Papers
4
Total Citations
16
H-Index
3
About
Jaehwi Jang is a rising researcher at the forefront of humanoid robotics and robot learning, whose work bridges the gap between simulation and real-world dexterity. His primary research areas include whole-body loco-manipulation, sample-efficient reinforcement learning, and inverse constraint learning. Jang’s most notable contribution, "Opt2Skill" (2025, 8 citations), pioneers a method to imitate dynamically-feasible whole-body trajectories, enabling humanoid robots to perform versatile tasks like manipulation while walking—a breakthrough in handling high-dimensional, contact-rich dynamics. He further advanced the field with "Learn to Teach" (2025, 3 citations), introducing a privileged learning framework that dramatically reduces the simulation samples needed for humanoid locomotion over uneven terrain, addressing a critical bottleneck in real-world deployment. Beyond locomotion, Jang has innovated in robot understanding, developing "Inverse Constraint Learning and Generalization by Transferable Reward Decomposition" (2023, 3 citations) to infer hidden constraints from demonstrations, and "SGGNet²" (2023, 2 citations) for speech-guided navigation, enhancing accessibility for non-expert users. His work consistently tackles the core challenges of sample efficiency and dynamic feasibility, positioning him as a key contributor to the next generation of capable, real-world humanoid robots.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4