Data engineer Somali Chaterji was recently awarded a grant to research better ways to integrate drones into farming. She is working to incorporate machine learning applications and on-device computation into drones used in agriculture. It’s part of an effort to use drones to increase crop yields. “The use of drones will become more and more important and cost-effective, especially once all these devices are connected with each other, do more autonomous operations, transmit data across each other and to the farmer, and integrate with other robotics on the ground,” Romeo Durscher, vice president of public safety at Auterion, a company which makes operating systems for drones, told Lifewire in an email interview. “The future is indeed in the data and what to make of the data.”
Getting Smarter
Chaterji aims to create a network of small devices to make drone data collection and analysis more sustainable. Under her plan, drones will determine their optimal trajectories, reducing wasted battery power and recharge time. Drones will move around a field, sensing soil and plant conditions to determine and spray the amount of water and nutrients needed. The system will enable the devices to reduce energy consumption using on-device intelligence. “Our innovation distributes the computation, and each device can decide to transmit only the useful quanta of data instead of a giant data deluge,” she said in the news release. “Improved efficiencies like these will benefit the farmers and the environment by reducing the frequency of charging these devices and decreasing the reliance on cloud computation and data centers.” Drones are already used extensively in farming. The flying machines can be used for aerial scouting of crops, Jarrod Miller, an assistant professor with the Plant and Soil Sciences Department at the University of Delaware, said in an email. “They can help to map fields for precision applications of fertilizers and pesticides while being used to measure crop response to various types of management,” he added. “Drones also allow for precision spot spraying of fields or aerial seeding of smaller fields.”
Better Drones for Better Crops
One problem with drones for farming is that they are often costly, with prices for a single model ranging up to $25,000. New cheaper and longer-lasting drones will help make them more available to farmers, Albert Sarvis, assistant professor and program lead for Geospatial Technology at Harrisburg University of Science and Technology, said in an email. “Five years ago, a 15 to 20 minute flight time was considered standard,” he added. “For the same, or lower, cost, current drones easily fly for 25 to 30 minutes. In the same way, sensor prices have dropped 25-50% in that same time period.” Future drones will become more important to farming and cost-effective once they are fully connected with remote sensors and each other, Durscher said. Once the data is collected, it must be more autonomous—have artificial intelligence (AI) capabilities, so it does not take a human to analyze the data and come up with a recommendation, he added. Software giants like Microsoft are investing in data analytics to boost farm productivity and reduce time and resources. Microsoft Azure FarmBeats enables developers to build artificial intelligence or machine learning models based on fused data sets. “That allows the assessment of farm health, get recommendations on how many soil moisture sensors to use and where to place them, track farm conditions and more,” Durscher said. High-speed wireless 5G networks that are being rolled out could make drones even more useful. Networks using drones with high-definition cameras are becoming popular for agriculture, Steven Carlini, vice president of innovation and data centers at Schneider Electric, which provides solutions for automated farming, said in an email interview. “With a private network, the owner can prevent things like data capping and speed throttling,” he added. “There is the potential for tremendous amounts of data generation enabled by 5G. It is impractical and costly to transmit data across long distances—local edge data centers with sufficient processing power are needed on-site.”
