Nathan L
- Research Program Mentor
PhD candidate at University of California Berkeley (UC Berkeley)
Expertise
Robotics; Reinforcement learning; Artificial intelligence; Linear Algebra
Bio
Learner - that is how I describe myself first. I love new experiences, both physical and mental. It is important to challenge how we view the world and make change. I work in robot learning, love to cook, and train for triathlons. I am always trying to optimize my physical and mental health. Formally, I am a PhD candidate at the University of California, Berkeley, Department of Electrical Engineering and Computer Sciences (graduation Fall 2021). I have the pleasure of being advised by Professor Kristofer Pister. I also do part time work at Facebook AI Research. I completed my undergraduate education at Cornell University's College of Engineering in 2017.Project ideas
Train a neural network to predict robot dynamics
With data from real robots, see what data science is needed when training a deep neural network to accurately predict forward dynamics in robotics. This will entail understanding what data looks like from a real robot (used in my research), and how that differs from simulation. You will learn about dynamics and free-body motion, how to model that with machine learning tools, and what that means for real-world autonomous agents.
Fundamentals of Artificial Intelligence in Python:
This course will distill what I learned teaching Intro to Artificial Intelligence at UC Berkeley (CS188) into functional, small coding problems. I want you to understand what decisions engineers will have to make when designing intelligent systems today. We will work with simple methods like search, constraint satisfaction problems, and linear regression.
Learn to fly a quadrotor
Use existing implementations of reinforcement learning algorithms to understand a specific robotics task - in simulation for now. I will start with lessons on robotics, then have a student implement key lines from state-of-the-art machine learning algorithms to see what does and does not work. This is labelled as advanced because reinforcement learning builds on many topics in machine learning and python, so you should be comfortable in python and have an idea what machine learning is before starting.