自动化
机械臂
计算机科学
人工智能
模拟
机器学习
人机交互
软件工程
工程类
机械工程
作者
Stefan Conrad,Philipp Auth,Tom Masselter,Thomas Speck
标识
DOI:10.1002/aisy.202401086
摘要
Laboratory automation is crucial for improving efficiency and enhancing reproducibility in scientific workflows. However, industrial solutions mostly do not fit the needs of scientific institutions, such as cost efficiency, customizability, and flexibility in fast iteration cycles. This study presents a laboratory automation system that integrates affordable robotics and artificial intelligence (AI)‐driven functionalities in a modular architecture to address key challenges in research environments. The system uses a robotic arm and a large language model (LLM) as a lab assistant, enabling natural language interaction and task orchestration. In contrast to fully autonomous systems, this approach emphasizes a collaborative human‐in‐the‐loop model, ensuring adaptability and reducing reliance on artificial intelligence for task planning. Key innovations include meta‐tools for dynamic task recording and playback, low‐level information management to reduce cognitive load on the LLM, and AI‐assisted data reading for real‐time measurement extraction. The system's ability to automate complex workflows is validated in three experimental scenarios, involving sample preparation, error handling, and multi‐step measurements. The system demonstrates the ability to perform tasks with minimal user input while maintaining flexibility and adaptability to changing experimental conditions. The findings pave the way for the future of laboratory automation, where human and AI‐driven systems work seamlessly together in optimized scientific workflows.
科研通智能强力驱动
Strongly Powered by AbleSci AI