Using a Raspberry Pi instead of Beagle Bone


#24

@wsnook I don’t want to hire you for anything. We are just looking to build a Raspberry Pi version of the PFC. We wanted some mentors/people who share common goals to collaborate with us. And yes, we are planning to make a large scale smart hydroponic farm. We don’t intend anyone to work with us dedicating their whole 24 hours on it. Thanks anyway for your help until now. And yes, we will be uploading forked repos too for the community.


#25

It looks like we’ll be the ones learning from you! Where will you be posting updates on your progress? Let me know so I can add it to my list of active projects to keep an eye on!


#26

In that case, you might be interested in some of the recent commits to https://github.com/OpenAgInitiative/openag-electrical.

I just noticed today that they’re working on an adapter plate to experiment with using a Raspberry Pi with boards that were designed to fit BeagleBone GPIO connectors. Related openag-electrical commits: c2aac39, 09fab23, 72c7547


#27

We will confirm as to where we will be posting our updates from 28th of January, since we have our Semester End Exams until then. Thanks again for taking interest. :slight_smile:


#28

For everybody following this thread, you might like to know that Jake from the OpenAg team at MIT has done a lot of work in the last week on modifying the PFC_EDU device software to be usable on other platforms. It looks like he intends to add support Raspberry Pi and macOS with an FT323H USB I2C adapter. You can follow the work in progress in the raspi-integration branch of openag-device-software on github. Here’s a screenshot of the openag-device-software network graph where you can see 17 Raspberry Pi related commits (blue line) since January 8:

I haven’t talked to Jake about this, and don’t have any further information about it other than reading the commit diffs on github. But, it looks like he’s making a lot of progress.


#29

@wsnook you are correct. @jake has made a lot of progress. He is modifying the brain to be platform agnostic. e.g. you can run it on any linux machine with a USB-I2C-FT323H and also specific support for the RasPi.

This is part of our continuing research and NOT a finished product. If you grab this branch of the code, expect to do some testing and possibly find some issues. We are still working on it!

Here is Jake’s big win for yesterday: a PFC_EDU 3.0 board being run by a RasPi instead of a beaglebone. The software is the latest brain that Will mentioned:

Great job, Jake!

CC @David_Z @hildreth @Poitrast


#30

I know that not all Hz are created equal.

However, it’s difficult to beat the power of a 1.4GHz raspberry pi (3B+) processor compared to BBB’s 1GHz

… and the fact that RPI has 3 additional cores!


#31

I’m glad to see RPI getting some love again, but I think it’s important to recall why the team moved away from this device. As I understand, data corruption on the SD card is common and the move to BBB was as much about data integrity as anything else (please let me know if I missed something else). The benefits of not having to continually reload the RPI outweigh the additional cost and lost power imo.


#32

I didn’t understand what you meant by continually reloading the RPI and lost power imo. I didn’t get you there. Could you please explain?


#33

In my experience the choice of which micro SD card to use is huge (SanDisk Extreme is my choice). I have never had an SD card issue which is why I scratched my head at that being one of the reasons to move to the BBB. Another nice feature of having everything running off a single card is that you can easily remove the card and take an image of it.

The Orange Pi looks interesting spec wise it seems like a cross between RPI & BBB.

There are many new micro controllers surfacing these days. I think it’s wise to not lock into any one at this point.


#34

I think this thread describes it best. Essentially, the RPI uses an SD card to store the OS and this design leads to OS corruption over time - either due to data transport errors or latent corruption on data already stored. This isn’t just the PFC, it’s a common problem for RPI users in general.

Agreed!


#35

@Drew you are correct. We switched to the BBB because of its rock solid file system and wifi support.

The BBB can simultaneously be a wifi access point and connected to your local wifi. That allows us to configure, set up and access it easily. The RPi can only be one or the other, which is klunky but @jake made it work.

The RPi does have an advantage over the BBB in terms of camera (and community) support.


#36

@Drew My team and I have started working on the code in full swing. Actually we started quite a while back. There are a few sensors left out. But in the CV part we actually have a problem, we are creating our own software for it. Could you please send us some resources that we could possibly refer to?


#37

I think Wsnook said it best:

Sadly, I’m in the same boat.


#38

I am not asking anything about what MIT have done @Drew. We are in the process of complement edge detection and area. We have leaf size and nutrient deficiency left over. We are having a problem with the leaf size part. When we went through a few blogs online, we saw that this is done by comparing the leaf to a known size of any other object. Hanging some other object in there doesn’t seem feasible. I basically wanted to ask whether you have any other ideas regarding this. Like any reference material or something. I don’t want anything related to MIT.


#39

I have zero experience or knowledge about image recognition. I’m afraid I cant be of any help at all :frowning: . Perhaps start a new thread on this topic to field a bigger audience?


#40

It’s fine @Drew. I’ll figure something out. I have another problem by the way. It’s about the chiller unit. I see in the BOM for PFC2.0 that they have mentioned that we can use PC water cooled chillers and PC water cooled radiators. So my question is that is therelike exact control of temperature in these chiller units. For example if I want the air temperature to be at 28 degrees celesius, can we set these PC chillers to 28 degree celsius?


#41

I too am looking to build the food computer with my raspberry pi 2b. How is your build doing?


#42

Just a few sensors are left out. We are trying to figure out the chiller unit part.


#43

Hey guys, we are hard at work on our model of the food computer. And about the chiller unit part, I actually got in touch with Kipp Bradford at kippkitts about the chiller unit that had been designed for $800 as quoted by him. He was actually kind enough to connect me to a dealer in China which makes a smaller version for less the price. I got a quotation of $600 from them which was way better. But, recently I went through a link on github by a person called Rob Baynes, where it has been said a PC water cooling chiller and a PC water cooling radiator can be used to build the chiller unit. Can anyone put me in contact with the person who said that it’s possible to control the air temperature with this? I just wanted to know how it was possible and just wanted to see if things worked out better for us. Thanks in advance.