Session: Internet of Things

Open Ag Technology GOAT 2018
1

Internet of Things

**Summary of things that people are working on: **

  • Open source versions of closed source sensor platforms

  • Carbon removal credits

  • Blockchain network for decentralization of data

  • Soil carbon

  • Crop monitoring using off-the-shelf moisture and temperature sensors

  • Dairy, mushroom farming

  • Sensing systems and monitoring systems (radios, cellular, streamed to internet)

  • Co-deploying lower cost IoT systems with scientific grade equipment

  • Tractor based monitoring, in-field long-term monitoring

  • Water quality and quantity monitoring

  • Research applications

**Topics and Applications people are interested in: **

  • Including uncertainty in data, data management, metadata (tagging from the beginning with provenance etc)

  • Sharing of data

  • Building, deploying, maintaining, on farm IoT,

  • Human-machine interaction

  • Operations

  • Automation

  • Greenhouses

  • Weather stations

  • Open source versions of closed source sensor platforms

  • Reducing “choppiness” of data streams, putting data streams put on the cloud, automation of modeling of data streams

  • Interoperability of hardware and software

  • Use cases of what the data is to know where it could be useful

  • Documentation of why a particular IoT sensor is selected and used (why and how), creation of modules for sensors has downstream implications, understand the decision-making process between that

  • On-the-fly processing of field data

  • Decentralization of data

  • Efficient irrigation (irrigation scheduling methods)

  • Sensors that can be deployed in remote areas and left alone for a long time

  • Human adaptation/ human connection to sensed data on farms

  • Data visualization that draws inferences from separately collected data (JSON, Elastic search), standardized communication language for IoT devices through an API

  • Field deployment of IoT: networked, weatherization, power resilience

  • Rapid on-boarding and off-boarding of devices onto IoT grids, moving quickly from a prototyping phase onto a product

  • Calibration, validation, scaling

  • Building stuff for researchers versus farmers (markets, quality of data)

  • Aaron’s talk: An architecture for real-time data exchange (goo.gl/TtxSuc)

IoT Topic One: Calibration and validation towards scale

  • Defining the topic:

    • How do we merge field sensor data and lab based protocols? On-boarding lower cost devices

      • Metadata around the process

      • Transparency of process

    • Ruggardization

    • Data versus analytics

      • Indicators

      • Measuring one thing by proxy of another

    • Building a low-cost version of an existing sensor

    • Building a training set

    • We also need data about protocols - how we got the data, in addition to the data itself

    • High levels of certainty are necessary for some applications but not others. How do we estimate uncertainty and error?

  • Action Items

    • Developing a standard for determining error ranges for field deployment of sensors

      • Figure out whether it is a data problem or a hardware problem. Opportunity: we can solve a lot with data methods now. It is not always a hardware issue. (Or sometimes it is an operator error issue.)

      • Look at guidance documents for EPA requirements for model year of each monitor, date of install, have to get it inspected once a year

      • Levels of error and uncertainty

      • Comparing sensors to other sensors - borrowing from medical equipment validation protocols, but does not have to be AS stringent

      • Many different standards for, say, weather station siting and all the data produced then comes with the assumption that it has been sited in an ideal space

      • Metadata Protocols: Software enforced or checklist?

      • ********* **W3C Sensor Network Ontology can be one way to describe properties of sensors, data and how it was collected **(https://www.w3.org/TR/vocab-ssn/)

      • Error ranges can be relative to itself

      • Suggestion: Open source label or standard that has similar reputational cache as “Cambridge Scientific” manufactured items?

        • OH Grade Level - whether it has been calibrated with scientific-grade equipment\

      • Flex-down: self-awareness, self-assignment in the mode of Creative Commons

      • Suggestion: Working group process for creating an open standard

      • Will this actually be adopted or useful?

  • Examples

    • Jack Chiu: Developing a wireless sensor network to monitor water levels in rice paddies

      • Irrigation management to improve water use efficiency

      • Production scale (400+ acres up into the thousands of acres, or in different cities)

      • Started off trying to build low-cost, but it wasted a lot of time. Started of deploying low-cost sensor and testing it against a calibrated, scientific grade sensor.

      • Ruggardization and field repairable

        • Had to figure it out himself, got help from Dr. Ken Fisher

        • Reverse engineering what was already out there from proprietary

        • Conformal coating on circuit boards helps in humid environments

        • Dessicant packs

        • NEMA ratingsare important for outdoor ruggedness: weather-sealed boxes to handle UV, seals that handle water splashing or pressurized water against the box

        • Panel mount connector that is completely weather sealed, modularity

        • Check for corrosion, visual inspection of devices and then determine if that is affecting the equipment and data output

      • Had to estimate the new error range and uncertainty when field deployment because the manufacturer error ranges were based on lab conditions

      • Scale-deploying: 80 units, development of building systems to deploy within 2 months

    • Ken Fisher: Tractor based monitoring system of plant height and canopy temperature

      • Ultrasonic monitoring works great in the lab but in the field, you can get a very accurate reading but you’re not sure what you are measuring in terms of canopy height. With a yardstick - do you pick the highest leaf or estimate the average height

      • Are you measuring what you want to measure? What are you actually measuring?

      • Interested mainly in relative differences

    • Greg Austic: Survey data on carrots, spinach, dirt

      • Working group creates the methods, Greg’s lab implements the methods

      • Can we standardize a way of labeling methods: include error range and how it is calculated etc.

IoT subsection: devices for real farmers

Notes by Jane Wyngaard

Deploying practical systems for actual production farms

Requirements:

- Guarantee ruggedness
- Guarantee uptime
- Onboarding thousands people/nodes
- Low costs
- Needs ot do edge compute
- Must be decentralised
- Data points (alerts) via text
- Lots ofbinary sensors: PLC on or Off
- Scale measurements
- Time spec: doesn't have to be real time often but 
- Need aggregation of data at node
- Price point...
    - Aftermarket tractor sensors systems are ~$500-1000k

Problems

  • Googles et al are not solving this problem with the NEST etc
    • Too much bandwidth required
    • Eg camera monitoring
    • OTS stuff is too outdated and not robust enough and TOO EXPENSIVE
  • No VC for this
  • DIY/hackerspace stuff completely unsuitable for hot moist areas
  • Massive regulatory overhead for a small maker to get a product to production level - possibly less so in Ag than other domains?
  • Alot of currently available commercial IoT ecosystems are gimmicks
  • Funding models?
    • Researchers create farm grade system in exchange for data?
    • Big Ag instead of funding lobbiests?
    • Wealthy farmer?
    • Big Ag conglomerates would fund build X system for their needs? Eg Famers business network was farmer funded originally
      • Open means everyone who needs the data gets it
    • USDA funding would have teeth
    • Funding in exchange for open data?
    • A model: hardware integration as a service
    • Can high school soldering club do the manufacturing for a system?
      • Entrepreneurship training
      • build experience
      • business experience
      • QC?
      • Would avoid alot of the regulatory challenges
      • Partnership with Research
      • 4H clubs, Future Farmers of America
      • Not a good model for a real bussiness?
      • Build a open kit with full instructions that a high schooler can follow - then anyone with no experience can follow
      • Research should be commensurate to anything: could they provide the gear in exchange for data

Tips

  • Conductive grease solves many things
  • LOW cellular data rate for LOW price $2/2M/month available

To dos/A way forward

  • could we create a workshop (Agathon) on IoT ecosystem for Ag?
  • EPSCOR events maybe for funding such?
  • An advisory group of big famrers to outline what the system would actually contain?
  • How to bring in the real world voice of the farmer into discussions - can we go to a conference they will attend?
1 Like
2

https://hologram.io/store/pilot-global-iot-sim-card/ offer 1Mb/ month for free.