The 1st Workshop @ IROS on
Embodied AI and Robotics for Future Scientific Discovery
October 24
, 2025 ยท
8:00 AM to 12:00 PM
(GMT+8) in
PressRoom A
The accelerating pace of scientific research demands more efficient, accurate, and comprehensive methodologies that can transcend the limitations of traditional research approaches. Current scientific practices face significant challenges: manual experimentation is time-consuming and resource-intensive, while multidisciplinary research requires integration of knowledge that often exceeds individual researchers' expertise. This workshop explores a transformative vision: the development of embodied AI systems capable of conducting end-to-end scientific research across multiple disciplines.
Recent advances in embodied AI and robotics present unprecedented opportunities to evolve from purely virtual AI scientists to embodied robotic researchers that can actively engage in a complete research lifecycle. This workshop brings together experts from robotics, AI, and various scientific domains to discuss the development of systems that can autonomously perform the scientific research lifecycle, literature reviews, generate hypotheses, conduct physical experiments, and synthesize findings into scholarly manuscripts.
The workshop will chart a transformative path: from traditional human-led research to collaborative AI scientists and Robot scientists working alongside humans,
and ultimately to fully autonomous AI and Robot scientists with capabilities. This developmental roadmap will help identify key technological barriers and research opportunities in creating embodied AI and Robotic systems capable of revolutionary scientific breakthroughs.
What is Embodied AI and Robotic for future scientifc discovery ?
see this table below:
Content
Topics
The workshop aims to address critical challenges and implications, including technological limitations, ethical considerations, and the changing landscape of scientific training. By fostering dialogue between researchers and practitioners, we aim to establish a roadmap for developing truly autonomous generalist scientist that can help address complex global challenges.
- Advanced AI agents and algorithms for each stage of autonomous scientific research, including:
- Operating system agents with advanced computing capabilities for scientific research automation
- Domain-specific robotic systems designed for autonomous experimentation in targeted scientific fields
- Embodied AI systems with dexterous manipulation capabilities, incorporating:
- Foundation models for perception and reasoning
- Advanced robotic hardware for precise manipulation
- Multi-modal sensing and interaction
- Real-world adaptation and learning
- Comprehensive datasets and benchmarks for:
- AI-driven scientific discovery
- Robotic manipulation in laboratory settings
- Performance evaluation frameworks for AI agent and robot-based autonomous scientific systems
- Broader impacts and ethical considerations
The workshop will also address critical challenges and implications, including technological limitations, ethical considerations, and the changing landscape of scientific training. By fostering dialogue between researchers and practitioners, we aim to establish a roadmap for developing truly autonomous generalist scientist that can help address complex global challenges.
