Papers
6
Total Citations
79
H-Index
4
About
Mengdi Li is an emerging researcher at the intersection of robotics, artificial intelligence, and human-robot interaction, with a particular focus on leveraging large language models (LLMs) to advance autonomous robotic systems. Her work addresses some of the most pressing challenges in modern robotics, including high-level planning, multimodal perception, and complex manipulation tasks. Li's most influential contribution, "Chat with the Environment: Interactive Multimodal Perception Using Large Language Models" (2023), has garnered over 51 citations and demonstrates how LLMs can enable robots to reason and plan within complex, real-world environments. Building on this foundation, her 2024 work on bimanual robot orchestration tackles the notoriously difficult problem of coordinating two-handed robotic manipulation through intelligent language-driven control policies. Her research also extends to explainability in reinforcement learning, where she investigates reward decomposition to make agent behavior more interpretable to humans — a crucial step toward trustworthy AI systems. Earlier contributions on occlusion reasoning and visually grounded human-robot dialogue reveal a consistent thread throughout Li's research: making robots more perceptive, communicative, and understandable. With a growing citation record and timely research agenda, she represents a valuable voice in shaping the future of intelligent, language-enabled robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2Large Language Models for Orchestrating Bimanual Robots9 citations · 2024
- 3
- 4Robotic Occlusion Reasoning for Efficient Object Existence Prediction5 citations · 2021
- 5
- 6