Tianhu Peng

Ph.D. Researcher in Robotics @ UCL

Focusing on legged robots, reinforcement learning, and sim-to-real transfer. Aspiring entrepreneur with strong industry background at Tencent Robotics X.

Tianhu Peng

01. Education Journey

Present

Ph.D. in Robotics (Year 3)

University College London (UCL)

Specializing in Legged Robotics, Reinforcement Learning, and advancing the state-of-the-art in autonomous locomotion and expressive motion generation.

Past

MSc Human and Biological Robotics

Imperial College London

Deepened expertise in biomechanics, advanced control systems, and human-robot interaction.

2017 - 2020

BEng Mechatronics and Robotics Engineering

University of Sheffield

Built a strong foundational knowledge in mechanical design, electronics, and foundational robotics engineering.

02. Professional Experience

Robotics Researcher Intern

Tencent Robotics X (Shenzhen)

8 Months

Gained rich industry experience working directly on advanced legged robot platforms. Contributed to cutting-edge research translating theoretical robotics into robust, real-world deployment.

  • Legged Robotics
  • Reinforcement Learning
  • Sim-to-Real Transfer
  • Motion Generation

03. Featured Moments

04. Selected Publications

39 Citations

Deep reinforcement learning for robotic bipedal locomotion: a brief survey

Overview of DRL applications in bipedal systems.

11 Citations

Learning bipedal walking on a quadruped robot via adversarial motion priors

Transferring complex walking gaits using AMP.

Project Page 5 Citations

Gait-Conditioned Reinforcement Learning with Multi-Phase Curriculum for Humanoid Locomotion

Advanced curriculum learning for humanoids.

3 Citations

Hierarchical Intention-Aware Expressive Motion Generation for Humanoid Robots

Generating context-aware, expressive motions.

SCDP: Learning Humanoid Locomotion from Partial Observations via Mixed-Observation Distillation

Robust locomotion learning from partial states.

A High-Fidelity Digital Twin for Robotic Manipulation Based on 3D Gaussian Splatting

Leveraging 3D Gaussian Splatting for high-fidelity twins.

Sim-to-Real Transfer in Deep Reinforcement Learning for Bipedal Locomotion

Bridging the reality gap in continuous control.

OmniDexter: A Modular Tendon-Driven Robotic Wrist with Enhanced Precision and Versatility

Innovative tendon-driven wrist design.

05. Get In Touch

Whether you have a question about my research, are interested in collaboration, or want to discuss startup opportunities.