Embodied AI and Robotics for Autonomous Scientific Discovery

Autonomous Scientific Discovery faces bottlenecks from manual experimentation and interdisciplinary barriers. While AI excels at in silico analysis, physical testing remains a challenge. Embodied AI, which integrates perception, reasoning, and action in real-world environments, offers a transformative path toward autonomous robotic scientists.

This Research Topic aims to advance embodied AI systems capable of performing complete scientific research cycles—from hypothesis generation to physical experimentation and analysis. It seeks to address key technological challenges in autonomous decision-making, precision manipulation, and human-AI collaboration to accelerate discovery across diverse scientific domains.

We welcome original research, reviews, and perspectives on:
• AI-driven autonomous experimentation
• Foundation models for scientific reasoning
• Robotic manipulation for lab procedures
• Human-AI-robot collaboration frameworks
• Closed-loop experimentation systems
• Benchmarks for robotic scientists
• Ethical and societal implications

This Research Topic is linked to the partnered workshop of the same name at IROS 2025 in Hangzhou, China. Authors who previously published their paper as a conference proceeding must expand their paper to include 30% original content to be considered.