Manipulator Control Using Human Arm Imitation

The use of serial manipulators in industries has seen a many-fold Increase in the recent years. They are mainly used if the task requires very high precision or involves lifting very heavy objects. These manipulators are either controlled by very complex control panels (which requires a permanent skilled controller to operate) or are automated using Image Processing (IP) and other sensor techniques which have to be run under controlled environment and have many drawbacks and technological limitations. We are attempting to build a Universal Controller for 5-7 DOF manipulators using 3D Image Processing captured from a Microsoft Kinect. We have proposed a new algorithm to calculate the joint angles characterizing the motion of human arm using analytical geometric calculations. We also created a platform to define mappings between Human Arm Motion and Robot Motion. The angles obtained from human arm motion (observed by a Kinect) are then transformed using the pre-defined mappings and fed to the robot. The robot thus imitates human motion and can now be very easily controlled without much training.