Research Project

Wavefront Sensorless Adaptive Optics using Reinforcement Learning

Core researcher
Advisor: Dr. Davide Spinello, Dr. Ross Cheriton, Dr. Colin Bellinger
August 2021 – Now
Keywords: Adaptive Optics, Reinforcement Learning, Soft Actor-Critic

1. Developing a budget-friendly wavefront sensorless adaptive optics system.
2. Online model-free off-policy reinforcement learning framework, soft actor-critic controller.
3. Tuning hyperparameters of soft actor-critic algorithm
4. Simulations in both static and dynamic atmospheres
   

Communications and Observations through Atmospheric Turbulence (COAT 2023) presentation

Research on Coverage Control of Multi-agent System

Core researcher
Advisor: Dr. Davide Spinello
March 2021 – August 2021
Keywords: Multi-agent Control, Multi-agent Reinforcement Learning

1. Reviewed method applied on multi-agent system control and modeling of unknown environment
2. Proposed a multi-robot control method based on multi-agent reinforcement learning (MARL) and Gaussian process regression (GPR).

Research on Energy Management Strategy of Hybrid Power Source

Core researcher
Advisor: Dr. Chenxing Hu
June 2019 – August 2019
Keywords: Hybrid Power Source, Energy Management,Reinforcement Learning

1. Proposed an accelerated energy management strategy NAF-DQL+PR based on deep Q learning and realized it by TensorFlow
2. Verified the proposed strategy by software simulation and hardware-in- the-loop simulation

Research on Optimization of hyper parameter in Deep Q Learning

Core researcher
Advisor: Dr. Yuan Zou
March 2020 – April 2020

1. Optimized the hyper parameter of NAF-DQL+PR by genetic algorithm and realized it by python
2. Verified the proposed strategy by software simulation

Monitoring and Regulation Interface Development of Hybrid Power Source

Core researcher
Advisor: Dr. Yuan Zou
September 2019 – October 2019

1. Established vehicle CAN protocol and improved system architecture
2. Achieved real-time monitoring and adjustment interface of the power source based on LabVIEW
3. Achieved pressure test of the interface by hardware-in- the-loop simulation

   Part of the interface	Part of the code

Development of a Double Mass Vibration Didactical Experimental Platform

Core researcher
Advisor: Dr. Yuan Zou
June 2018 – December 2018

1. Completed parameter matching and selection of electrical components according to the syllabus
2. Established vehicle CAN protocol of platform
3. Realize real-time parameter adjustment and operation status display of the platform based on LabVIEW
4. Accomplished verification of the platform by based on the requirements of the course