Cloud platform for openAg


#1

Hello guys, I was wondering, does anybody know about plans to create an openAg cloud platform?

I am creating an environmental controller (basically a brain box) and wanted to connect it to the cloud. I want to potentially connect to a wireless sensor to the environment controller and use the info to control a relay. We almost have a commercial product ready, but we still need either a server or a cloud service to start selling. An OpenAg cloud platform seems like the perfect next step, someone must be working on it, no?


Can I help you with anything — cloud based data gathering, analysis, product development?
#2

After searching the Wiki for “cloud” I found a number of results:

Current state of cloud software platform they’re working on.
Roadmap of cloud software platform development.

They’re definitely working on it and it’ll be robust when it comes out.


#3

I can confirm what @Webb.Peter says, we at OpenAg are working on a cloud solution. No timeline for it yet, as we are a pretty small team and doing a lot of design / integration / research tasks at the same time.


#4

Check out https://plant-os.com developed on Google Cloud. Has similar principles but our goal is to make it simple enough for everyone to use. We could use some love from this community.


#5

Is your work open source? The website looks great but if the platform isn’t accessible it doesn’t really help our community that much at this point.


#6

Yes we started PlantOS after watching Caleb on The Verge(2014) with similar principles to MIT OpenAG however the difference is in the approach where we want to make it easy enough for people to get involved by refining the technology first. On a side note we have been working with Fenome LLC folks and our adviser is from MIT. Our hardware (sensor computer and NDVI camera) are open source with code and details in our PlantOS Github. We would prefer users with Atlas Scientific sensors as we are concerned with data fidelity. If anyone has any sensor suggestions with similar performance please let me know.

We are currently working on publishing our APIs to enable access for everyone on our dashboard. The advantage would be access to a dashboard designed by our partners which includes plant scientists and Google Cloud(Singapore) on algorithms to apply . The goal is for users to select the crops they want to grow from simple icons with pre-loaded plant recipes and algorithms so users don’t have to reinvent the wheel to optimize their VF operation with a dashboard showing pertinent information. Yes we do pursue commercial interests with our clients however these tend to be bespoke work from design to operations.

We would like to start a discussion on dashboard requirements from the community so we can implement them for the greater good. If it is a good idea we can start another thread on this topic alone.


Omaha Mavericks Food Computer 2.0 - (MavFC)
#7

This is great to hear. I’m curious what your work with Fenome has included, and who your adviser is. I’ve seen some discussion on the OpenAg Gitter about the cloud platform which included a demo, but other than that there has been complete silence about a cloud platform so far as I’ve seen. I’m excited to hear that you’re working with Google Cloud Platform, I don’t have much experience but have built a few simple apps with it and loved how quickly I could develop a scalable solution.

I am the owner of two V2 PFC’s which I would love to connect to a cloud platform. Additionally, we just finished the $300 Food Computer MVP Project. I would love to utilise your cloud UI for that as well, we are running CouchDB and a simple HTML server right now: http://openagbloom.ddns.net:8080/ We developed the MVP with the idea that it would be more accessible and less costly, obviously, this meant we had to sacrifice a lot of sensors/actuators to do this. I’d like to try and hook it up to your UI though if that is feasible, I’m going to look through your GitHub now. Are you running the same openag_brain or something different?


#8

We spoke to Richard and Sowmya in Fenome prior to work on the dashboard and algorithms for specific metrics on our dashboard. We have been informed of changes in structure and currently in contact with Caleb to hopefully work together soon. Our advisors are listed on the site, do click on the images to access their respective LinkedIn profiles. Thanks and Google Cloud(Singapore) has been an awesome partner supporting us with expertise and resources. It took our team 6 months but we finally have our cloud platform streaming data from our customers VF site.

That sounds great! We are working on publishing plantOS cloudconnectors (our APIs) soon so the community can access our cloud platform easily. The process would be following our instructions on Github to setup API on hardware and sign up to our platform with your Google account(ex: Gmail or GSuite). We have screenshots on our Instagram of data streaming from our customers that are bespoke engagements from design to operations.


#9

Who is it that you are working with on the ground? It sounds like so far what you’ve done is stream data from an existing farm, not a Food Computer?

I went through the Github, and the only stuff there is a CAD drawing of the Raspberry shield and some sensor control software - really nowhere near enough to run an actual farm. I don’t see anything for the cloud or data storage or data formats. I guess my question comes, what is your goal for the usage of PlantOS as it relates to the OpenAg Open Source community.


#11

I’ve bootstrapped operations since late 2014 with my co-founder. plantOS team

Yes are currently streaming data from existing farms. We have 3 user segments

  1. Community [Open Source]
    PlantOS Raspberry Pi Shield - Print circuitboard and connect it to a Raspberry Pi 3 Model B to assemble plantOS sensor computer. We also use Atlas Scientific sensors
    IoT Github - Code to stream data to plantOS platform
  2. Vertical Farms
  3. Plantations (ex: Oil Palm)

You are right and our Github was a mess. In fact Caleb gave the same criticism after his engineer reviewed our Github. We’ve finally completed our cloudconnector module and significantly restructured our Github. Please check it out

I do have a day job and have been working on this in private with my small team. We do incur operations costs notably Google Cloud charges however we are fortunate to have support from Google Cloud team in Malaysia/Singapore assisting us with credits as we scale. This enables us to make this announcement to welcome everyone to sign-up as our community user. Over time we believe our business model allows us to support operations costs for our community users without relying on assistance.

We need help in several areas:

  1. Completing our docker image. We are having issues with it
  2. Mockups and ideas for our dashboard design. We use Stencil for mockups
  3. Machine Learning and Computer Vision
  4. Web application development (Javascript, HTML and CSS). Our dashboard hosted on Google Cloud is based on Flask.
  5. Open Hardware certification

Please drop me a PM with your email address to be invited into our Slack channel as we are on the Free tier of Slack which has limitations. Alternatively we would be happy to answer questions here and work with all of you to make us better. We are registered in Malaysia and serve a segment of customers that won’t be able to afford USD 800 sensors when an average farmer here earns USD 400 monthly. We are currently exploring options to manufacture our hardware at scale and lower costs significantly to democratize digital agriculture.


#12

Hi @Everyone, we would like to drive discussions towards how we can improve our dashboard to support OpenAg PFCs? Ideas are great but mockups are better. We recommend using Pencil which is Open Source.

  1. What else are we missing to support a PFC? (ex: sensors, actuators)
  2. We are looking to expand our KPI feature to incorporate digital plant recipes. We would feedback on designing this. Have seen OpenAg examples of it being a JSON file with upper and lower limits for a few parameters. However we are looking at implementing ML models in Tensorflow(Python) for crop types we support and perform predictive analytics based on our dataset.

We look forward to this exciting phase of understanding how users are using these systems and measure optimization techniques achieved on plantOS platform. Currently we have drafts for Chinese Cabbage and building one for Oil Palm but we hope to work with the community to increase the number of crops plantOS platform can support.


#13

Hello everybody, is there a place where the owners of the PFC’s could voluntarily share their data into the cloud, so that everybody could take a look at the data and see what they could learn from it? I think that might be a helpful tier between the individual PFC’s and any apps that use the data.


#14

No, not yet. But, here are some things you could check out to get a sense of the current progress:

  1. @rbaynes from the Media Lab released the cloud platform code he’s working on at https://github.com/OpenAgInitiative/openag_cloud. There’s related discussion in the comments for the thread about “plantOS Welcomes Community Users - Code & Schematics…”.

  2. There’s some flavor data from the OpenAg group at https://github.com/OpenAgInitiative/flavor-data. Somebody please correct me if I’m wrong, but I think that’s the only public dataset so far from OpenAg.

  3. The OpenAg publications page is at https://www.media.mit.edu/groups/open-agriculture-openag/publications/

  4. Read the discussions about plantOS.


#15

Hey Sichem,

We do not support that functionality as of now. You can however, connect to our dashboard and create insight into your own data. If you’d like to share a visualization with someone, then you can create a screenshot which you can share with whomever you like.

Hope that is okay for now :slight_smile:. I must say that I do love the idea and I’m discussing it with Devtar as of now. We have some other priorities at the moment (think of easy deployment and an updated dashboard which we’re currently working on). But we’re going to put it on the list of functionalities that are to be implemented.

If you’d like to try our platform, then please follow the instructions on our GitHub. We’d love to get your feedback on our (current) platform!


#16

It is in the works, but we are missing a few pieces (cloud reporting) and I want some time for design issues to ‘bake’, and run more testing. I could see a lot of it being out in several months (any interest in beta testing?).
You correctly identify a need, and something we have a vision for. Just remember, the MVP was a ‘Minimal’ product, with the intention that it will grow as we have time and resources.


#17

Sure, I’d love to help with beta testing and other areas as well. This is still all very new and very interesting. Thanks!


#18

Let me know when you have things set up and running. There are a few modifications that need to be made (swap out some code), and set you up on the cloud.