CV

Profile

PhD researcher in Robotics and AI / Control Systems and Robotics at the University of Manchester, specialising in world models, model-based reinforcement learning, predictive planning, and safe control for autonomous systems. My work focuses on compact representations of dynamic, interactive environments and action-conditioned learned dynamics for multi-step prediction and online planning through MPPI and related sampling-based control methods. Expected PhD completion: December 2026.

I work primarily with Python, C++, PyTorch, CARLA, MATLAB/Simulink and closed-loop simulation, with additional automotive engineering experience at Lotus Technology and NIO.

Core technical skills

Area	Methods / tools
World Models & ML	Model-Based RL, learned dynamics, latent world models, representation learning, BEV, occupancy models, multi-agent interaction modelling
Planning & Control	MPPI, MPC, sampling-based planning, motion planning, CBF, CVaR, risk-aware control, vehicle dynamics and control
Robotics & Simulation	CARLA, IPG CarMaker, closed-loop simulation, autonomous navigation, scenario-based evaluation
Programming	Python, C++, PyTorch, MATLAB/Simulink, SciPy, Git, Docker
Automotive Systems	Vector CANoe, CAN bus, DBC, diagnostic analysis, DFMEA
Languages	Chinese (native), English (professional), German (A1)

Professional & research experience

PhD Researcher — University of Manchester, UK | 2023–Present

Developed world-model-based planning methods for autonomous systems, using compact latent dynamics for online multi-step decision-making through MPPI-based predictive control.
Developed an interaction-aware MBRL framework combining latent representations, recurrent action-conditioned dynamics, pedestrian-interaction modelling, collision-risk prediction, reward/value estimation, and sampling-based planning.
Built CNN-based BEV and occupancy world-model variants using TD-MPC-style latent predictive-control architectures.
Developed safety-aware predictive-control methods combining sampling-based MPC, Control Barrier Functions, and risk-sensitive optimisation; reduced collisions by up to 65.7% relative to a nominal controller in evaluated studies.
Designed a multi-obstacle discrete-time CBF safety filter, reducing collision rate from 7.8% to 3.8% across 500 evaluated scenarios with 9.30 ms reported runtime.
Implemented and benchmarked autonomous-navigation agents in Python / PyTorch using custom environments and CARLA.
Supervised Master’s dissertation projects in autonomous systems and control.

Graduate Teaching Assistant — University of Manchester, UK | 2023–Present

Supported teaching and assessment in Robotic Systems, Applied Mechanics & Industrial Robotics, Digital Control, and Model Predictive Control.

ADAS Engineer — Lotus Technology, Hangzhou, China | 2022

Supported development of an ADAS speed-limit information feature integrating HD-map and camera-derived inputs.
Developed C++ software functions for feature logic and vehicle-signal interfaces.
Used Vector CANoe for CAN-signal monitoring, trace analysis, diagnostics, functional debugging and system integration.
Conducted DFMEA for the ADAS domain controller, including failure-mode analysis and mitigation actions for assisted steering, automated-driving and HMI functions.

BMS Software Intern — NIO, Shanghai, China | 2021

Developed a three-point high-voltage insulation-detection algorithm in MATLAB/Simulink using model-based development / code-generation workflows.
Worked with DBC files and CAN signals for BMS monitoring, validation and debugging.
Analysed chamber, bench and vehicle-level data in CANoe, including high-temperature investigation of PTC insulation faults.

High-Voltage Lead — Hunan University Formula Student Electric Team | 2017–2019

Led system-level concept design and integration of the EV high-voltage architecture.
Used MATLAB and AVL CRUISE for powertrain modelling and performance evaluation.
Applied DFMEA in high-voltage-system development.
Led a lightweight HV battery structural redesign that achieved a 21% mass reduction while maintaining performance and safety requirements.

Education

PhD Candidate in Electrical Engineering — University of Manchester, UK
2023–Present; expected completion December 2026. Research focus: world models, model-based reinforcement learning, autonomous driving, predictive planning and safety-critical control. Supervisors: Prof. Guido Herrmann and Prof. Alessandra Parisio.

MSc in Advanced Control and Systems Engineering — University of Manchester, UK
2020–2022. Thesis: A Hierarchical Forecasting Model of Pedestrian Crossing Behaviour for Autonomous Vehicles. Research Excellence Award and Top Student Award, 2022.

BEng in Vehicle Engineering — Hunan University, China
2016–2020. Thesis: Motion Planning and Dynamics Control of Autonomous Racing Car. Four-wheel vehicle model, Pacejka tyre-force modelling, MPC trajectory planning, and IPG CarMaker closed-loop evaluation.

Selected publications

Yang, G., Chen, A. S., Parisio, A., & Herrmann, G. (2026). A Safety Framework for Uncertainty-Aware Model Predictive Path Integral Control in Shared-Space Navigation. IEEE CDC 2026. Accepted.
Yang, G., Chen, A. S., Parisio, A., & Herrmann, G. (2026). PIP-CN: Prediction-Integrated Planning for Crowd Navigation. IEEE/RSJ IROS 2026. Accepted.
Yang, G., Chen, A. S., Parisio, A., & Herrmann, G. (2026). Risk-Bounded Predictive Control for Automated Vehicles in Shared-Space Areas. AVEC 2026.
Zhang, C., Ma, Y., Yang, G., & Chen, T. (2025). An Integrated Industrial PV Panel Cleaning Recommendation System for Optimal Dust Removal. Applied Energy, 377, 124692.
Yang, G., Lopez Pulgarin, E. J., & Herrmann, G. (2024). A Hierarchical Forecasting Model of Pedestrian Crossing Behaviour for Autonomous Vehicle. IEEE Access, 12, 9025–9037.

Honours & awards

Full PhD Scholarship, University of Manchester.
Research Excellence Award and Top Student Award, University of Manchester, 2022.
Overall Champion, Formula Student Electric China, 2019.