Self-Localization and Autonomous Driving with Multiple RGB-D Cameras in Orchard

Abstract

Recently, the declining and aging of farmers in Japan has become more serious year by year. Development of agricultural robots for labor saving and high efficiency in agriculture is underway, but the development of agricultural robots for fruit tree cultivation is delayed due to the diversity of cultivation methods. In this research, we realize autonomous driving of UGV（Unmanned Ground Vehicle）in orchard of V-shaped joint tree. In order to build a low-cost and versatile system, this paper proposes self-localization methods with RGB-D cameras without artificial landmarks. Traveling along the fruit tree row is realized by local self-localization based on the position information of the fruit tree. When turning on headland, the self-localization is realized by the EKF (Extended Kalman Filter) using map matching. The usefulness of this methods is verified by field experiments.