Hand-eye 3D pose estimation for a drawing robot

Abstract

In order to draw pictures on a plane, it is important for a robot arm to know the pose of the end-effecter relative to the world frame which is settled on the workplace. Visual servoing is frequently used in the hand-eye configuration of robot arms. However, 3D pose is preferred for a drawing robot. It is not only required to obtain the three dimensional position of the end effecter so that the pen can touch the paper without drilling a hole, but also required to get the direction so that the pen is always perpendicular to the drawing plane.In this paper, we proposed a monocular camera vision system for a 6DOF drawing robotic arm to estimate 3D pose of the end effecter robustly. First, a coplanar asymmetric polygonal landmark, which is a rectangle with a corner cut off to remove rotation ambiguity, is introduced. Second, the corner points of the landmark are indirectly obtained by intersection of the edges obtained using RANSAC. After that, corner points are filtered by KALMAN filter to reduce detection errors. Third, the end effecter's three dimensional pose relative to the world frame is estimated through PnP algorithm and the kinematics of robotics arm.Experiments show that the 3D vision system has been successfully applied in our robotic arm to draw a circle. The stability and robustness of hand-eye is improved in real time drawing usage. Index Terms -Hand-eye vision, Pose estimation, Perspectiven-point, KALMAN filter.

An Improved Unscented Kalman Filtering Combined with Feature Triangle for Head Position Tracking

“…Sultan M.S. [28] combines robot kinematics, vision, and force sensors to independently perform hand-eye coordination tasks, and the proposed method accurately determines the camera pose of single-view objects. The corners are indirectly detected from the intersection of lines, and the lines are obtained using RANSAC (Random Sample Consumus) algorithm.…”

Parallel vision-based pose estimation for non-cooperative spacecraft

“…According to the most rigid objects' feature on the ground with the mutually perpendicular or nearly perpendicular planar structures, Lv et al 12 adopted the normals of planar areas on the target surface to estimate the 3D pose of the target with ladar range image. Sultan et al 13 proposed a monocular camera vision system for a 6-degree of freedom (DOF) drawing robotic arm to estimate 3D pose of the end effecter robustly. Kirac et al 14 had demonstrated an implementation of regression forests for estimation of the articulated 3D pose of the human hand.…”

Three-dimensional pose estimation model for object with complex surface

“…Dan Lv et al 2 achieve a 3D pose estimation model for the rigid objects on the ground to recognize the military vehicles automatically. Malik Saad Sultan et al 3 propose a monocular camera vision system for a 6-degree-of-freedom (DOF) drawing robotic arm by estimating the 3D pose of the end effecter robustly. Furkan Kirac et al 4 detect the 3D pose of hand gesture from single frame depth data to realize the humancomputer interaction.…”