The intelligent grasping robot can efficiently and reliably perform various handling tasks, reducing the labor intensity of the staff, and accurate object positioning was the basis for the robot to perform the handling task. A robotic grasping system was studied by using Kinect sensor. The robotic grasping system consisted of three subsystems (object detection system, object positioning system and robot motion system). The image information of the object acquired by Kinect sensor was firstly used to train a single multibox detection (SSD) model, and then the object’s category was detected according to the SSD model, the border of the object in the image and pixel coordinate and depth value of the border were obtained. In order to obtain the mapping between the coordinate information of the object in the threedimensional space and the pixel information in the image, the camera can be calibrated by ZHANG Zhengyou calibration method to obtain the intrinsic parameters, extrinsic parameters and distortion parameters. The robot system and the vision system were connected by the Kinect camera handeye calibration, and then the threedimensional coordinates of the object in the robot base coordinate system were obtained by coordinate transformation to realize the object positioning. Finally, the robot inverse kinematics was introduced to solve the joint angle, and the robot motion was driven to complete the grasping and carrying task. Object’s positioning and conveying experiments were conducted. The mean absolute error of the object in x and y directions was 5.2mm and 2.8mm, respectively. The mean absolute error of object height was 4.5mm. The success rate of the object grabbing and carrying experiment was 97%. The experimental results showed that the robotic grasping system proposed was valid to perform object detection, object positioning and conveying task.