The Design And Implementation Of An Intelligent Assisted Driving System For Railed Minecart

In traditional railed minecart transportation,manual supervision and control are necessary,which is labor-intensive and inefficient.Besides,it might also cause some serious accidents due to negligence of staffs.Therefore,developing a assisted driving systems for railed minecart can reduce consumption of labor force and risk of accidents caused by negligence.This thesis aims to design and carry out an intelligent assisted driving system for railed minecarts.According to the actual mine,this system uses sensors and imaging equipment on the running vehicle.Using wireless network in the mine,it shows the real-time road conditions in front of the railed minecart on a monitor outside the well.The operator can know the vehicle's running status in time through the feedback of the display and manage the overall system through a series of operation modes.It can also give the management authority to the system itself.So,the system can take over all the control process automatically.This system contains the following modules: Data Acquisition Module uses the industrial computer's concurrency and processing capabilities to timely sort,filter,and package all sensor data.Network Transmission Module is mainly responsible for data transmission and instruction communication of each part,network status detection and network disconnection processing in emergency.Peripheral Control Module,under the encouragement of some external instructions,controls external linkage equipment to ensure the safety and reliability of the transportation process.Display and Processing Module shows the information gathered by the system to the driver.It is also responsible for forwarding and performing related operations of the driver.User Permission Module is responsible for assigning permissions to system resources.Equipment Management Module is mainly responsible for the loading and unloading management of the moving unit in the system.To facilitate the driver's understanding of the environment and road conditions ahead,this system provides real-time identification of obstacles.Identifying obstacles that may appear in the running mine could warn the operator.Using a mature and reliable engineering target detection algorithm to achieve this function is important.Because of the high needs for realtime performance and small target detection,we use the currently more suitable YOLOv3 target detection algorithm for target recognition.Similarly,combined with the particularity of the environment in the mine and the particularity of identifying obstacles,after targeted collection and processing,we have trained the training set on the darknet framework.In the simulated environment,some progress has been made to recognize specific environments and special obstacles.We have finished development of this system.It is under Waiting Status for permission of the Mining Products Safety Approval and Certification Center.According to the requirements of safety production in coal mine,any products are not allowed to be applied in mines without the permission.So,we built a simulated environment in a corridor with similar scenes to test the system.The usability and stability have reached design standards.