Zhiyu Huang

Zhiyu Huang

University of California, Los Angeles

I am currently a postdoctoral scholar at the UCLA Mobility Lab, working under the guidance of Prof. Jiaqi Ma. Previously, I was a research intern at the NVIDIA Research Autonomous Vehicle Research Group and a visiting student researcher at UC Berkeley in the Mechanical Systems Control (MSC) Lab. I earned my Ph.D. from Nanyang Technological University (NTU), where I conducted research in the Automated Driving and Human-Machine System (AutoMan) Lab under the supervision of Prof. Chen Lyu.

My research focuses on the intersection of robotics, mobility, and artificial intelligence (AI). I aim to develop algorithms and techniques that enable machines/robots to interact with humans naturally and make intelligent decisions. My research interests include deep learning, reinforcement learning, and generative AI, applied to areas such as perception, prediction, decision-making, simulation in autonomous driving, and human-machine interaction. My work has led to the publication of over 30 papers in top AI, ITS, and robotics journals and conferences.

🔥 News

  • 2024.09:  🎉🎉 Our NAVSIM paper on end-to-end driving benchmark has been accepted at NeurIPS 2024 Datasets and Benchmarks Track!
  • 2024.07:  🎉🎉 Our ITSC invited session on Learning-powered and Knowledge-driven Autonomous Driving has received 11 paper submissions, all of which were accepted. Congratulations to all the authors! Looking forward to seeing you in Edmonton, Canada!
  • 2024.06:  🎉🎉 Our team secured first place in the Waymo Open Dataset Occupancy Flow Challenge and second place in the Sim Agents Challenge! Check out our technical reports on the Waymo challenge website and CVPR 2024 Workshop on Autonomous Driving.
  • 2024.05:  🎉🎉 Our paper on online belief prediction and POMDP planning has been accepted by RAL!
  • 2024.01:  🎉🎉 Our paper on joint prediction and planning for tree policy has been accepted by ICRA! See you in Yokohama, Japan!
  • 2023.11:   I was invited by zdjszx.com to give a public lecture on “Scalable, Learnable, and Interactive Decision-making for Autonomous Driving”. The recorded version of the lecture (in Chinese) is available for viewing on bilibili.
  • 2023.10:  🎉🎉 Our paper on brain-inspired reinforcement learning for safe autonomous driving has been accepted by TPAMI!
  • 2023.09:  🎉🎉 We won the best paper runner-up award in ITSC 2023!
  • 2023.09:  🎉🎉 Our paper on human-guided reinforcement learning for robot navigation has been accepted by TPAMI!
  • 2023.08:  🎉🎉 Our GameFormer paper has been accepted by ICCV as Oral presentation!
  • 2023.06:  🎉🎉 Our team won the innovation award in the nuPlan Planning Challenge! Check out our report and presentation on our GameFormer Planner.

📝 Publications

Highlights

sym

Gen-Drive: Enhancing Diffusion Generative Driving Policies with Reward Modeling and Reinforcement Learning Fine-tuning

Zhiyu Huang, Xinshuo Weng, Maximilian Igl, Yuxiao Chen, Yulong Cao, Boris Ivanovic, Marco Pavone, Chen Lv

arXiv, 2024

Paper | Project

  • We introduce the Gen-Drive framework, which shifts from the traditional prediction and deterministic planning framework to a generation-then-evaluation planning paradigm.
sym

Versatile Behavior Diffusion for Generalized Traffic Simulation

Zhiyu Huang, Zixu Zhang, Ameya Vaidya, Yuxiao Chen, Chen Lv, Jaime Fernández Fisac

arXiv, 2024

Paper | Project |

  • We propose VBD, a novel traffic scenario generation framework that utilizes diffusion generative models to predict scene-consistent and controllable multi-agent interactions in closed-loop settings.
sym

Learning Online Belief Prediction for Efficient POMDP Planning in Autonomous Driving

Zhiyu Huang, Chen Tang, Chen Lv, Masayoshi Tomizuka, Wei Zhan

IEEE Robotics and Automation Letters, 2024

Paper

  • We propose an online belief-update-based behavior prediction model and an efficient planner for POMDPs. We develop a Transformer-based prediction model, enhanced with a recurrent neural memory model, to dynamically update latent belief state and infer the intentions of other agents.
sym

DTPP: Differentiable Joint Conditional Prediction and Cost Evaluation for Tree Policy Planning in Autonomous Driving

Zhiyu Huang, Peter Karkus, Boris Ivanovic, Yuxiao Chen, Marco Pavone, Chen Lv

IEEE International Conference on Robotics and Automation (ICRA), 2024

Paper |

  • We employ a tree-structured policy planner and propose a differentiable joint training framework for both ego-conditioned prediction and cost evaluation models, resulting in a direct improvement of the final planning performance.
sym

Learning-enabled Decision-making for Autonomous Driving: Framework and Methodology

PhD Thesis, 2024

Thesis

  • This thesis presents a comprehensive framework and a series of learning-based methodologies for decision-making in AVs, with the objective of improving the scalability, adaptability, and alignment of their decision-making systems.
sym

GameFormer: Game-theoretic Modeling and Learning of Transformer-based Interactive Prediction and Planning for Autonomous Driving

Zhiyu Huang, Haochen Liu, Chen Lv

IEEE/CVF International Conference on Computer Vision (ICCV), 2023

Oral presentation (top 3%)

Paper | Project | | GameFormer Planner

  • We address the interaction prediction problem by formulating it with hierarchical game theory and implementing it with TransFormer networks.
sym

Learning Interaction-aware Motion Prediction Model for Decision-making in Autonomous Driving

Zhiyu Huang, Haochen Liu, Jingda Wu, Wenhui Huang, Chen Lv

IEEE International Conference on Intelligent Transportation Systems (ITSC), 2023

Best Paper Runner-up Award

Paper |

  • We propose an interaction-aware motion prediction model that is able to predict other agents’ future trajectories according to the ego agent’s future plans, i.e., their reactions to the ego’s actions.
sym

Conditional Predictive Behavior Planning with Inverse Reinforcement Learning for Human-like Autonomous Driving

Zhiyu Huang, Haochen Liu, Jingda Wu, Chen Lv

IEEE Transactions on Intelligent Transportation Systems, 2023

Paper

  • Distinguished from existing learning-based methods that directly output decisions, we introduce a predictive behavior planning framework that learns to predict and evaluate from human driving data.
sym

Differentiable Integrated Motion Prediction and Planning with Learnable Cost Function for Autonomous Driving

Zhiyu Huang, Haochen Liu, Jingda Wu, Chen Lv

IEEE Transactions on Neural Networks and Learning Systems, 2023

Paper | Project |

  • We propose an end-to-end differentiable framework that integrates prediction and planning modules and is able to learn the cost function from data.
sym

Multi-modal Motion Prediction with Transformer-based Neural Network for Autonomous Driving

Zhiyu Huang, Xiaoyu Mo, Chen Lv

IEEE International Conference on Robotics and Automation (ICRA), 2022

Paper

  • We propose a neural prediction framework based on the Transformer structure to model the relationship among the interacting agents and extract the attention of the target agent on the map waypoints.
sym

Efficient Deep Reinforcement Learning with Imitative Expert Priors for Autonomous Driving

Zhiyu Huang, Jingda Wu, Chen Lv

IEEE Transactions on Neural Networks and Learning Systems, 2022

Paper | Project |

  • We propose a novel framework to incorporate human prior knowledge in DRL, in order to improve the sample efficiency and save the effort of designing sophisticated reward functions.
sym

Driving Behavior Modeling using Naturalistic Human Driving Data with Inverse Reinforcement Learning

Zhiyu Huang, Jingda Wu, Chen Lv

IEEE Transactions on Intelligent Transportation Systems, 2021

Paper |

  • We propose a structural assumption about internal reward function-based human driving behavior and employ sampling-based maximum entropy inverse reinforcement learning (IRL) algorithm to infer the reward function parameters from naturalistic human driving data.
sym

Multi-modal sensor fusion-based deep neural network for end-to-end autonomous driving with scene understanding

Zhiyu Huang, Chen Lv, Yang Xing, Jingda Wu

IEEE Sensors Journal, 2020

Paper

  • We propose a novel deep neural network-based system for end-to-end autonomous driving, consisting of multimodal sensor fusion, scene understanding, and conditional driving policy modules.

All Publications

Journal

Conference

Preprint

🎖 Honors and Awards

📖 Education

  • 2019 - 2024, Doctor of Philosophy, Robotics and Intelligent Systems, Nanyang Technological University, Singapore
  • 2015 - 2019, Bachelor of Engineering, Vehicle Engineering, Chongqing University, Chongqing, China

📚 Academic Services

Program Committee

Associate Editor

  • OJ-ITS – IEEE Open Journal of Intelligent Transportation Systems

Journal Reviewer

  • IEEE Transactions on Intelligent Transportation Systems
  • IEEE Transactions on Neural Networks and Learning Systems
  • IEEE Transactions on Intelligent Vehicles
  • IEEE Transactions on Cybernetics
  • IEEE Robotics and Automation Letters
  • Transportation Research Part C: Emerging Technologies
  • Engineering Applications of Artificial Intelligence
  • Artificial Intelligence Review

Conference Reviewer

  • IEEE International Conference on Robotics and Automation (ICRA) 2022 – 2025
  • IEEE Intelligent Vehicles Symposium (IV) 2022 – 2024
  • IEEE Intelligent Transportation Systems Conference (ITSC) 2022 – 2024
  • IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2023
  • European Conference on Computer Vision (ECCV) 2024
  • Conference on Robot Learning (CoRL) 2024
  • IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2025