Robot Object Tracking Algorith M Based On Monocular

Object tracking has always been an important research direction in the field of computer vision.In recent years,with the rapid development of deep learning,object tracking algorithms based on deep learning have continuously improved in performance,and have gradually replaced traditional object tracking algorithms.However,because the complex structure of deep learning will affect the real-time performance of the tracking algorithm,the object tracking algorithm based on deep learning has not been well applied in mobile robots.It was not until the emergence of the object tracking algorithm based on Siamese Network in recent years.It provides the possibility for mobile robots to deploy object tracking algorithms based on deep learning.In this paper,using Jetson Nano as the hardware platform,a lightweight convolutional neural network is designed,combined with the Siamese Network,and the object tracking algorithm based on deep learning is applied to the mobile robot.The main work of this paper is as follows:1)Aiming at the problem of weak computing power of mobile hardware,an object tracking algorithm based on deep learning that can be run in real time is designed.The Roofline model is used to calculate the performance of AlexNet,ResNetl8,MobileNetV2 and SqueezeNet,and based on this,a lightweight convolutional neural network based on the Fire module is designed,and the lightweight network is used as the backbone of the Siamese Network,combined with the Regional Proposal Network(RPN)to construct an object tracking algorithm.In order to improve the speed of the object tracking algorithm,TensorRT is used to accelerate the algorithm implementation.The experimental results show that the lightweight convolutional neural network and object tracking algorithm designed in this paper have greatly improved in speed,and the object tracking algorithm can reach up to 18 frames per second on the Jetson Nano platform.2)In order to deploy the object tracking algorithm on mobile hardware,the robot hardware system is designed,and the robot object tracking software system is designed and implemented around the hardware system.The hardware system uses Jetson Nano development board as the core,combined with vision sensors,four-wheel mobile chassis,wireless modules,etc.to build a robot hardware system,which can realize forward and turn by controlling the speed of the motors on both sides of the robot.The robot object tracking software system based on the Robot Operating System(ROS),and designs object tracking nodes and motion control nodes for the robot.The object tracking node is responsible for processing the image data and calculating the object position,and the motion control node calculates the rotation speed of the motors on both sides according to the object tracking result.The experimental results show that the mobile robot can achieve the function of object tracking.