Papers
2
Total Citations
458
H-Index
2
About
Tianbao Zhang is a pioneering researcher in the field of biomechanics and legged robotics, with a focus on energy-efficient locomotion and reinforcement learning for dynamic walking. His major contributions lie in bridging passive dynamics with active control, demonstrating that simple, underactuated robots can achieve stable 3D walking through minimal actuation and learning. In his seminal 2005 paper on stochastic policy gradient reinforcement learning, Zhang showed that a physical biped robot could learn a robust walking policy from scratch in just 20 minutes—a breakthrough that has garnered over 270 citations for its practical impact on real-world robot learning. His earlier 2004 work on actuating a 3D passive dynamic walker, with 188 citations, introduced reduced-order models that enabled stable gait tuning with minimal degrees of freedom, laying the groundwork for efficient, human-like walking machines. Zhang’s work is notable for its emphasis on simplicity and biological inspiration, achieving rapid learning and stable locomotion without complex sensors or high-torque actuators. His research continues to influence modern reinforcement learning approaches in robotics and prosthetics.
Research Focus
Key Achievements
Top Papers
- 1Stochastic policy gradient reinforcement learning on a simple 3D biped270 citations · 2005
- 2Actuating a simple 3D passive dynamic walker188 citations · 2004