Uploaded May 2018 | Updated September 2026, 2 weeks ago
ICRA 2018 Spotlight Video
Interactive Session Thu PM Pod F.3
Authors: Subramani, Guru; Gleicher, Michael; Zinn, Michael
Title: Recognizing Geometric Constraints in Human Demonstrations Using Force and Position Signals
Abstract:
Identifying geometric constraints in human demonstrations is useful for programming by demonstration. This paper introduces a method for recognizing constraints from position and force measurements of a human demonstration. Our key idea is that position information alone is insufficient to determine that a constraint is active. Therefore, we must consider the reaction forces to correctly distinguish constraints from movements that just happen to follow a particular geometric shape. Our techniques distinguish plane, arc, and line constraints by using force and position information together. Our method uses the principle of virtual work to determine reaction forces from force and position data. Our approach fits geometric constraints locally using reaction force and position information and clusters these over the whole motion for global constraint recognition. We validate our approach with experiments that confirm its ability to identify constraints in demonstrations with significantly better precision than a similar position only technique.
ICRA 2018 Spotlight Video
Interactive Session Thu PM Pod F.3
Authors: Subramani, Guru; Gleicher, Michael; Zinn, Michael
Title: Recognizing Geometric Constraints in Human Demonstrations Using Force and Position Signals
Abstract:
Identifying geometric constraints in human demonstrations is useful for programming by demonstration. This paper introduces a method for recognizing constraints from position and force measurements of a human demonstration. Our key idea is that position information alone is insufficient to determine that a constraint is active. Therefore, we must consider the reaction forces to correctly distinguish constraints from movements that just happen to follow a particular geometric shape. Our techniques distinguish plane, arc, and line constraints by using force and position information together. Our method uses the principle of virtual work to determine reaction forces from force and position data. Our approach fits geometric constraints locally using reaction force and position information and clusters these over the whole motion for global constraint recognition. We validate our approach with experiments that confirm its ability to identify constraints in demonstrations with significantly better precision than a similar position only technique.
