Papers
2
Total Citations
4
H-Index
2
About
Hao Sha is a rising researcher in autonomous robotics, specializing in goal-driven navigation and decision-making under uncertainty. His work addresses two fundamental challenges in robot autonomy: operating in unknown environments and handling sensor imperfections. In his 2024 paper "Learning Hierarchical Graph-Based Policy for Goal-Reaching in Unknown Environments," Sha tackles the dual demands of large-scale perception and long-horizon decision-making—critical as robots move beyond controlled labs into complex, real-world spaces. His approach uses hierarchical graph representations to bridge local sensing with global planning, a contribution that has already garnered attention with 2 citations in its first year. In parallel, his work on "RGBD-based Image Goal Navigation with Pose Drift" introduces a drift-resisting topo-metric graph that enables robots to navigate using only relative poses, even when odometry errors accumulate. The key innovation—an error-sharing mechanism within the graph representation—allows the system to maintain robust localization without external infrastructure. Sha’s research sits at the intersection of reinforcement learning, graph-based mapping, and robust perception, offering practical solutions for long-duration autonomous missions. His work is particularly relevant for field robotics applications where GPS is unavailable and sensor noise is inevitable.
Research Focus
Key Achievements
Top Papers
- 1
- 2