Towards Next-Generation Scientific Discovery: AI-Assisted Autonomous Experimentation

Autonomous laboratories and AI-driven experimentation are increasingly transforming how scientific discovery is conducted in chemistry, materials science, biology, and related domains. While existing efforts have achieved notable progress in either algorithmic components—such as Bayesian optimization and active learning—or engineered automation pipelines that reliably execute predefined protocols, true experimental autonomy remains limited. Achieving scalable autonomy requires tightly integrating scientific objectives, experimental design, robotic execution, and data-driven adaptation within a unified closed-loop framework. Recent advances in foundation models, multimodal learning, planning, and robotics offer new opportunities to build general-purpose experimental agents capable of reasoning over goals, operating diverse instruments, and learning from outcomes. However, challenges in system integration, uncertainty handling, safety, standardization, and evaluation persist, motivating focused research in this area.Recent advances in artificial intelligence, robotics, and laboratory automation are reshaping how scientific experiments are conceived, conducted, and iteratively improved. Beyond workflow automation and isolated optimization loops, the field is increasingly moving toward agentic systems that can autonomously plan experiments, execute protocols, observe outcomes, and adapt strategies with minimal human intervention. The goal of this Research Topic is to consolidate emerging methods and integrated systems that enable such autonomous scientific experimentation, spanning both physical laboratory robotics and software-based scientific agents.We aim to bring together interdisciplinary contributions across robotics, machine learning, and experimental sciences, covering the full experimental stack—from high-level scientific reasoning and experimental design to reliable execution on instruments and robots, as well as data interpretation and decision-making under uncertainty. This Research Topic seeks to highlight advances that support closed-loop discovery, improve robustness and reproducibility, and enhance the scalability and accessibility of autonomous experimentation across diverse scientific domains and laboratory infrastructures.This Research Topic welcomes original research articles, methods papers, system papers, and reviews on AI-assisted and agentic approaches to scientific experimentation. Topics of interest include, but are not limited to, closed-loop experimental design, scientific agents for experiment planning, execution, and monitoring, and the integration of AI with laboratory automation platforms, instruments, and robotic systems. We also encourage work on multimodal sensing, data representation and management, uncertainty-aware decision-making, safety and human-in-the-loop supervision, as well as standardization, benchmarking, and evaluation of autonomous experimental systems. Contributions may target chemistry, materials science, biology, or other experimental sciences, and may involve both simulated and real-world laboratory settings. Topic editors: Yan Xia, Yanyong Zhang, Wenjun Wu, Ning Han Source: https://www.frontiersin.org/research-topics/77878/towards-next-generation-scientific-discovery-ai-assisted-autonomous-experimentation