Comparison of Sick and Hokuyo UTM-30LX laser sensors in canopy detection for variable-rate sprayer

Abstract

Accurate acquisition of canopy structures, sizes and shapes of plants are significant in providing technical data for variable-rate precision spraying and plant growth estimation in modern agriculture. This research work proposes a comparative analysis of two mainstream brands of laser sensor scanners for canopy detection. A Hokuyo UTM-30LX laser sensor and a Sick LMS151-10100 laser sensor was used to detect spray targets including two artificial trees and a cuboid foam box, respectively. Two data acquisition and storage algorithms based on C++ language have been developed to collect real-time data from the given targets based on the sensors. A 3-dimentional image reconstruction algorithm was proposed to construct the detection targets using MATLAB software. In this experiment, the detection distances between laser sensors and targets range from 1.8 to 2.2 m, and the traveling speed of laser sensors ranges from 0.5 to 2.0 m/s, and the size of trees 2.15 × 1.24 × 0.70 m3 have been taken for verification of proposed method. The detection accuracies of both laser sensors have been compared by utilizing the given targets under indoor laboratory conditions, and detection accuracies of 3-dimentional reconstruction images of the target are analyzed by the root-mean-square error (RMSE), the coefficient of variation (CV) and edge similarity score (ESS). The experimental results show that the laser sensors have different detection accuracies under the same experimental conditions, and has different detection accuracy under different experimental conditions. The LMS151-10100 laser sensor is more accurate and more suitable for detection in the case of fast detection speed. However, both sensors have the capability to measure the targets accurately and can be applied for the detection of trees in the area of variable-rate precision spraying. Keywords: Agriculture spraying automation, Crops and tree spraying, Lessor sensors, Canopy detection, Orchard