A novel image-based object orientation estimation algorithm for robotic manipulator applications

Abstract

In many robotic manipulator applications, the robot manipulator usually requires knowing the orientation of an object-of-interest to perform grasp planning. This paper presents a novel image-based object orientation estimation algorithm that efficiently and accurately estimates the orientation of a geometric object from vision information only. First, a boundary-point set is defined from the boundary of a geometric object in a segmented binary image. Next, a new objective function is proposed to evaluate the maximum projection length of the boundary-point set onto a closed set with respect to an orientation angle. Finally, the orientation information of the geometric object is estimated via a one-dimensional minimization process. Simulation and experimental results show that the proposed method is able to provide accurate estimates for geometric objects with maximum absolute error less than 0.35°, which is enough to satisfy the requirement on robot grasping control tasks.