Design And Implementation Of Heavy-duty AGV System

With the rapid development of the economy,the cost of labor has been risen and laborbased logistics transportation has gradually become the lowest-efficiency and highest-cost transportation method.While as a representative of modern and automated transportation technology,Automatic Guided Vehicle(AGV)has promoted the application of artificial intelligence technology in the transportation field and has become an inevitable choice for the sustainable development of China’s manufacturing industry and logistics transportation industry.Specifically,heavy-duty vehicle AGV is large in size and high in load,which can complete the task of carrying large objects.Firstly,according to the actual models and requirements,this paper designs the hardware scheme of heavy-duty AGV system.The kinds and characteristics of obstacle avoidance sensors are analyzed,and the millimeter wave radar which can work "all weather and all day" is selected as the obstacle avoidance sensor.The advantages and disadvantages of different AGV navigation modes are discussed.Combined with the application scenarios of heavy-duty AGV indoor workshop and outdoor road,the hybrid navigation mode of magnetic sensor and visual sensor is selected.According to the requirements of heavy-duty AGV transportation management system,choose the model of GPS,absolute value encoder and surveillance camera.Secondly,the realization of heavy-duty AGV software system is studied in detail,including the hybrid navigation algorithm based on visual navigation algorithm and magnetic navigation algorithm,the safe obstacle avoidance algorithm based on millimeter wave radar,and the path planning algorithm based on transportation management system.For the visual navigation algorithm,the basic structure of the convolutional neural network and the optimization algorithm of the loss function are analyzed,and the training process of the heavyduty AGV corner model is studied to complete the processes of image acquisition,image training and simulation test.As for the magnetic navigation algorithm,it is divided into three steps: data processing of magnetic sensor,vehicle pose calculation and PID controller.Incremental PID is applied to vehicle rectifying control,and it is improved by limiting,rounding and integral separation.While improving the rectifying efficiency,it conforms to the actual situation of vehicle control.For the transportation management system,the route planning,information display and remote control functions of vehicle transportation are completed by analyzing the path planning algorithm and analyzing the sensor information protocol.Finally,the AGV system is implemented on the heavy-duty Dongfeng truck platform to complete the performance test of the heavy-duty AGV in this paper.To be specific,the vehicle driving function,autonomous navigation function and obstacle avoidance function are tested,meanwhile,the no-load test and loading test of the overall system are completed.A series of test results show that the system designed in this paper can achieve autonomous navigation and safety obstacle avoidance functions,and meet the application requirements of port logistics transportation.