Portrait of Keyu Chen

Ph.D. Candidate · Tsinghua University

Keyu Chen (陈可禹)

Traffic Simulation · RL Post-Training · Autonomous Driving

I am a third-year Ph.D. student in the School of Vehicle and Mobility at Tsinghua University, advised by Prof. Sifa Zheng.

My research focuses on traffic simulation and RL post-training for autonomous driving, with an emphasis on realistic, interactive, and controllable traffic scenarios. I develop reliable closed-loop simulation frameworks to support the safe, robust, and scalable advancement of autonomous driving systems.

Beijing, China 中文简历 LinkedIn ORCID ResearchGate
01

Academic background

Education

Tsinghua University logo

Tsinghua University

Ph.D. candidate in Mechanical Engineering, School of Vehicle and Mobility

Beijing, China
Nanjing University of Aeronautics and Astronautics logo

Nanjing University of Aeronautics and Astronautics

B.Eng. in Vehicle Engineering

Nanjing, China
02

Recent updates

News

ForSim was accepted to ICRA 2026.

FREA was accepted to CoRL 2024 as an Oral presentation (4.3%).

IGT was accepted to Knowledge-Based Systems.

03

Research output

Publications

Selected Publications

Representative work
CRAFT framework overview

CRAFT: Counterfactual-to-Interactive Reinforcement Fine-Tuning for Driving Policies

Keyu Chen, Nanfei Ye, Yida Wang, Wenchao Sun, Danqi Zhao, Hao Cheng, Sifa Zheng

arXiv, 2026

CRAFT improves driving policies by combining dense counterfactual proxy supervision with residual correction from true closed-loop interaction.

Paper Project Code Stars
ForSim stepwise forward simulation overview

ForSim: Stepwise Forward Simulation for Traffic Policy Fine-Tuning

Keyu Chen, Wenchao Sun, Hao Cheng, Zheng Fu, Sifa Zheng

IEEE International Conference on Robotics and Automation (ICRA), 2026

ForSim achieves high-fidelity traffic simulation through stepwise closed-loop trajectory propagation, combining physical dynamics for multimodal diversity with interaction-aware agent updates for realistic behavioral consistency.

Paper Project Code Stars
RIFT realistic and controllable traffic simulation

RIFT: Group-Relative RL Fine-Tuning for Realistic and Controllable Traffic Simulation

Keyu Chen, Wenchao Sun, Hao Cheng, Sifa Zheng

arXiv, 2025

RIFT achieves realistic and controllable traffic simulation by combining imitation-learning pre-training in a data-driven simulator with reinforcement-learning fine-tuning in a physics-based simulator.

Paper Project Code Stars
FREA feasibility-guided safety-critical scenario generation

FREA: Feasibility-Guided Generation of Safety-Critical Scenarios with Reasonable Adversariality

Keyu Chen, Yuheng Lei, Hao Cheng, Haoran Wu, Wenchao Sun, Sifa Zheng

Conference on Robot Learning (CoRL), 2024 · Oral presentation (4.3%)

FREA incorporates feasibility as guidance to generate adversarial yet AV-feasible, safety-critical scenarios for autonomous driving.

Paper Project Code Stars
IGT illumination-guided RGB-T object detection

IGT: Illumination-Guided RGB-T Object Detection with Transformers

Keyu Chen, Jinqiang Liu, Han Zhang

Knowledge-Based Systems (KBS), 2023 · SCI Q1, IF 7.4

IGT uses illumination intensity to guide multimodal feature fusion, enabling the comprehensive use of complementary cross-modal information in object detection.

Paper

All Publications

Additional collaborative work
SparseDriveV2 scoring-based end-to-end autonomous driving

SparseDriveV2: Scoring is All You Need for End-to-End Autonomous Driving

Wenchao Sun, Xuewu Lin, Keyu Chen, Zixiang Pei, Xiang Li, Yining Shi, Sifa Zheng

European Conference on Computer Vision (ECCV), 2026

SparseDriveV2 advances scoring-based autonomous driving with a scalable factorized trajectory vocabulary and coarse-to-fine scoring for dense yet efficient planning.

Paper Code Stars
DriveCamSim camera simulation overview

DriveCamSim: Generalizable Camera Simulation via Explicit Camera Modeling for Autonomous Driving

Wenchao Sun, Xuewu Lin, Keyu Chen, Zixiang Pei, Yining Shi, Chuang Zhang, Sifa Zheng

arXiv, 2025

DriveCamSim introduces explicit camera modeling to generate controllable multi-view, multi-frame driving imagery that generalizes across camera parameters and frame rates.

Paper Code Stars
Emergency Index traffic conflict assessment

Emergency Index (EI): A Two-Dimensional Surrogate Safety Measure Considering Vehicles’ Interaction Depth

Hao Cheng, Yanbo Jiang, Hailun Zhang, Keyu Chen, Heye Huang, Shaobing Xu, Jianqiang Wang, Sifa Zheng

Transportation Research Part C: Emerging Technologies (TRC), 2025

EI provides a two-dimensional surrogate safety measure that models interaction depth and evasive urgency for more faithful traffic-conflict assessment.

Paper Code Stars
04

Open source

Projects

Highway-Adversarial safety-critical scenarios

Highway-Adversarial

We manipulate critical background vehicles in Highway-Env to exhibit adversarial behaviors that challenge the ego vehicle, creating safety-critical scenarios.

Code Stars

05

Industry and research

Internships

Li Auto logo

Li Auto

Closed-loop RL post-training for autonomous driving

Beijing, China

06

Recognition

Honors & Awards

Outstanding Graduate Top 1%

Nanjing University of Aeronautics and Astronautics

Outstanding Bachelor Thesis Award Top 5%

Nanjing University of Aeronautics and Astronautics

National Scholarship Top 1%

Nanjing University of Aeronautics and Astronautics