Papers
17
Total Citations
323
H-Index
10
About
Jeremy Dao is a leading researcher in bipedal locomotion, whose work bridges reinforcement learning, sim-to-real transfer, and dynamic robot control. His primary research focuses on developing agile, robust locomotion skills for bipedal robots like Cassie, with key contributions in high-speed running, vision-based terrain adaptation, and loco-manipulation. Dao’s most cited paper (63 citations) introduces an iterative reinforcement learning framework for designing dynamic locomotion skills, addressing the practical challenges of reward function tuning. His work on learning task-space actions for bipedal locomotion (45 citations) and optimizing gaits for the 100m dash (30 citations) demonstrates his ability to push robots to athletic performance levels comparable to humans. Notably, his 2024 paper on vision-based locomotion (32 citations) enables robots to anticipate and adapt to challenging terrain using visual perception, a critical step beyond blind controllers. Dao’s sim-to-real transfer techniques, including handling unsensed dynamic loads (29 citations) and humanoid box loco-manipulation (25 citations), showcase his commitment to deploying learned policies on real hardware. His work has been widely recognized, with over 290 total citations, and has set new benchmarks for dynamic bipedal maneuvers, standing, and walking under disturbances.
Research Focus
Key Achievements
Top Papers
- 1
- 2Learning Task Space Actions for Bipedal Locomotion45 citations · 2021
- 3Learning Vision-Based Bipedal Locomotion for Challenging Terrain32 citations · 2024
- 4
- 5Sim-to-Real Learning for Bipedal Locomotion Under Unsensed Dynamic Loads29 citations · 2022
- 6Sim-to-Real Learning for Humanoid Box Loco-Manipulation25 citations · 2024
- 7
- 8
- 9Learning Dynamic Bipedal Walking Across Stepping Stones15 citations · 2022
- 10Dynamic Bipedal Turning through Sim-to-Real Reinforcement Learning14 citations · 2022