Vision-Controlled Robot Using Hand Pose Estimation without Marker Loading Requirement
Kiyoshi Hoshino · Applied Mechanics and Materials · 2012
The author proposes a visual-servoing and vision-controlled robot. It claims no sensors installed or special control means used, instead of that a high-precision and high-speed 3D hand pose estimation permits real time operation with two cameras installed at positions of loosely orthogonal relationship, using one PC of the normal specifications. Two cameras have their own database. Once sequential hand images are recorded with these two high-speed cameras, the system first selects one database with bigger size of hand region in each recorded image. Second, a coarse screening is carried out according to the proportional information on the hand image which roughly correspond to wrist rotation, or thumb or finger extension. Third, a detailed search is performed for similarity among the selected candidates. The estimated results are transmitted to a robot so that the same motions of an operator is reconstructed in the robot without time delay.