Precision Agriculture with IoT, Machine Learning, Drones, and Networking Research

Precision irrigation is the idea of instead of homogeneously watering a field, to measure and water only the areas that need it.  (If there is a little dip in the field, water will roll down and that area will receive more water.)  The same concept holds for applying fertilizer, etc. and in general, this concept is known as precision agriculture. 

Ranveer Chandra in Microsoft Research has done some amazing work in precision agriculture using a combination of sensors and imagery from drones flying overhead.  In summary, you can use a small number of sensors in the ground measuring moisture, pH and fertilizer levels, etc.  The reason prohibiting a lot of sensors is that connectivity (of the sensors sending their data to the cloud) is very expensive (the sensors themselves are cheap).  Aerial imagery from the drones along with machine learning helps us find the best places to put the sensors to get the most information.  In addition, unused television channel frequencies can be used to send data packets in place of conventional WiFi to improve the connectivity situation.  Television operates on a much lower frequency and carries much further than WiFi.   

Resources:

- Project Page: https://www.microsoft.com/en-us/research/project/farmbeats-iot-agriculture/

- Article in The Economist: https://www.economist.com/news/science-and-technology/21707242-unused-tv-spectrum-and-drones-could-help-make-smart-farms-reality-tv-dinners

- YouTube video: https://www.youtube.com/watch?v=pDgjOHY7sMI