Update:
The critical first step for Drone Spot Spraying is to identify and locate weeds on crop fields. One of the Case Studies from the 2018 MSB Grant showed that overhead weed detection and identification over mature Soybean fields is possible. On August 11th, a Soybean Field in Caroline County was used to evaluate overhead weed detection and identification. In one area of the Soybean field there were Palmer Amaranth and Redroot Pigweeds growing together. The test site picture shows Palmer and Redroot weeds growing about 4 feet apart.
An aerial drone image from 125 ft altitude showed the Palmer Amaranth and Red Root Pig weeds imbedded in the soybean canopy. An expanded view of the weeds showed that they are distinguishable by color, leaf pattern and other discriminators. A close-up at 25 foot helps to verify their distinguishing characteristics. Through various types of analysis, these types of weeds can be detected and, in some cases, identified. In the future, more advanced sensors and image processing will improve this capability.
Drone drop testing was used to simulate precision spraying accuracy. An initial scout drone was used to locate several test targets. The coordinates for the targets were then given to a second drone that hovered at 50 feet over each target and took pictures to designate where a spot spray discharge would have hit the ground. The target error of the simulated spray drone to the scout drone location was less than 7 feet. Better calibration between the two drones might provide 4-foot accuracy. However, wind and other conditions most likely will be the limiting factor for spot spray accuracy.
Simulated spot spray target testing was conducted at Kinder Park in Severna Park Maryland on January 10th, 2019. Test target and drop errors appeared to be biased in one general direction, therefore, calibration should be able to reduce the error. If needed Ground Control Points (GCP) or RTK techniques could be used to improve drop accuracy. Further investigation needs to be done between weed detection and Spray accuracy.
Drone analysis can evaluate seedling emergence maps. For corn this provides an early estimate of yield. There are several techniques for this with varying degree of accuracy and cost. Two applications were investigated. The first was a strong, health corn field with an estimated 97% emergence. In spring of 2018 many farms had severe seedling damage due to heavy rain. A second field with severe crop damage was evaluated. Damage assessment and recommended re-seed maps were provided. The damaged field was re-seeded with a strong recovery. This type of analysis should be of value to both farmers and insurance companies.
On mid-May drone measurements were made on a corn field which had a strong seedling germination. The resulting maps showed almost uniform emergence with an estimated 29,760 plants per acre. The crop stand maps showed several small areas that had missing plants and others where there was no emergence. This strong field was used as a baseline drone analysis which showed a crop stand of ~98% over the 10 acres tested.
There was severe rain during the spring of 2018 and many freshly planted seeds were damaged. Drone measurement and analysis were done to assess the damage to corn crops. The drone maps showed a seedling emergence of ~ 47% over the 7.5 acres tested. Drone maps were used to recommend a reseeding plan. In late July, drone overhead and ground measurements showed that there had been a significant recovery in the crop stand after reseeding.
View uploaded report 
The Precision Drone Technology for Improved Crop Yield, Input and Environmental Impact project tested various capabilities of agricultural drones, including weed detection and identification. Drones were able to successfully detect and identify Palmer Amaranth and Redroot Pigweed growing in a field of mature soybeans. The two weed species were distinguishable by color, leaf pattern and other traits on drone-captured images. Information on location of weeds was successfully handed off to a simulated spray drone and tests were conducted to determine the accuracy of simulated spray drops on targeted locations. Target error between scouting drone and spray drone was less than 7 feet and this accuracy could likely be improved to 4 feet with calibration.
Drone capabilities were also tested for crop yield estimates and crop damage estimates. These tests were performed on corn fields. In mid-May, a 10-acre corn field was assessed using a drone to determine crop emergence. The drone images were used to determine that the crop had almost uniform emergence (97% emergence) and an estimated 29,760 plants per acre. Crop maps developed from drone images were able to show small areas where plants were either missing or had no emergence. Severe rain during the spring of 2018 damaged many freshly planted seeds. Drone measurement and subsequent analysis of an affected corn field on June 21st to assess seedling damage showed a seedling emergence of ~ 47% over the 7.5 acres tested. Maps developed from the drone flights were used to recommend a reseeding plan. The field was reseeded accordingly and a follow-up flight and ground measurements in late July showed a significant recovery in the crop stand after reseeding.
There are many ag drone capabilities that can be implemented today to improved Maryland crop farming productivity. Based on this investigation, drone-based seedling emergence analysis can be valuable to farmers in assessing crop yield. It can also be used to evaluate crop damage with recommended re-seeding maps. Agricultural precision drone technology will have a significant impact on crop productivity over the next several years. Based on this project research, drone-based precision spraying appears to be viable and should be available to Maryland crop farmers by 2022. This new capability will reduce chemicals and crop damage while also reducing environmental impact.