Multi-task scheduling of self-driving laboratories under scientific constraints

Area of Science:

• Robotics and Automation
• Artificial Intelligence in Science
• Materials Science

Background:

• Self-driving laboratories (SDLs) automate scientific experiments using robotics and AI.
• Concurrent multi-task execution in SDLs faces challenges with scheduling, leading to resource conflicts and data quality issues.
• Accurate scheduling is crucial for maintaining experimental fidelity, especially for time-sensitive processes like nucleation.

Purpose of the Study:

• To develop and validate a multi-task scheduling algorithm for SDLs that accounts for scientific constraints.
• To improve the coordination between scheduling and robotic execution in SDLs for concurrent experiments.
• To ensure high-fidelity data generation for AI-driven scientific discovery.

Main Methods:

• Developed a multi-task scheduling algorithm integrating scientific constraints like precedence, station allocation, and time synchronization.
• Integrated the algorithm into an SDL with a closed-loop communication architecture.
• Validated the algorithm through concurrent synthesis of gold nanoparticles and metal-organic frameworks.

Main Results:

• The algorithm successfully synchronized robotic operations with experimental stations and chemical workflows.
• Consistent material quality was achieved across concurrent multi-task executions, outperforming conventional scheduling.
• The approach preserved chemical fidelity essential for reproducible scientific data.

Conclusions:

• The developed scheduling algorithm provides a practical framework for concurrent multitasking in SDLs.
• This framework ensures experimental consistency, crucial for generating high-fidelity datasets for autonomous discovery.
• The study advances the capability of SDLs for reliable, AI-driven scientific research.