Software was developed for a Fujitsu HOAP-2 humanoid robot to demonstrate that such "human form" robots can be used to perform useful tasks in an unknown environment; specifically, that they can operate with or in the place of humans to carry out construction operations necessary for space exploration, such as building or maintaining planetary surface habitats for humans. This paper describes the autonomous vision and hand-eye coordination software that enables the HOAP-2 to locate and manipulate objects. Simple vision and pattern detection algorithms including color filtering, noise filtering, image segmentation, orientation filters, and a feedback controlled tracking system enable the robot to locate marked building materials in the workspace. Experimentally determined spatial mappings associate locations in the visual field with the arm joint angles and trajectories necessary to reach these locations, allowing the robot to reach for and manipulate the building materials. These capabilities, when combined with locomotion capabilities, enable the robot to operate autonomously in an unknown workspace.
Vision and "Hand-Eye" Coordination for the Fujitsu HOAP-2 Humanoid Robot
2005-08-01
Aufsatz (Konferenz)
Keine Angabe
Englisch
The Mechanisms in a Humanoid Robot Hand
British Library Online Contents | 1997
|Robot Hand-Eye Coordination Based on Stereo Vision
British Library Online Contents | 1995
|