AUTOMATED NAVIGATION FOR UNMANNED GROUND VEHICLES IN LOGISTICS

Abstract

The use of ground-based courier robots is widespread in the systems that integrate the complex automation of transportation management. Most often, the robot's main problems are finding and avoiding obstacles on the way during cargo transportation. Many computing resources are used to model the route taking into account the features of the terrain during logistic navigation. In research, the choice of gear motors, motor drivers, microcontrollers, and microcomputers was justified to ensure movement. To ensure effective autonomous navigation, the robot must contain ToF, LiDAR, ultrasonic sensors, GPS modules, cameras, an accelerometer, and a magnetometer. The selected combination of the investigated hardware components will allow for accurately determining the location of the vehicle. The article analyses the components of the robot. The principle of operation of drivers and engines is defined. Two models of drivers and two models of engines were considered. A schematic of the device with all connected sensors, a controller, and a microcomputer was developed. The principle of interaction of all system components to ensure autonomous navigation was developed. For the long-term operation of the system with autonomous navigation, the power supply system and the capacity of the necessary battery were calculated. A neural network was created and used for object detection. For neural network calculations, a server-side was created that receives frames from the device's camera and returns data on detected objects to the autonomous vehicle.

Keywords

• autonomous navigation
• robotics
• computer vision
• neural networks