An Intelligent Robotic Grasping and Manipulation System with Sensor Fusion

Complex robotic grasping tasks, for example, deformable objects grasping and adaptive grasping, which are close to practical robotic application, remain a challenge due to unknown object geometries. In this paper, an intelligent robotic grasping and manipulation system with sensor fusion that consists of visual-based object detection and 6-D pose estimation, grasping pose estimation, robot control, and safe grasping force framework based on deep learning is proposed. It is designed to guarantee safe grasping by estimating appropriate grasping force. The results show that our system has a good performance on the detection and grasping pose adjustment of trained objects with different texture. The safe grasping force estimation has a better performance on rough texture and limited deformable objects.