Nathan Lichtlé

PhD student in Deep Learning @ UC Berkeley
Berkeley AI Research (BAIR)

I'm graduating this May and actively seeking industry roles. Feel free to reach out!

nathan.lichtle@gmail.com

PhD @ Ecole des Ponts ParisTech
MS @ ENS Paris-Saclay (MVA)

CV / Email

About me

Hi, I'm Nathan Lichtlé (pronounced "leesht-lay")! I'm a Ph.D. candidate in EECS at UC Berkeley, advised by Alex Bayen. I'm part of the Berkeley AI Research (BAIR) lab and Berkeley Deep Drive (BDD). Previously, I did my MS at Ecole Normale Supérieure Paris-Saclay (MVA Master), then received a Ph.D. from Ecole des Ponts ParisTech under the supervision of Amaury Hayat.

My main research interests are in applied deep learning and control. Specifically, I have spent a large part of my Ph.D. focusing on reinforcement learning (RL) and its real-world applications in autonomous driving and traffic optimization. This culminated with the CIRCLES experiment, in which I designed and trained RL agents to control 100 autonomous vehicles (AVs) deployed in live highway traffic as part of the largest traffic smoothing experiment ever conducted.

I have a strong interest in applying AI to control multi-agent systems and building fast, data-driven simulators for RL. Beyond that, I've worked on traffic forecasting, combining PDE-inspired representations with neural networks for sequence modeling, achieving accurate long-term predictions. Lately, I've also become increasingly interested in exploring the potential of language models for control systems.

If you're interested in chatting with me, drop me an email!

My Research

CIRCLES project

CIRCLES : 100 cars highway deployment

circles-consortium.github.io

Using deep RL, we train traffic-smoothing controllers and deploy them on 100 AVs in rush hour highway traffic to improve flow and reduce fuel consumption for everyone on the road.

Nocturne project

Nocturne : data-based traffic simulation

github.com/facebookresearch/nocturne

Nocturne is a fast 2D driving simulator built in C++, leveraging real-world Waymo data to create complex control tasks that require multi-agent coordination under human-like partial observability.

I-210 project

I-210 flow optimization

We optimize fuel economy in a large, calibrated model of Interstate 210. Using multi-agent RL, we reduce fuel consumption by 25% using 10% of AVs and without decreasing the system throughput.

Publications

Reevaluating Policy Gradient Methods for Imperfect-Information Games

Max Rudolph*, Nathan Lichtlé*, Sobhan Mohammadpour*, Alexandre M. Bayen, J. Zico Kolter, Amy Zhang, Gabriele Farina, Eugene Vinitsky, Samuel Sokota
Submitted to ICML 2025 (Forty-Second International Conference on Machine Learning).
Paper / Code / DEMO

Reinforcement Learning-Based Oscillation Dampening: Scaling Up Single-Agent Reinforcement Learning Algorithms to a 100-Autonomous Vehicle Highway Field Operational Test

Kathy Jang*, Nathan Lichtlé*, Eugene Vinitsky, Adit Shah, Matthew Bunting, Matthew Nice, Benedetto Piccoli, Benjamin Seibold, Daniel B. Work, Maria Laura Delle Monache, Jonathan Sprinkle, Jonathan W. Lee, and Alexandre M. Bayen.
IEEE Control Systems Magazine (CSM), Feb 2025, Volume 45, Issue 1.
Paper

Traffic Control via Connected and Automated Vehicles (CAVs): An Open-Road Field Experiment with 100 CAVs

Jonathan W. Lee*, Han Wang*, Kathy Jang*, Nathan Lichtlé*, Amaury Hayat*, Matthew Bunting*, Arwa Alanqary, William Barbour, Zhe Fu, Xiaoqian Gong, George Gunter, Sharon Hornstein, Abdul Rahman Kreidieh, Matthew W. Nice, William A. Richardson, Adit Shah, Eugene Vinitsky, Fangyu Wu, Shengquan Xiang, Sulaiman Almatrudi, Fahd Althukair, Rahul Bhadani, Joy Carpio, Raphael Chekroun, Eric Cheng, Maria Teresa Chiri, Fang-Chieh Chou, Ryan DeLorenzo, Marsalis Gibson, Derek Gloudemans, Anish Gollakota, Junyi Ji, Alexander Keimer, Nour Khoudari, Malaika Mahmood, Mikail Mahmood, Hossein Nick Zinat Matin, Sean McQuade, Rabie Ramadan, Daniel Urieli, Xia Wang, Yanbing Wang, Rita Xu, Mengsha Yao, Yiling You, Gergely Zachár, Yibo Zhao, Mostafa Ameli, Mirza Najamuddin Baig, Sarah Bhaskaran, Kenneth Butts, Manasi Gowda, Caroline Janssen, John Lee, Liam Pedersen, Riley Wagner, Zimo Zhang, Chang Zhou, Daniel B. Work, Benjamin Seibold, Jonathan Sprinkle, Benedetto Piccoli, Maria Laura Delle Monache, and Alexandre M. Bayen.
IEEE Control Systems Magazine (CSM), Feb 2025, Volume 45, Issue 1.
Paper

Traffic Smoothing Controllers for Autonomous Vehicles Using Deep Reinforcement Learning and Real-World Trajectory Data

Nathan Lichtlé, Kathy Jang, Adit Shah, Eugene Vinitsky, Jonathan W. Lee, and Alexandre M. Bayen
2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC).
Paper

Nocturne: a scalable driving benchmark for bringing multi-agent learning one step closer to the real world

Eugene Vinitsky*, Nathan Lichtlé*, Xiaomeng Yang*, Brandon Amos, Jakob Foerster
36th Conference on Neural Information Processing Systems (NeurIPS 2022) Track on Datasets and Benchmarks.
Paper / Supplementary material (videos) / Code (Collab with Facebook Research)

Deploying Traffic Smoothing Cruise Controllers Learned from Trajectory Data

Nathan Lichtlé*, Eugene Vinitsky*, Matthew Nice*, Benjamin Seibold, Dan Work, Alexandre M. Bayen
2022 IEEE International Conference on Robotics and Automation (ICRA)
Paper / Code

Fuel Consumption Reduction of Multi-Lane Road Networks using Decentralized Mixed-Autonomy Control

Nathan Lichtlé, Eugene Vinitsky, George Gunter, Akash Velu, Alexandre M. Bayen.
2021 IEEE Intelligent Transportation Systems Conference (ITSC)
Paper / Visualize the highway scenario

Optimizing Mixed Autonomy Traffic Flow with Decentralized Autonomous Vehicles and Multi-Agent Reinforcement Learning

Eugene Vinitsky*, Nathan Lichtlé*, Kanaad Parvate, Alexandre M. Bayen
ACM Transactions on Cyber-Physical Systems, Vol. 7, No. 2, Article 13. April 2023.
Paper

From Sim to Real: A Pipeline for Training and Deploying Traffic Smoothing Cruise Controllers

Nathan Lichtlé*, Eugene Vinitsky*, Matthew Nice*, Rahul Bhadani, Matthew Bunting, Fangyu Wu, Benedetto Piccoli, Benjamin Seibold, Daniel B. Work, Jonathan W. Lee, Jonathan Sprinkle, Alexandre M. Bayen
IEEE Transactions on Robotics (T-RO), Volume 40, September 2024.
Paper / The I-24 Trajectory Dataset

Reinforcement Learning in Control Theory: A New Approach to Mathematical Problem Solving

Kala Agbo Bidi, Jean-Michel Coron, Amaury Hayat, and Nathan Lichtlé*
37th Conference on Neural Information Processing Systems (NeurIPS 2023) Workshop on MATH-AI. Submitted to Systems & Control Letters (SCL).
Paper / Code (coming soon)

Beliefs and Level-k Reasoning in Traffic

Eugene Vinitsky, Angelos Filos, Nathan Lichtlé, Kevin Lin, Nicholas Liu, Alexandre M. Bayen, Anca Dragan, Rowan McAllister, Jakob Foerster
Talking to Strangers: Zero-Shot Emergent Communication, 4th Emergent Communication Workshop at the 34th Conference on Neural Information Processing Systems (NeurIPS 2020), Vancouver, Canada.
Paper

Inter-Level Cooperation in Hierarchical Reinforcement Learning

Abdul Rahman Kreidieh, Samyak Parajuli, Nathan Lichtlé, Yiling You, Rayyan Nasr, Alexandre M. Bayen
Preprint. 2019.
Paper

Integrated Framework of Vehicle Dynamics, Instabilities, Energy Models, and Sparse Flow Smoothing Controllers

Jonathan W. Lee, George Gunter, Rabie Ramadan, Sulaiman Almatrudi, Paige Arnold, John Aquino, William Barbour, Rahul Bhadani, Joy Carpio, Fang-Chieh Chou, Marsalis Gibson, Xiaoqian Gong, Amaury Hayat, Nour Khoudari, Abdul Rahman Kreidieh, Maya Kumar, Nathan Lichtlé, Sean McQuade, Brian Nguyen, Megan Ross, Sydney Truong, Eugene Vinitsky, Yibo Zhao, Jonathan Sprinkle, Benedetto Piccoli, Alexandre M. Bayen, Daniel B. Work, Benjamin Seibold
DI-CPS'21: Proceedings of the Workshop on Data-Driven and Intelligent Cyber-Physical Systems
Paper
Antalya, Turkey