Mengxi Li
Papers
6
Total Citations
60
H-Index
5
About
Mengxi Li is a robotics researcher whose work lies at the intersection of human-robot interaction, robot learning, and multi-agent coordination. Her major contributions include developing methods for robots to learn human objectives from sequences of physical corrections—a paradigm that allows humans to intuitively guide robots through touch, making human-robot collaboration more natural and efficient. She has also pioneered frameworks for learning primitive-based skills from demonstrations for complex tasks like insertion, achieving data-efficient generalization. In the realm of team dynamics, Li introduced mathematical models to capture leading and following behaviors in human-robot teams, and developed decentralized approaches for robot teams to implicitly communicate through actions. Her work on learning from demonstrations under different dynamics addresses the critical challenge of transferring skills between agents with mismatched capabilities. With over 60 citations across her most-cited papers, Li’s research is shaping how robots understand human intent, adapt to human corrections, and coordinate in teams—advancing the vision of robots that can seamlessly integrate into collaborative environments.
Research Focus
Key Achievements
Top Papers
- 1Learning Human Objectives from Sequences of Physical Corrections23 citations · 2021
- 2
- 3Influencing Leading and Following in Human-Robot Teams11 citations · 2019
- 4Learning from My Partner's Actions: Roles in Decentralized Robot Teams8 citations · 2019
- 5Learning Feasibility to Imitate Demonstrators with Different Dynamics5 citations · 2021
- 6Learning Human Objectives from Sequences of Physical Corrections2 citations · 2021