About
Experiences
2023.09 – 2026.07 (expected)
M.Eng Student & Researcher @ School of Automotive Studies, Tongji University
AI4Science for PEM fuel cells and PEM water electrolyzers.
Focus on ion contamination diagnosis, physics-informed modeling, and multi-modal data fusion.
Instructors: Xuezhe Wei, Jiangong Zhu, Xiaoping Wu
Instructors: Xuezhe Wei, Jiangong Zhu, Xiaoping Wu
2025.09 – 2026.01
Robotics Locomotion & Manipulation Algorithm Intern @ Agibot (智元机器人), Shenzhen, China
Project: Whole-body Loco-manipulation (Sim2Real)
• Debugged AIMRT–ROS SDK communication for full-stack HW/SW integration.
• Developed dynamics-constrained motions with Crocoddyl + Pinocchio for dataset expansion.
• Built retargeting evaluation methods based on GMR & PHUMA.
• Developed VR teleoperation whole-body controller for Agibot Lingxi X2; validated Sim2Sim & Sim2Real.
Instructor: Shaohe Du
• Developed dynamics-constrained motions with Crocoddyl + Pinocchio for dataset expansion.
• Built retargeting evaluation methods based on GMR & PHUMA.
• Developed VR teleoperation whole-body controller for Agibot Lingxi X2; validated Sim2Sim & Sim2Real.
Instructor: Shaohe Du
2025.04 – 2025.09
Embodied Locomotion Control Algorithm Intern @ Xiaomi Robotics, Beijing, China
Project: Biped Locomotion with Toe Modeling
• Investigated functional roles of toe DoF in biped locomotion; validated with Cassie-style simulation for stability & gait diversity.
• Built motion data processing & visualization pipeline; converted retargeted mocap into RL-ready trajectories for PPO training.
• Integrated imitation learning methods (AMP, OMNIH2O) in Legged Gym; refactored DataLoader to support custom trajectory formats.
• Trained and debugged PPO locomotion policies in Isaac Gym; optimized rewards, contact forces, and PPO hyperparameters.
Instructor: Jianhua Feng
• Built motion data processing & visualization pipeline; converted retargeted mocap into RL-ready trajectories for PPO training.
• Integrated imitation learning methods (AMP, OMNIH2O) in Legged Gym; refactored DataLoader to support custom trajectory formats.
• Trained and debugged PPO locomotion policies in Isaac Gym; optimized rewards, contact forces, and PPO hyperparameters.
Instructor: Jianhua Feng
2025.01 – 2025.03
Embodied Perception Algorithm Intern @ Anker Innovations (安克创新), Shanghai, China
Project: 3D Simulation Asset Generation & Multi-engine Learning Survey
• Surveyed heterogeneous simulation/rendering stacks: Gazebo, MuJoCo, Isaac Sim; Unreal/Unity; ROS2/LCM; NeRF & 3D/4D Gaussian Splatting.
• Implemented ROS2 control interfaces in Isaac Sim for robot arm control, sensor publishing, and command streaming.
• Built an end-to-end pipeline: image capture → 3DGS reconstruction → point cloud editing → foreground/background separation → sim scene generation.
Instructor: Minyi Chen
• Implemented ROS2 control interfaces in Isaac Sim for robot arm control, sensor publishing, and command streaming.
• Built an end-to-end pipeline: image capture → 3DGS reconstruction → point cloud editing → foreground/background separation → sim scene generation.
Instructor: Minyi Chen
2024.03 – 2024.10
ADAS Functional Test & Automation Intern @ NIO (蔚来), Shanghai, China
Project: Regulatory Scenario & State-machine Transition Automated Testing for Production ADAS
• Deployed WorldSim and communication modules (DDS, feature_app) on Ubuntu to support automated ADAS testing.
• Modeled regulatory scenarios and authored test cases for state-machine transitions and boundary conditions across AEB/BSD/LCA/SDOW, ACC/NP(NOP)/AHB/ALC, and parking features.
• Developed automation scripts for release → run → result collection → report generation; improved efficiency and reproducibility.
• Performed log replay (LogSim) for regression testing and bug localization.
Instructor: Xinghai Huang
• Modeled regulatory scenarios and authored test cases for state-machine transitions and boundary conditions across AEB/BSD/LCA/SDOW, ACC/NP(NOP)/AHB/ALC, and parking features.
• Developed automation scripts for release → run → result collection → report generation; improved efficiency and reproducibility.
• Performed log replay (LogSim) for regression testing and bug localization.
Instructor: Xinghai Huang
2023.04 – 2023.08
Data Communication Intern @ Porsche Engineering, Shanghai, China
Project: ADAS Simulation Scenarios & Data Communication Pipeline
• Built a dual-system co-simulation stack (Windows + WSL Ubuntu) with CARLA + ROS for cross-platform communication and integration testing.
• Reconstructed real-world Jiading (Shanghai) road segments in CARLA; implemented typical FCW/SCW scenarios (hard brake, cut-in, etc.).
• Designed a multi-node middleware pipeline: CARLA sensors → ROS perception → decision logic → socket → Windows QT dashboard visualization.
• Developed UI regression automation with Behave + Selenium; built QT/QML dashboard for speed & warning indicators.
Instructor: Mojun Qian
• Reconstructed real-world Jiading (Shanghai) road segments in CARLA; implemented typical FCW/SCW scenarios (hard brake, cut-in, etc.).
• Designed a multi-node middleware pipeline: CARLA sensors → ROS perception → decision logic → socket → Windows QT dashboard visualization.
• Developed UI regression automation with Behave + Selenium; built QT/QML dashboard for speed & warning indicators.
Instructor: Mojun Qian
2018.09 – 2023.06
B.Eng Student @ School of Automotive Studies, Tongji University
Major in Automotive Engineering, with early research in hydrogen energy and fuel cell systems.
Instructor: Haifeng Dai,
Instructor: Haifeng Dai,
Publications ( / )
In preparation, 2025 (to be updated)
Discover Applied Sciences, 2025
[Title to be updated]
[Venue / Year to be updated]
Selected Honors & Awards
- (to be updated)