Research

Current Work

Enhance RL with LLM Feedback

Paper Reinforcement Learning LLM

Proposed and verified an RL framework which exploits LLM feedback for reward specification based on potential functions.

Action Advising for Efficient RL in Ad Hoc Teaming

Reinforcement Learning Transfer Learning Multi-Agent PyTorch

Develop policy transfer strategy based on action advising to enhance reinforcement learning efficiency in multi-agent ad hoc teaming.

---

Past Work

Active Vision of Coordinated Ground Vehicle Systems

Multi-Robot Planning and Control Active Vision ROS

Scheduled active vision tasks for a UGV team to enhance vision-based localization of an aerial vehicle for safe navigation.

Efficient Large-Scale Team Behavior Manipulation

Multi-Agent Reinforcement Learning Large Scale Distributed Control

Established a distributed behavior control pipeline using social network insights and reinforcement learning for large-scale multi-agent teams.

Adaptive Control of UAVs with Parameter Estimation

Paper Drone Adaptive Control Parameter Estimation MatLab

Proposed an adaptive control method with mass-inertia estimation and disturbance rejection tailored for multi-rotor UAVs in aerial transportation tasks.

Design of an Integrated Unmanned Quadruped-Hexarotor System

Paper Robotic System Design Aerial-Ground Robot Computer Vision

Designed a hexarotor UAV capable of grasping and transporting a quadruped robot through an adaptive docking structure.