Temporal Estimates of Maize Plant Growth in a Breeding Program Using Ground Based and Unmanned Aerial Vehicle Systems.

Plant height is generally measured manually using a measuring stick at the end of the season before harvest (traditionally the inflorescence or flag leaf at maturity in maize), which provides only a terminal growth estimate. Yet, terminal plant height is of practical importance for subtropical maize as this trait shows 40 to 60% correlation with yield. Measuring plant height manually is labor intensive, making it infeasible to collect in-seasons measures of growth within a field breeding program; however, sensor technologies deployed on ground and unmanned aerial vehicles (UAVs) can now make temporal estimates of growth and height practical throughout the entire growing season. We developed and deployed a novel ground platform that can clear mature maize fit with ultrasonic and other sensors, fixed wing and copter UAVs with imaging sensors allowing height to be estimated using structure from motion, and collected manual height measurements at three locations temporally across growth stages. These locations had an overlapping set of 16 to 35 diverse hybrids replicated multiple times in one, two and four row plots. This pilot study was initiated to compare detection ability and practicality of different technology platforms and identify challenges in routinely implementing this data into a breeding program.  We observed the repeatability of sensor-measured plant height increased as the crop grew (likely due to increased differences between genotypes), noted different growth trajectories of the different genotypes, and found good correlation between structure from motion and manual measurements.