I am a final-year Ph.D. candidate at the Safe AI Lab, advised by Prof. Ding Zhao. I also work closely with Google Deepmind, Prof. Bo Li and Prof. Huan Zhang at UIUC, Prof. Steven Wu at CMU. Before that, I finished my bachelor's degree with honor (President's Award) from Beihang University in 2019. I also had the opportunity to work as a research intern at Amazon Web Services, Amazon Lab126, Nuro, and DJI.
My research expertise lies at the intersection of reinforcement learning, optimization, and embodied intelligence. I am interested in how to safely deploy learning-based systems to real-world decision-making applications. Currently, I'm exploring how to efficiently utilize the power of large pretrained models to enhance downstream task-solving capabilities of autonomous agents.
I'm actively exploring job opportunities at the moment. Please feel free to reach out via email if you believe there's a suitable match.
News & Updates
Amazon Web Service. Applied Scientist Intern, 2023 Summer.
Work on foundation models for decision-making.
Amazon, Lab126 Astro Robot Team. Applied Scientist Intern, 2022 Summer.
Work on RL for household robot exploration and planning.
Nuro, Inc. Machine Learning Research Intern. Host: Dr. Wei Liu, 2021 Summer.
Work on safe RL for self-driving vehicle behavior planning.
Dajiang Innovations (DJI) Technology Co., Ltd. Algorithm Engineer Intern, 2017 Summer.
Work on autonomous robot decision-making and motion planning.
Selected Publications and Preprints
TAIL: Task-specific Adapters for Imitation Learning with Large Pretrained Models
Zuxin Liu, Jesse Zhang, Kavosh Asadi, Yao Liu, Ding Zhao, Shoham Sabach, Rasool Fakoor.
Towards Robust and Safe Reinforcement Learning with Benign Off-policy Data
Zuxin Liu*, Zijian Guo*, Zhepeng Cen, Yihang Yao, Hanjiang Hu, Ding Zhao.
Constraint-Conditioned Policy Optimization for Versatile Safe Reinforcement Learning
Yihang Yao*, Zuxin Liu*, Zhepeng Cen, Jiacheng Zhu, Wenhao Yu, Tingnan Zhang, Ding Zhao.
Learning Shared Safety Constraints from Multi-task Demonstrations
Konwoo Kim, Gokul Swamy, Zuxin Liu, Ding Zhao, Sanjiban Choudhury, Zhiwei Steven Wu.
2023 ICML ILHF Workshop (oral).
Learning to Explore (L2E): Deep Reinforcement Learning-based Autonomous Exploration for Household Robot
Zuxin Liu, Mohit Deshpande, Xuewei Qi, Ding Zhao, Rajasimman Madhivanan, Arnie Sen.
2023 RSS Robot Representations Workshop.
On the Robustness of Safe Reinforcement Learning under Observational Perturbations
Zuxin Liu, Zijian Guo, Zhepeng Cen, Huan Zhang, Jie Tan, Bo Li and Ding Zhao.
2022 ICML Safe Learning for Autonomous Driving Workshop (Best Paper Runner-up).
2022 NeurIPS ML Safety Workshop (AI Risk Analysis Award).
Trustworthy Reinforcement Learning Against Intrinsic Vulnerabilities: Robustness, Safety, and Generalization
Mengdi Xu*, Zuxin Liu*, Peide Huang*, Wenhao Ding, Zhepeng Cen, Bo Li, Ding Zhao.
MAPPER: Multi-Agent Path Planning with Evolutionary Reinforcement Learning in Mixed Dynamic Environments
Zuxin Liu, Baiming Chen, Hongyi Zhou, Guru Koushik, Martial Hebert and Ding Zhao.
Some Undergraduate Projects
We developed a visual semantic SLAM system based on orb-slam2 and Mask-RCNN. The robot can build a semantic map for the environment and remember each object's position. With a speech recognition module, we can use voice command to let the robot find a particular object if he has seen before. This project won the first prize in the 2018 International Conference on Optics and Photonics(ICOPEN) 3-D Sensor Application Design Competition (1 out of 20 teams around the world).
Do you want to fly? Our VR-Multicopter system can help you to experience the feeling of fly. We use a VR device to control orientation of a gimbal that mounted on our multicopter. The stereo camera on the gimbal will send videos back to our VR equipment in real-time. Just move your head and enjoy the view from sky! This project won the first prize in the 2017 International Design and Innovation Competition (1 out of 14 teams around the world).
This project is designed for ICRA DJI Robotmaster AI challenge. The robots are required to autonomously find enemy robots and hit them (shoot rubber ball). More exciting information and videos about this robot platform and the relevant robot competition can be found here.
This cute robot could make different expressions according to user’s voice command. The movements of its eyes, eyebrow, and mouth etc are fully controlled by servo motors. Most of the materials are 3D printed.
Selected Honors and Awards
 ShenYuan Medal Award (10/4000), Beihang University
[2016&2017&2018] National Scholarship (top 1%), Ministry of Education of the People's Republic of China
[2016&2017&2018] University-level Outstanding Student, Beihang University
 Beijing Outstanding Student, Ministry of Education of Beijing
 28th First prize of the Feng Ru Cup Competition of Academic and Technological Works (top1%), Beihang University
 Dean's Award, Beihang University
[Conference Reviewer] CVPR 2023, ICCV 2023, ICML 2022-2023, NeurIPS 2022 (top reviewer, 8%), AISTATS 2022, ICRA 2020-2023, IROS 2020-2021
[Journal Reviewer] Reviewer for Journal of Field Robotics, IEEE RA-L, IEEE T-KDE, IEEE T-VT, IEEE T-SMC, Autonomous Robots, Pattern recognition