LEADERS IN INNOVATIVE TECHNOLOGY SOLUTIONS
Facebook LinkedIn YouTube
4-D Mapping for Crop Monitoring

Researchers at the Georgia Tech Research Institute and the University of Georgia are using 4D mapping techniques to develop models of entire crop fields. The task of assessing crops and collecting field samples is inherently co-robotic. Whereas the actual collecting of the sample may be automated, the decision regarding which plants to sample is best left to the grower or crop consultant.

 

To accomplish autonomous sampling in a field, a 4D model of the crop (3D model with an additional time dimension) is developed based on weekly in-field imaging, using the Georgia Tech Smoothing and Mapping library. This model monitors variables such as plant height, leaf area index, canopy coverage, plant discoloration (chlorosis/yellowing and necrosis/browning), and their rate of change. These and other physical measures are critical for identifying plants under biotic (pest organisms, plant pathogens, and weeds) or abiotic (inadequate moisture and nutrient levels) stresses.

 

The 4D model of the crop can then be reviewed by the grower to specify in which areas of the field leaf or soil samples are to be collected by a robotic system, such as the UGA rover with integrated Georgia Tech robotic arm (see 3D-Printed Robot Arm for Crop Inspection).

 

The team has completed an initial feasibility study of a broccoli crop at UGA’s agricultural research farm in Tifton, Georgia (see photos on right).

 

Project Contact: Gary McMurray

 

 

4-D Mapping for Crop Monitoring

 

4-D Mapping for Crop Monitoring

 

4-D Mapping for Crop Monitoring

Preliminary 3D reconstruction results on a broccoli crop, where image data sets were collected weekly. Top: original image from first week of study; Middle: result of running 5-point algorithm to establish correspondence across frames; Bottom: preliminary monocular 3-D reconstruction result from 70 successive images (in week 9), where scale is obtained from GPS.