Smart agriculture [1-10] is becoming more and more popular, from monitoring systems in greenhouses to agricultural robots. Today, the application of robots to agriculture is not unfamiliar, particularly with the development of AI, that we can create robots that automatically harvest agricultural products . Some parts of the world focus on NIR cameras to process identification under models RCNN, Faster R-CNN, VOC ... in this article, we mention the use of Yolo v3 to identify strawberries by Real sense D435i 3D camera, then through the evaluation function to select a strawberry to locate the strawberry in the space for easy harvest robot, to identify the strawberry in the space we combine the 3 direction property The axis in the IMU has built-in camera and 3D image space, the harvesting process is a dynamic process, so we will use anti-noise filters, target orientation, IMU sensor noise and follow the trajectory. transfer, ensuring the picking mechanism does not occur instability when identification error and sensor impact on the system. Where the anti-noise filter for IMU sensor status is extended Kalman and for the motion process is a smooth line incorporating gauges to eliminate peak noise. In this section, we only focus on the application of strawberry identification and identification in space; In the next direction, we will evaluate the SLAM combination robot system for self-propelled problem.