Crop stem width estimation in highly cluttered field environment
We present two methods for estimation of crop stem width on small mobile robots. Stem width is an important phenotype needed by breeders and plant-biologists to measure plant growth, however, its manual measurement is cumbersome, in-accurate, and inefficient. The presented methods use a common image processing core that is designed to extract the foreground in the presence of significant leaf and stem clutter, view of other rows, and varying lighting, from a side-facing USB camera mounted on a small mobile robot. Using the extracted foreground, one approach uses estimates of robot velocity from wheel encoders and dense optical flow to estimate depth, while the other employs filtering of the LIDAR 2-D point cloud to estimate the depth. Both methods have been exhaustively validated against available hand-measurements on biomass sorghum (Sorghum bicolor) in real experimental fields. Experiments indicate that both methods are also applicable to other crops with cylindrical stems without significant modifications. The width estimation match on sorghum is 92.5% (using vision) and 98.2% (using vision and LIDAR) when compared against manual measurements by trained agronomists. Thus, our results clearly establish the feasibility of using small robots for stem-width estimation in realistic field settings. Furthermore, the techniques presented here can be utilized for automating other phenotypic measurements.