Modeling skills for pushing objects and open various types of doors

2017 Fiscal Year Final Research Report

• Project/Area Number: 16K12502
• Research Category: Grant-in-Aid for Challenging Exploratory Research
• Allocation Type: Multi-year Fund
• Research Field: Intelligent robotics
• Research Institution: Nara Institute of Science and Technology
• Principal Investigator: Takamatsu Jun 奈良先端科学技術大学院大学, 情報科学研究科, 准教授 (90510884)
• Research Collaborator: Yamaguchi Akihiko
• Project Period (FY): 2016-04-01 – 2018-03-31
• Keywords: ヒューマノイドロボット / 動作スキル / 不確かさ / 学習 / 全身動作

Outline of Final Research Achievements

In this research, we achieve 1. the method for modeling the uncertainty of the object to manipulate and 2. the method for generating the whole body motion of a humanoid robot. To achieve the former method, we develop the method to estimate the probabilistic distribution of the uncertainty using neural networks and the method to optimize not only symbolic-level skill types (e.g., push and thrust), but also skill parameters. To achieve the latter, we develop the method to quickly calculate the center of mass and the whole body motion, which satisfy the inputted hand trajectories and force profiles. We achieve that the humanoid robot HRP4 pushes a box.

Free Research Field

ロボティクス