Design Of Intelligent Vehicle Based On ROS System

The rise of the Internet and artificial intelligence has accelerated the development of intelligent vehicles.The intelligent vehicle is more and more widely used in the fields of warehouse logistics,industrial manufacturing,security and so on.The application of intelligent vehicle in the industrial field not only improves the work efficiency,but also greatly reduces the cost of the transportation industry and manufacturing industry.It also has a wide range of prospects and high practical application value for people who work in the toxic and harmful environment.The design of intelligent vehicle adopts ROS system,with simultaneous localization and mapping system and visual tracking system,which can replace people to work in complex environments.First of all,based on the analysis of the common chassis structure of the intelligent vehicle,the differential four-wheel steering structure is used to design the chassis structure of intelligent vehicle,considering the cost,reliability,development difficulty and other factors of intelligent vehicle design.Then the hardware system of the intelligent vehicle is designed,which consists of two parts: the embedded system of the host computer and the chassis motion control system of the slave computer.For the host computer's embedded system,Raspberry Pi 3B + is used,and it is equipped with a Kinect depth camera and lidar;for the slave computer's chassis motion control system,the STM32F103 control system is used.Then,based on the analysis of the basic characteristics,communication mechanism and file system of ROS,the Ubuntu mate 16.04 and the corresponding ROS kinetic system are built in the embedded system raspberry PI 3B + of the host computer of the intelligent vehicle.Based on the platform,the ROS software architecture of the intelligent vehicle is established,and the software of the main function modules of the intelligent vehicle is designed in the way of independent nodes.In the aspect of chassis motion control node design,through the establishment of serial communication protocol between the host computer ROS system and the slave computer STM32,the function of chassis motioncontrol is realized.In the design of simultaneous localization and mapping node and visual tracking node,the software of main function modules of intelligent vehicle is designed.In addition,through the installation of Kinect depth camera and laser radar driver,the external sensor nodes are designed.Secondly,the design of the intelligent vehicle simultaneous localization and mapping system.First,the SLAM algorithm is introduced.Due to the high requirement of sensor data processing in SLAM algorithm,the kinematics model of differential four-wheel steering chassis is established,based on the kinematic model,the odometer model of the system is established and the odometer is calibrated.After the front-end of the laser is registered,the map-optimized SLAM algorithm is used to design the simultaneous localization and mapping system of the intelligent vehicle.Finally,a 4.8m × 3.5m experimental environment is built to carry out the map building experiment and autonomous navigation experiment of the intelligent vehicle.The experimental results show that the intelligent vehicle can build a map in a strange environment,and can carry out autonomous navigation to avoid obstacles.Finally,the intelligent vehicle vision tracking system is designed.First,the KCF tracking algorithm is introduced,and then a vision tracking system based on KCF algorithm is established.By using the diagonalization property of circulant matrix in Fourier space,the matrix operation is transformed into the dot multiplication of the elements,the calculation amount is reduced and the operation speed is improved,which can meet the real-time requirements and realize the vision following function of intelligent vehicle.Through the experiment and data analysis,this mobile intelligent vehicle can better achieve the visual tracking function and complete the tracking of the specified target.