Hardware-software system for autonomous tractor control with obstacle avoidance capabilities

Authors:

Abstract
BACKGROUND: The development of reliable autonomous agricultural machines requires a comprehensive solution to the problems of navigation, perception of the environment and real-time control. Despite the existing products, the issues of development of a unified hardware-software system that ensures motion and obstacles avoidance in real conditions remain relevant. The aim of the study is to develop such a system for an unmanned tractor. AIM: Development and testing of the hardware-software system for autonomous tractor control that ensures motion along a given path with high accuracy, an all-round view of the environment, distinguishing static and dynamic obstacles, and performing a detour or emergency stop maneuver in accordance with the selected operating mode. METHODS: During the study, a modular software architecture, implemented on an NVIDIA microcomputer, was developed. The hardware package includes a GNSS receiver with RTK correction support (Global Navigation Satellite System with Real-Time Kinematic correction), a computer vision system (4 all-round cameras and a radar), an operator’s tablet console and a wireless communication system. Computer vision algorithms for detecting and distinguishing obstacles, path control algorithms based on the GNSS RTK positional data, and a finite state machine for controlling the operating modes of the system were used for data processing. The functionality was tested with virtual simulation methods. RESULTS: As a result of the development, the hardware-software system that meets all the technical requirements was created. The hardware architecture is specified. The software is implemented as a set of interconnected modules responsible for navigation, motion control, computer vision and operator interface. The developed web interface allows the operator to set routes, select operating modes (“stop” or “detour”) and monitor the system status in real time. Integration tests have confirmed the system’s operability and fault tolerance. CONCLUSION: The hardware-software system that demonstrated the ability to make the tractor move autonomously along a given path and to avoid obstacles during virtual tests has been developed. The key advantages of the solution are modularity and scalability.