DeepNet-Based 3D Visual Servoing Robotic Manipulation

The fourth industrial revolution (industry 4.0) demands high-autonomy and intelligence robotic manipulators. goal is to accomplish autonomous manipulation tasks without human interventions. However, visual pose estimation of target object in 3D space one the critical challenges for robot-object interaction. Incorporating estimated into an control scheme another challenge. In this paper, a deep-ConvNet algorithm developed estimation. Then, it integrated servoing achieve long-range mobile task using single camera setup. proposed system integrates (1) training only synthetic images, (2) 6DOF as sensing feedback, (3) control. consists two main steps. First, perception network trains on datasets then efficiently generalizes real-life environment postrefinements. Second, execution step takes generate continuous translational orientational joint velocities. has been experimentally verified discussed Husky base UR5 manipulator. Experimental findings from simulations real-world settings showed efficiency task.