Conference Presentation, Keynote
Chandra Krintz
- Analytics is expected to drive economic growth and deliver tremendous success in web services, e-commerce, and social networking, serving as the key to unlocking commercial opportunities in emerging markets.
- With global population projected to reach 9 billion by 2050 against a current baseline of 7 billion and limited arable land, the agricultural sector requires significant technological improvements to maintain efficiency.
- A $50 billion California agricultural market was noted for last year, with future stability contingent on adequate rainfall to maintain current levels.
- The strategy involves turning consumers into producers and developing specific agricultural analogues to platforms like Amazon, Google, and Tinder for tasks such as irrigation scheduling, crop disease detection, and pest management.
- Research aims to converge IoT, robotics, big data, machine learning, statistical analysis, and image processing to solve farming problems efficiently.
- Data analysis will identify high-probability problem locations to direct farmer energy, while multispectral drone imaging and thermal spectrum imaging will track crop health and detect leaks through temperature differentials.
- A hybrid computing model combining on-farm and public cloud resources is deemed necessary to address low bandwidth, high-cost connectivity, and the volume of data generated.
- On-farm systems must be self-managing, fault-tolerant, and "lights out easy to use," functioning like consumer appliances without requiring system administrators.
- The distributed system architecture, built from NUCs with 24 processors and over a terabyte of disk, is designed to continue operating intelligently even when losing individual boxes or components.
- Software applications written in Hadoop, Spark, R, or MATLAB must be portable and run transparently on both cloud and farm environments to support a diverse population of contributors.
- Real-world validation will be conducted through partnerships with growers rather than relying solely on laboratory simulations.
- Hardware costs must decrease significantly to ensure solutions are super inexpensive, and the industry must provide models allowing farmers to control their data regarding sharing or selling.
- The approach prioritizes solving specific, real problems identified by farmers directly and emphasizes the need to build a community to successfully feed the planet.