Automated root mapping to enhance field trial evaluation of citrus rootstocks in the HLB era

Automated root mapping to enhance field trial evaluation of citrus rootstocks in the HLB era

Report Date: 12/23/2020
Project: 18-033C   Year: 2020
Category: Horticultural & Management
Author: Yiannis Ampatzidis
Sponsor: Citrus Research and Development Foundation

In the quarter 8, we evaluated the performance of the automated (remote-controlled) GPR system in the field, and compared it with the manual scanning procedure. In summary, the advantages of the remote-controlled system over the manual process are: (i) lesser time taken for scanning, (ii) more precise movement, and (iii) less human effort required. The experiments were conducted using a 1,600 MHz GPR antenna in 5 different trees in a citrus grove located at SWFREC. The time taken to perform one scanning circle was reduced by approximately 3 times when using a remote-controlled system (compared to a manual process).  Materials and EquipmentA ground penetrating radar (GPR) control unit was mounted on an aluminum alloy chassis and equipped with a 1,600 MHz antenna. The aluminum alloy chassis was moved by a vehicle, which is powered with two 12V DC motors that have an output rate of 300 RPM. The motors were connected to a 30 A dual channel motor driver.  Time taken for one scanning circleFor both the remote-controlled and manual process, the time required for one scanning circle from start to the end was recorded. The detailed results of these experiments are presented in the Final report. In summary, the time taken for one scanning circle can be reduced by approximately 3 times by utilizing the remote-controlled unit. This is due to the fact that in manual process the operator has to manually draw the circle and move around the tree, which takes a lot of time when compared to remote-controlled process. The only effort required by the  remote-controlled process is to hook up either ends of the rod to the tree and the GPR system, and then remotely move the GPR unit around the tree. On an average, it can take up to 30-40 seconds to draw one scanning circle around the tree manually. The scanning circle can often end up in an irregular shape due to the difficulty of reaching under the tree. Depending on the size of the tree, it might be required to perform the scans with 1-3 concentric circles. For example, a tree with an age of 3 years will need 3 scanning circles around it which can take up to 90-120 seconds. All this effort and time taken in the manual process can be avoided by using the remote-controlled system as it has markers for circles of different radius from 1 foot to 3 feet in increments of 0.5 feet.  Effort needed for one scanning circleBefore performing a GPR scan for a tree, the operator might have to trim the low hanging branches which might prevent a person from crouching under the tree and perform the scan manually as it is often quite difficult to get under the tree and perform a complete continuous scan without stopping. It takes a lot of human effort to move around the tree in a crouching position. This can be avoided by using the remote-controlled system as it just needs to be hooked to the tree trunk and no other human effort is needed to make it perform a complete continuous scan. Furthermore, the remote-controlled process requires no stops, which can increase the quality of the collected data and eliminate human errors related to the precise movement of GPR in a circular scan. Comparison of scan line data from remote-controlled and manual processThe scan line data collected from both the remote-controlled and manual process were compared after processing the data using an auto root detection software (Tree Radar Inc., Maryland, USA). The results of root detection for both processes were almost the same and there were no significant differences between them (please see the Final report for a detailed analysis). ConclusionThe experiments show that the remote-controlled process can reduce the application time by 3 times when compared to a manual process. It can also reduce the human effort required and increase the precision of data collection. For both procedures (remote-controlled and manual), initial cleanup is required before data collection to clear the debris and fruit drops under the tree. This can take up to an average of 60 seconds per tree. The presence of irrigation lines close to the trees could also increase data collection time. These lines can be moved to the side, but in some cases the operator has to manually lift up the irrigation lines to allow the scanner to move under it. Further development has to focus on solving this issue.


Your browser does not support pdfs, click here to download the file.