab robotics
• open-source lab assets
• vision-language-action (VLA) learning