Visual guidance of robot motion

Future robots are expected to cooperate with humans in daily activities. Efficient cooperation requires new techniques for transferring human skills to robots. In this paper, we present a system that provides a robot with the ability to imitate a motion by observing how a human instructor conducts it. The system consists of the following modules: data acquisition, feature extraction and tracking, motion extraction, motion representation, and trajectory generation. Two cameras are used as visual sensors to capture image sequences of a moving rigid object. Feature extraction and tracking are first performed in the image sequences in order to extract 3D motion of the rigid object 3D motion parameters are then calculated from a set of matched 3D points on the moving object. Once the motion trajectory of the moving object is available, its corresponding characteristics are extracted and transferred to the robot. We use a parametric cubic B-spline curve to represent the motion data since characteristics of a parametric cubic B-spline curve are completely determined by its control points, which can be calculated from given measurement data points using the least-squares method. Given the robot configuration, a motion trajectory can then be generated from the obtained control points for the robot to replicate the given motion in its own workspace with the required smoothness and characteristics. Experimental results show that a robot with such a system can be actually taught to imitate a motion and a robot arm has successfully replicated a human arm movement with similar trajectory characteristics.