Abstract:In order to solve problems and to achieve the Autonomous Grinding Robot obtain model of threedimensional (3D) environment, identify the grinding object and ascertain positions and orientations of the end of grind tool by the object in unknown and unstructured environments, a method for realtime measuring positions and orientations of the end of grind tool and leading it based on orthogonal binocular stereovision is presented. The measuring coordinate system was set on laser plane, and the point of intersection from two cameras, whose optical axes were perpendicular to each other, was the reference point that could lead the positions and orientations of the end of grind tool. The final grinding position was ascertained by iteration. And then the positions and postures of the end of grind tool in the image coordinate systems of two cameras could be obtained, by processing the pictures of the end of grind tool could be acquired from the information fusion of positions and postures of the end of grind tool in two image coordinate systems according to the model of orthogonal binocular stereovision.Experimental result shows that the positions and orientations of the end of grind tool can be fed back and leaded by orthogonal binocular stereovision consisting of two CMOS cameras with Eye-in-Hand, and it can offer the solution that grind the surface with robots guided by machine vision.