Wenhao Ding

Wenhao Ding (丁文浩)

Email: wenhaod AT nvidia DOT com

My research centers on enhancing the safety and robustness of physical autonomous systems, with a current emphasis on autonomous vehicles. I work to improve the reasoning capabilities of learning-based models by using multi-modal large language models, reinforcement learning, and causal discovery.

2024 -

I am a Research Scientist at the Autonomous Vehicle Group of NVIDIA Research. I work on the closed-loop simulation and safety for autonomous systems.

2019 - 2024

My Ph.D. research focused on safety-critical scenarios generation. My thesis title is ''Generative AI for Critical Digital Twins''.

2023 - 2023

I was a Research Intern in the Autonomous Vehicle Group led by Prof. Marco Pavone. I collaborated with Chaowei Xiao and Yulong Cao on a scenario generation project.

2022 - 2022

I was an Applied Scientist Intern of the Astro team at Lab126, where I collaborated with Nathalie Majcherczyk and Mohit Deshpande on the active perception project.

2021 - 2021

I was a Machine Learning Research Intern at Bosch Center for Artificial Intelligence, where I worked on traffic flow analysis and clustering.

2014 - 2018

I received my Bachelor degree from the Department of Electronic Engineering at Tsinghua University. I was a member of the Spark Program and a former leader of Skyworks.

2025/04 - CAT-K is accepted to CVPR as an oral paper.

2024/09 - Two papers about Offline RL are accepted to NeurIPS.

2024/08 - One paper about scenario generation is accepted to ECCV as an Oral paper.

2024/07 - We are organizing the 2024 IEEE International Automated Vehicle Validation Conference. Consider submitting your paper and special session proposals!

2024/06 - I join NVIDIA Research as a Research Scientist at the Autonomous Vehicle Group.

2024/03 - One paper using causal-aware representation for driving is accepted to RA-L.

2024/02 - One paper about privacy in robotics is accepted to ICRA 2024.

2023/09 - One paper combining causality and robustness is accepted to NeurIPS 2023.

2023/04 - One paper about reward-conditioned offline RL accepted to ICML 2023.

2023/03 - We are hosting the Secure and Safe Autonomous Driving (SSAD) Workshop and Challenge at CVPR 2023!

2023/01 - One paper about embodied AI accepted to ICRA 2023.

2022/12 - My research is covered by CMU Engineering News Driving autonomy into the metaverse!

2022/10 - Glad to present our work about causal discovery in Wayve!

2022/09 - Two papers got accepted to NeurIPS 2022. See you in New Orleans!

2022/09 - One paper about using causal graph to generate safety-critical scenarios got accepted to CoRL 2022.

2022/08 - My research proposal wins 2022 Qualcomm Innovation Fellowship, North America.

2022/03 - I will start my Applied Scientist Intern in Amazon Lab126 working on Astro.

2022/02 - I gave a talk about scenario generation in Upenn hosted by Prof. Rahul Mangharam.

2022/01 - We are holding the Season Depth Challenge in IROS 2022 and ICRA 2022.

2022/01 - I gave a talk about scenario generation in Stanford SISL Lab.

RealDrive: Retrieval-Augmented Driving with Diffusion Models

Wenhao Ding, Sushant Veer, Yuxiao Chen, Yulong Cao, Chaowei Xiao, Marco Pavone
Under Review

website / arXiv / code / bibtex

MoDoMoDo: Multi-Domain Data Mixtures for Multimodal LLM Reinforcement Learning

Yiqing Liang, Jielin Qiu, Wenhao Ding, Zuxin Liu, James Tompkin, Mengdi Xu, Mengzhou Xia, Zhengzhong Tu, Laixi Shi, Jiacheng Zhu
Under Review

website / arXiv / code / bibtex

CrashAgent: Crash Scenario Generation via Multi-modal Reasoning

Miao Li, Wenhao Ding, Haohong Lin, Yiqi Lyu, Yihang Yao, Yuyou Zhang, Ding Zhao
Under Review

website / arXiv / code / bibtex

Surprise Potential as a Measure of Interactivity in Driving Scenarios

Wenhao Ding, Sushant Veer, Karen Leung, Yulong Cao, Marco Pavone
Under Review

website / arXiv / code / bibtex

Closed-Loop Supervised Fine-Tuning of Tokenized Traffic Models

Zhejun Zhang, Peter Karkus, Maximilian Igl, Wenhao Ding, Yuxiao Chen, Boris Ivanovic, Marco Pavone
(Oral) IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2025

website / arXiv / code / bibtex

BECAUSE: Bilinear Causal Representation for Generalizable Offline Model-based Reinforcement Learning

Haohong Lin, Wenhao Ding, Jian Chen, Laixi Shi, Jiacheng Zhu, Bo Li, Ding Zhao
Conference on Neural Information Processing Systems (NeurIPS) 2024

website / arXiv / code / bibtex

OASIS: Conditional Distribution Shaping for Offline Safe Reinforcement Learning

Yihang Yao*, Zhepeng Cen*, Wenhao Ding, Haohong Lin, Shiqi Liu, Tingnan Zhang, Wenhao Yu, Ding Zhao
Conference on Neural Information Processing Systems (NeurIPS) 2024

website / arXiv / code / bibtex

RealGen: Retrieval Augmented Generation for Controllable Traffic Scenarios

Wenhao Ding*, Yulong Cao*, Ding Zhao, Chaowei Xiao, Marco Pavone
(Oral) European Conference on Computer Vision (ECCV) 2024

website / arXiv / code / bibtex

Safety-aware Causal Representation for Trustworthy Reinforcement Learning in Autonomous Driving

Haohong Lin, Wenhao Ding, Zuxin Liu, Yaru Niu, Jiacheng Zhu, Yuming Niu, Ding Zhao
IEEE Robotics and Automation Letters (RA-L) 2024

arXiv / code / bibtex

Seeing is not Believing: Robust Reinforcement Learning against Spurious Correlation

Wenhao Ding*, Laixi Shi*, Yuejie Chi, Ding Zhao
Conference on Neural Information Processing Systems (NeurIPS) 2023, New Orleans

website / arXiv / code / bibtex

Your Room is not Private: Gradient Inversion Attack for Deep Q-Learning

Miao Li, Wenhao Ding, Ding Zhao
International Conference on Robotics and Automation (ICRA) 2024, Japan

arXiv / code / bibtex

Bayesian Reparameterization of Reward-Conditioned Reinforcement Learning with Energy-based Models

Wenhao Ding*, Tong Che*, Ding Zhao, Marco Pavone
International Conference on Machine Learning (ICML) 2023, Hawaii

arXiv / code / bibtex

Semantically Adversarial Scene Generation With Explicit Knowledge Guidance

Wenhao Ding, Haohong Lin, Bo Li, Ding Zhao
IEEE Transactions on Intelligent Transportation Systems (T-ITS), Feb, 2025

arXiv / Xplore / code / bibtex

Learning to View: Decision Transformers for Active Object Detection

Wenhao Ding, Nathalie Majcherczyk, Mohit Deshpande, Xuewei Qi, Ding Zhao, Rajasimman Madhivanan, Arnie Sen
IEEE International Conference on Robotics and Automation (ICRA) 2023, London

arXiv / code / bibtex

A Survey on Safety-Critical Driving Scenario Generation -- A Methodological Perspective

Wenhao Ding, Chejian Xu, Mansur Arief, Haohong Lin, Bo Li, Ding Zhao
IEEE Transactions on Intelligent Transportation Systems (T-ITS), March, 2023

arXiv / Xplore / code / bibtex

Generalizing Goal-Conditioned Reinforcement Learning with Variational Causal Reasoning

Wenhao Ding, Haohong Lin, Bo Li, Ding Zhao
Conference on Neural Information Processing Systems (NeurIPS) 2022, New Orleans

arXiv / code / bibtex

SafeBench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles

Chejian Xu*, Wenhao Ding*, Weijie Lyu, Zuxin Liu, Shuai Wang, Yihan He, Hanjiang Hu, Ding Zhao, Bo Li
Conference on Neural Information Processing Systems (NeurIPS) 2022, New Orleans

website / arXiv / code / bibtex

CausalAF: Causal Autoregressive Flow for Safety-Critical Driving Scenario Generation

Wenhao Ding, Haohong Lin, Bo Li, Ding Zhao
Conference on Robot Learning (CoRL) 2022, New Zealand
Abridged in ICML 2022 Workshop on Safe Learning for Autonomous Driving

arXiv / code / bibtex

Multimodal Safety-Critical Scenarios Generation for Decision-Making Algorithms Evaluation

Wenhao Ding, Baiming Chen, Bo Li, Kim Ji Eun, Ding Zhao
IEEE Robotics and Automation Letters (RA-L) with ICRA 2021, Xi'an

arXiv / code / bibtex

Task-Agnostic Online Reinforcement Learning with an Infinite Mixture of Gaussian Processes

Mengdi Xu, Wenhao Ding, Jiacheng Zhu, Zuxin Liu, Baiming Chen, Ding Zhao
Neural Information Processing Systems (NeurIPS) 2020, Vancouver

arXiv / code / bibtex

Conference Reviewer: ICML, ICLR, NeurIPS, AISTATS, CVPR, ECCV, ICCV, CoRL, ICRA, IROS, IJCAI, ICASSP, ITSC, ICME

Journal Reviewer: TMLR, IEEE RA-L, IEEE TNNLS, IEEE T-ITS, IEEE T-IV, IEEE Access, IEEE TII, IEEE MM, Scientific Report

Scholar / Github / LinkedIn / Twitter

Wenhao Ding (丁文浩)

News & Updates

Selected Publications

RealDrive: Retrieval-Augmented Driving with Diffusion Models

MoDoMoDo: Multi-Domain Data Mixtures for Multimodal LLM Reinforcement Learning

CrashAgent: Crash Scenario Generation via Multi-modal Reasoning

Surprise Potential as a Measure of Interactivity in Driving Scenarios

Closed-Loop Supervised Fine-Tuning of Tokenized Traffic Models

BECAUSE: Bilinear Causal Representation for Generalizable Offline Model-based Reinforcement Learning

OASIS: Conditional Distribution Shaping for Offline Safe Reinforcement Learning

RealGen: Retrieval Augmented Generation for Controllable Traffic Scenarios

Safety-aware Causal Representation for Trustworthy Reinforcement Learning in Autonomous Driving

Seeing is not Believing: Robust Reinforcement Learning against Spurious Correlation

Your Room is not Private: Gradient Inversion Attack for Deep Q-Learning

Bayesian Reparameterization of Reward-Conditioned Reinforcement Learning with Energy-based Models

Semantically Adversarial Scene Generation With Explicit Knowledge Guidance

Learning to View: Decision Transformers for Active Object Detection

A Survey on Safety-Critical Driving Scenario Generation -- A Methodological Perspective

Generalizing Goal-Conditioned Reinforcement Learning with Variational Causal Reasoning

SafeBench: A Benchmarking Platform for Safety Evaluation of Autonomous Vehicles

CausalAF: Causal Autoregressive Flow for Safety-Critical Driving Scenario Generation

Multimodal Safety-Critical Scenarios Generation for Decision-Making Algorithms Evaluation

Task-Agnostic Online Reinforcement Learning with an Infinite Mixture of Gaussian Processes

Academic Services