Scholarlink

<p>This paper develops a navigation algorithm for self-driving orchard robots that can be used even in situations where GPS signals are poor. RTK GPS can be used to estimate the location of outdoor mobile vehicles and robots, but in orchard environments, location accuracy may be reduced due to trees. To solve this problem, we present a monocular camera-based local localization method that estimates the position and heading of the robots relative to the row center line. Local localization results have inaccuracies due to the absence of accurate distance information. To complement this, we propose a Kalman filter-based global localization algorithm that combines vision, RTK GPS, IMU, and BLDC motor hall sensors. The proposed algorithm under poor GPS signal is compared with the ground truth results that always use RTK GPS signal. Additionally, we developed a path tracking control algorithm based on preview distance. The combined localization and control algorithm verified through experiments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Orchard Navigation Algorithm for Self-Driving Vehicles Under Poor GPS Signal

  • 期刊论文

  • 发表日期：

    2025年4月2日

查看全文
  • Hyobin Suk,
  • Donghun Lee,
  • Minkyu Kim,
  • Sangmuk Lim,
  • Mooncheol Won

摘要

This paper develops a navigation algorithm for self-driving orchard robots that can be used even in situations where GPS signals are poor. RTK GPS can be used to estimate the location of outdoor mobile vehicles and robots, but in orchard environments, location accuracy may be reduced due to trees. To solve this problem, we present a monocular camera-based local localization method that estimates the position and heading of the robots relative to the row center line. Local localization results have inaccuracies due to the absence of accurate distance information. To complement this, we propose a Kalman filter-based global localization algorithm that combines vision, RTK GPS, IMU, and BLDC motor hall sensors. The proposed algorithm under poor GPS signal is compared with the ground truth results that always use RTK GPS signal. Additionally, we developed a path tracking control algorithm based on preview distance. The combined localization and control algorithm verified through experiments.