Design And Implementation Of Express Delivery Driverless Car In Community Based On ROS

Based on the ROS system and automatic driving technology,this paper designs and implements a driverless vehicle to solve the problem of express delivery in the community.According to hardware trolley platform made by self,it has the ability of mapping and locating,self-navigation,etc.,and uses the mechanical arm to realize the function of loading/unloading articles from the trolley.Then,through simulation modeling,the key links of the system are simulated and tested in the simulation environment.Finally,the performance of each key link and the experimental effect of the whole system are verified and tested on the self-made unmanned vehicle,which basically achieves an ideal effect.The main work of this paper is as follows:(1)The self-made refitted trolley platform is used to analyze and model the kinematics and dynamics of the mechanical arm device,control its movement based on the ROS system and embedded system,and independently calibrate and jointly calibrate the binocular camera and 2D radar respectively to obtain the structural parameters of each key device of the trolley.(2)Based on the normal distribution transformation principle of NDT,the automatic mapping and positioning of the trolley are realized.In view of the drift problem that is easy to occur in the automatic mapping process,this paper uses the loopback detection method to optimize the automatic mapping,so as to improve the stability of the mapping system.In the aspect of vehicle path planning,the vector map containing environmental information is drawn according to the point cloud map,which provides more abundant path reference information for vehicle path planning..(3)The autonomous navigation system is divided into three parts: target detection,routing and control.The sense of sight and point cloud fusion are used to sense the ambience,this means the target segmentation network and point cloud clustering method are integrated to get obstacles,and the Kalman filter is used to track obstacles.The global path is planned according to the vector map,and the local path is planned according to the obstacle information obtained from the global path and target detection.Finally,according to the state decision and the pure pursuit algorithm,the autonomous driving function of the car is satisfied.(4)The corresponding test experiments are designed for the functions of each part of the trolley system to verify and analyze the result of the surveying and position system in different circumstances,the accuracy and integrity of the navigation and obstacle avoidance of the autonomous navigation system,and the accuracy of the loading/unloading functions of the robotic arm device.The unmanned vehicle system designed and implemented in this paper can realize the functions of mapping and positioning,autonomous navigation and loading/unloading goods in a real-time environment.After system testing,the key performance of the car has basically achieved a relatively ideal effect.