Joint GPS,DR And Vision For Navigation Of Unmanned Vehicles

With the development of the times,more and more cars,accompanied by car safety driving problems,environmental pollution problems,traffic congestion and so on.In order to solve these problems,people began to study unmanned vehicles.Unmanned vehicles have broad application prospects in the defense and civilian areas.Autonomous navigation technology is one of the key technologies of unmanned vehicles.Global positioning system(GPS)has global coverage and high positioning accuracy.However,GPS signals is prone to be blocked in urban environments.Dead reckoning(DR)system is a self-service navigation,which does not require external information and is generally not affected by the environment.It can be used to compensate for the lack of GPS navigation,but its positioning error will accumulate over time and can not be used for long time.Visual navigation through the detection of lanes and traffic signs in the road,guiding unmanned vehicles driving autonomously.It is a passive measurement method,which does not emit light and radiation,with a good concealment,rapid and accurate measurement of the advantages.But it’s vulnerable to light,road environment.Any kind of single navigation has its shortcomings,which can not meet the demand of autonomous driving.The integrated navigation can make use of the advantage of various navigation and carry on the complementary advantages,which is an effective method to realize the navigation of unmanned vehicles.This thesis designs a platform of unmanned vehicles,which provides hardware support for navigation realization.The vehicle is equipped with GPS,camera,displacement encoder,electronic compass and other sensing devices.The host computer uses high-performance computer to process the image data and positioning data.The lower machine using microprocessor controls motor and steering gear and processes other sensor data.The thesis divides the system into three parts: environment-aware system,path planning system and bottom-layer control system.The environment-aware system system collect environment data.The path planning system uses two navigation algorithms proposed in this paper to process the perception data and get the control variables.The underlying control system uses the improved PID control algorithm to process the control variables and control the steering and speed of the vehicle.Combined with the hardware of this thesis,a navigation system based on GPS / DR is proposed and implemented.This thesis designs a differential GPS precision test scheme.The test results show that the positioning accuracy of the GPS in this paper can reach the centimeter level which meets the demand of unmanned vehicles.The navigation algorithm obtains the angle between the unmanned vehicle’s heading and the desired heading by obtaining the front and the target points of the planned route,the current positioning and heading of the unmanned vehicle.Then the algorithm passes it to the unmanned vehicle’s steering control section.The navigation algorithm set a threshold of the distance between the vehicle and the target point.When the distance isless than the threshold,the vehicle updating the front and the target points.When the vehicle reaches the end,the navigation algorithm is over.Steering control of the unmanned vehicle adopts the PID control method with a dead zone,and the speed control adopts the incremental PID control method.The two kinds of control algorithms make vehicle driving robustly.In view of the shortcomings of GPS/DR navigation which is susceptible to the occlusion of GPS signals,a vision-based unmanned vehicles navigation is designed and the GPS/DR/vision based unmanned vehicles navigation is obtained.The Canny operator edge detection is used to get the edge image,and the Hough transform is used to extract straight lines.The lane line is extracted based on the parallel and width features of the lane line from straight lines.The image is processed by the image preprocessing method.The centerline between two lane lines is used to represent the desired direction of the unmanned vehicles.The center line of the image represents the current heading of the unmanned vehicles.Then the heading’s deviation angle is obtained and transmitted to the unmanned vehicle’s steering control section.Visual navigation is used when the GPS signal is obstructed which causes GPS unlocated or non-differenced positioning.When the GPS signal is normal,unmanned vehicles use GPS/DR navigation.In order to verify the effectiveness of the navigation algorithm proposed in this paper,a test is designed.The test selected a campus rectangle section.The GPS signal is blocked in a part of the section where the lane line laid.Two kinds of algorithms are tested,namely integrated navigation of unmanned vehicles based on GPS/DR and integrated navigation of unmanned vehicles based on GPS/DR/vision.In addition,according to the safety requirements of unmanned vehicles test,a monitoring station software system is designed to monitor whether the environment-awareness system of unmanned vehicles is running in real time and display the position of unmanned vehicles on Baidu map.The experimental results show that the integrated navigation algorithm based on GPS/DR/vision can make unmanned vehicles travel steadily according to the established route and effectively solve the problem that the autonomous driving of the vehicle is not stable or deviates from the established route when the GPS signal is obstructed.