| With the rapid development of the global automobile industry and the improvement of people’s purchasing power,the world has witnessed substantial increases in the number of cars.Meanwhile,diverse problems such as traffic congestion,frequent accidents,environmental pollution and resource waste have emerged,leading to a research boom on driverless cars.of Nevertheless,it is difficult for high-level driverless vehicles to fully adapt to the extremely complex environment in consequence of their significant cost.In view of that,the driverless market is very broad under the given path with the specific scenes such as sightseeing in large parks and cargo handling in wharfs and ports.Aiming at this,the author has established the hardware platform of driverless environment perception and vehicle control system and developed the software algorithm of environment perception and vehicle control with the purpose of realizing driverless driving under the specified scene and established path.The paper is mainly divided into three parts: vehicle modification,installation and debugging of sensors and control hardware,design and application of sensing system and vehicle unmanned driving control.This paper designs the overall scheme of driverless vehicle.The author chooses electric vehicles as a modification of the unmanned vehicle through the market research,and completed the perception and control circuit at the beginning.Then,the multiline laser radar,cameras,millimeter wave radar and GPS differential are selected as unmanned perception components to optimize the layout of sensors’ location according to the requirements of the vehicle environment perception.The digital control of vehicle steering system is also achieved by adding electric power steering mechanism(EPS)to the vehicle while the digital control of brake and gas pedal is realized by controlling voltage through DA converter.Environment perception is the premise of autonomous driving of driverless vehicles.This paper designs a image target recognition algorithm based on SSD network,which can be employed for the classification and recognition of target objects such as pedestrians,cars and bicycles in front of vehicles.At the same time,in order to obtain the depth information of the obstacle,the author completed the joint calibration between the lidar and the camera and the distance of the target obstacle was calculated after the information fusion.When the obstacle is within the safe range determined according to the speed of the vehicle,the braking measures shall be taken in time.The millimeter-wave radar is installed around the body of the vehicle to detect the lateral and backward targets of the vehicle.With the purpose of ensuring the vehicle to travel along the predetermined path,the centimeter-level differential GPS is adopted to locate the vehicle in real time.In this paper,a fuzzy control algorithm based on the preview mechanism is designed to make sure that the unmanned vehicle can travel along the fixed route stored in advance.Firstly,a two-degree-of-freedom vehicle linear dynamics model was established,and the pre-sighting deviation was calculated by single-point pre-sighting method.A fuzzy controller was established with the pre-sighting deviation and its change rate as the input,and the electric power steering system using the steering wheel angle as the output control vehicle realized the lateral control of the vehicle.The driverless platform developed based on the above methods has carried out real vehicle experiments around the mechanical laboratory of Beijing Jiaotong University.The results reveal that the driverless platform sensing system can complete the braking within 1.5 seconds when the obstacles enter the warning range,so as to ensure the safety of the driverless vehicle.The control algorithm proposed in this paper can quickly and stably ensure the autonomous driving of the unmanned vehicle on the fixed path.The maximum lateral deviation of the experiment is within 0.7m,which can ensure the vehicle to drive in the current lane. |
