Safe Robotic Manipulation with Geometrical Constraint of Objects

2019 Fiscal Year Final Research Report

• Project/Area Number: 17H04699
• Research Category: Grant-in-Aid for Young Scientists (A)
• Allocation Type: Single-year Grants
• Research Field: Intelligent robotics
• Research Institution: Sasebo National College of Technology
• Principal Investigator: Makita Satoshi 佐世保工業高等専門学校, 電子制御工学科, 准教授 (60580868)
• Project Period (FY): 2017-04-01 – 2020-03-31
• Keywords: マニピュレーション / 把持 / ケージング / 物体操作 / 動作計画 / 物体認識 / グラスピング / 幾何学的拘束

Outline of Final Research Achievements

This study presents a strategy of robotic manipulation based on caging, geometrical constraints for object grasping and manipulation. The objective of the study is to execute manipulation tasks according to the strategy with higher reliability. First we acquire three dimensional images of objects to recognize geometrical features of the objects for caging constraint. In this study, we focus on loops or holes of objects such as a handle of mug and detect the features to plan robot motion of caging. Second we propose an evaluation index of caging constraints, which is available to evaluate the quality of object constraint by both mechanical and geometrical effects simultaneously. The index is calculated as robustness of objects in grasping and manipulation. The robustness of objects can be equivalent to difficulty of movement of objects, that is, higher robustness ensures to complete firm robotic tasks.

Free Research Field

知能ロボティクス

Academic Significance and Societal Importance of the Research Achievements

本研究課題で提案する，幾何学的な囲い込み・閉じ込めを利用する物体操作計画の学術的な意義の一つは，力学条件と幾何学的条件を統一的に扱う，物理的整合性のある評価指標を新たに提案する点にある．これまで力学条件の不確定要素，例えば接触力の変動などに対して高度なセンシングや制御で対応する事例は少なくないが，現実の運用上，それらを有効活用できるハードウェアは現時点では多くない．力学条件に追加する形で幾何学的条件を加味することで，システム構成を変えることなく物体操作の安心感を高めようとするところが本研究の独創的な点である．これにより，ロボットによる作業のさらなる自動化，また家庭内作業への応用が期待できる．