Endong is a PhD student under the supervision of Dr Matthew Howard. He received his MSc degree in Applied Machine Learning at Imperial College London and his BEng in EEE at Swansea University in 2022 and 2021, respectively. He joined the Robot Learning Lab at King’s College London and received NMES PhD Studentships in June 2023 where he is conducting his PhD research on imitation learning and reinforcement learning in Robot control.
- Imitation learning and robot learning from demonstration.
- (Inverse) Reinforcement learning
- The applications of machine learning in dynamic robot control
Thesis title and abstract
Train the novice to give a better reward in Reinforcement Learning from Demonstration.
Learning from demonstrations is a good robotics learning method as it does not require specialized knowledge such as control theory to program the robot automation. It also improves interactive experiences and gains better societal acceptance, making AI and Robotics more trustworthy. But the demonstration quality is crucial for student learning. Our research is to train the novice to give a better reward in reinforcement learning from demonstration so that the training data are more accurate for the robot agent to learn. The methodology is to propose a feedback method to improve the reward behaviour of the novice. Then the novice will be like an expert to teach the robot any skill.
First supervisor: Dr Matthew Howard
Second supervisor: Dr Hak-Keung Lam