Research Project
Demonstration-driven Process Knowledge Translation
This research focuses on demonstration-driven process knowledge translation for intelligent assembly tasks. The proposed system converts human operation demonstrations into robot-executable assembly instructions. By integrating visual perception, action understanding, atomic action parsing, and modular low-level skills, the system automatically identifies atomic actions, operation sequences, and key task parameters in assembly procedures, and translates them into executable robot control workflows, thereby reducing the need for manual programming and repeated tuning in conventional robot deployment.