Changshuo Li
Papers
2
Total Citations
46
H-Index
2
About
Changshuo Li is a researcher in robot learning and manipulation, focusing on enabling robots to acquire complex skills from minimal human demonstrations. His work bridges hierarchical task decomposition and constraint-based motion planning, addressing key challenges in autonomous robotics. In his highly cited 2018 paper, "Simultaneous learning of hierarchy and primitives for complex robot tasks" (28 citations), Li introduced a framework that jointly discovers task hierarchies and reusable motion primitives, allowing robots to generalize learned behaviors across diverse scenarios. This approach reduces the need for extensive manual programming or large datasets. His earlier 2017 work, "Learning Object Orientation Constraints and Guiding Constraints for Narrow Passages from One Demonstration" (18 citations), pioneered methods for extracting geometric and kinematic constraints from a single human demonstration, enabling robots to navigate tight spaces and handle orientation-sensitive objects—a critical capability for assembly and manipulation tasks. Li’s contributions are particularly impactful in the context of one-shot and few-shot learning, where his constraint-based techniques have influenced subsequent research in imitation learning and task planning. His work continues to shape how robots efficiently acquire and adapt complex manipulation skills.
Research Focus
Key Achievements
Top Papers
- 1Simultaneous learning of hierarchy and primitives for complex robot tasks28 citations · 2018
- 2