Electronic nose to detect fruit ripening
Concrete Jungle started in 2009 as a volunteer-run, fruit and nut distribution organisation in the city of Atlanta. Utilising the vast number of urban fruit trees, the team, started by Craig Durkin and Aubrey Daniels, collect and deliver neglected fruit to shelters and food banks across the city.
While some urban trees are located in backyards, easily observed and maintained by their owners, others are less cared-for. And given the different harvest times for the multitude of fruit and nuts, knowing when to take the time (and a team) to pick the delicious bounty can be a bit of a hassle.
So for the last few months, Craig Durkin has been working on a Pi-powered means of using tech to notify him when fruit is ripe and ready for picking.
We’ve teamed up with Carl DiSalvo’s Public Design Workshop at Georgia Tech to try to create an electronic nose that can smell fruit ripening in a tree. This way, the tree could let us know when it’s ready to be picked.
The build has gone through several variations, all documented via a Hackaday project page. Titled ‘Electronic nose to detect fruit ripening’, Craig has tried and tested several boards, sensors, and fruits while trying to establish what gases he’s looking for, and how to detect them.
With the project still in the testing phase, it would be interesting to see what our community can suggest to help Concrete Jungle with their build.
What are your thoughts?
My first thought is the gravitational field of those apples is really strong.
Perhaps a camera with filter appropriate to the ripened colour of fruit, maybe an infrared signal?
Possible change in leaves.
A sensor in the tree that measures sugar levels in sap?
Philip Le Grand
A lot of fruit (all?) give off a gas called ethylene when they ripen. In fact this acts as a trigger for other fruit in the vicinity to also start to ripen and how the supermarket’s can guarantee to get you fresh ripe fruit throughout the year when a lot of it is picked when under ripe overseas.
Maybe a sensitive electronic ‘nose’ to detect a build up of this gas, although levels would be affected by the weather e.g. rain and wind.
Hi Phillip, if you follow the link to their website you will find some interesting info including.
” And ripening is a complex process: we know that ethylene is the main gas released by many fruits as they ripen, but there are likely many other gases released as well. For example, Hirano et al used sensors similar to ours to monitor popcorn cooking and waffle cooking: “
* ‘Listen out’ for fruit to fall naturally in a breeze?
* crowd source the state of ripening with a location sensitive smartphone app
You could always just use the Pi camera, point it at a bunch of fruit, and set it to upload a picture every few hours. For fruit that change colour, that should save you a lot of time. You won’t get a yes/no decision from the computer, but you can look at a picture on the Internet rather than physically visiting the tree.
It would be interesting to know if you could train an image recognition algorithm to tell ripe fruit from unripe fruit. I expect the answer is yes, but you might find it hard getting enough training images to make it work.
I think the gas based solutions are going to be really hard when you try to make a sensor that can be used outdoors. There are so many things that can interfere with it: rain, wind, temperature changes, interference from other trees, interference from other gases such as traffic exhaust.
Whatever system you choose, you’ll need to think about power. If you’re using a Pi, it will deplete a reasonable size battery quite quickly. You might be able to bring the power usage down a bit by using other hardware, but anything that has a radio (for WiFi or 3G) is going to use significant amounts. One option might be an electronic timeswitch where the Pi can ask to be powered down, then turned back on in a few hours time. That way you could conserve your batteries by only running the Pi for the time necessary to take a reading and upload it.
Many varieties of pear and apple indicate readiness for picking when the small pores (stomata) on their surface change from green to rust-brown in color. The aroma of truly ripe apples is due to largely to a mixture of short-chain alkyl esters, aldehydes and ketones (e.g. ethyl acetate). Rather than using an adsorptive sensor, perhaps you could explore using an optical sensor that looks at IR wavelengths specific to the ester linkage for the presence of these compounds as indicators of ripeness (ca. wavenumbers 1715-30). Obviously this would need to done on samples harvested at various times, since “on-tree” monitoring would probably not yield usable results because of fluctuating environmental factors.
The first picture looks like a persimmon. Those are easy to figure out if they’re ripe — bite into it. If your mouth feels like you just bit into a cotton ball, it wasn’t ripe.
Worse news: they don’t all ripen at once.
Pi Zero boards are going for $5.00 at Micro Center. They come camera ready. Not as powerful as Pi 3 Model B, but powerful enough to do the job and cheap enough to deploy in large numbers. https://www.raspberrypi.org/blog/wearable-pi-zero-camera/
Comments are closed