Scenario controller for daily assistive humanoid using visual verification, task planning and situation reasoning
Kei Okada, Tokutsu Satoru, Ogura Takashi, Kojima Mitsuharu, Yuto Mori, Maki Toshiaki, Inaba Masayuki · IOS Press eBooks · 2008
This paper describes scenario description and control method for a daily assistive humanoid robot. High level representation of a task scenario help user to develop various daily assistive tasks of a humanoid robot. Four key techniques are presented. 1) Three layered architecture integrating a vision based behavior control layer, a task planning layer, a situation based scenario generation layer. 2) Visual verification system for adapting symbol level description into real-world situation. 3) Task level planner providing high-level action primitives for scenario description, 4) Vision and auditory based situation interpretation for controlling the scenario. Finally, we demonstrated a kitchen assistive task by a humanoid robot.