Research Overview
Investigating the integration of Large Language Models (LLMs) into multi-agent robotic systems to enhance coordination, planning, and adaptability in swarm robotics scenarios.
Research Focus
Multi-Agent Coordination
Enhanced multi-agent coordination by integrating LLMs into existing Hierarchical Reinforcement Learning (HRL) policies, enabling:
- Natural language-based task specification
- Improved inter-agent communication
- Adaptive behavior in dynamic environments
Safe AI Lab
Conducted research under Prof. Ding Zhao at Carnegie Mellon’s Safe AI Lab, focusing on:
- AI safety in multi-agent systems
- Verification and validation of learned behaviors
- Robust coordination under uncertainty
Technical Approach
- Hierarchical Reinforcement Learning (HRL) for task decomposition
- LLM-based planning and coordination
- Multi-agent simulation environments
- Safety-critical decision making
Impact
Exploring how language models can improve the interpretability and flexibility of multi-agent systems while maintaining safety guarantees—a critical challenge for deploying autonomous systems in real-world scenarios.