If Acorn discovers 300 aphids across 2, 15m^2 patch, an optimal placement of ladybugs exists based on experiments, historical data, ladybug availability, etc. Does this project aim to encompass community experiments? I like the idea of setting up multi-farm experiments for those interested in participating since that could help propel farm profitability and lower the amount of time required to manage a farm. So in the ladybug example, one might estimate the impact of ladybugs on aphids over time as a function of distance between ladybug disbursement and how many were released considering crops, and weather.
So in the future when aphids are discovered by Acorn, the software could estimate the financial impact and enable a farmer to choose the best course of action for them or go with default recommendations of maximizing profit (ie purchase ladybugs and ladybug disbursement container and apply them at a rate of 1ladybug/10 aphids detected ).
Managing experiments might be easier with something like AWS’ IoT Greengrass.
In terms of raising money and creating value, the project might provide users access to research that could be applied by the robot. Users who participate in experiments with proper evidence get discounts.
Yes community experiments will be very important. I’m really imagining the project going the way 3D printer development went from 2010 on, when experimenters from all over the world posted their results and everyone learned from it. I hope we can get to that point!
Interesting thoughts on ladybugs! Would be cool to make a ladybug deployment tool.
Very interesting idea. We have sufficient volunteer ladybugs so we can’t help with your experiment. Have never seen an aphid in the field but we do see them on imported plants in the nursery. Outside, ladybugs reproduce at the base of young Ponderosa Pine trees, 6’ tall or less. When they hatch, the tree turns red and appears to be emitting smoke as they climb up the tree and fly away from the tips of the branches. An advantage of organic farming.
More relevent to your experiment, the overwhelming majority leave the property. Indeed, readings from 2 decades ago informed me that releasing ladybugs on one’s own property has little benefit as they tend to travel considerable distances before settling in. I have no experience with such things myself but that is certainly what we see from the natural swarms. Sounds like you have had different experiences when releasing small numbers. That could be the key difference. A crowded population may all depart while a sparce one may remain.
@RhubarbFarmer That is interesting about ladybugs. I’m not familiar with them or farming but am attempting to buy a smallish farm and would like equipment that can operate based on evolving open source experimentation. Having worked as a Process Engineer, processes or experiments that involve people are non-ideal since anything that can happen, given enough time will. I came across oatscenter.org that I think generalizes the ladybug experiment problems of “trust, automatable data exchange and interoperability” and might be leveraged to communicate Acorn data to Farm Management Software. Just trying to understand how the data generated in this project can be leveraged in conjunction with other projects.
@taylor If I had a farm I wouldn’t know what to plant or what to do. Reviewing experiments and designing them takes time. If I had data from 5 farms that produced 10 crops and I knew properties of the farm and history and had pictures of the plants and I knew the yields, I could find correlates of yields and the top 5 could be provided based on my priorities. If trusted verifiable experiments were already ran, risk goes down and I might make a bigger bet on fewer experiments. I’ve worked with Reinforcement Learning which I think provides a nice framework for running experiments and managing short vs long term goals and knowing how well a crop is doing 1 week in and what is the expected value of different actions.
Where is the most appropriate project to put shared data that could be analyzed? It seems like one of the Farm Management Software platforms might be best.
