The ‘ WILDCARD ‘ Gro-Bot Organeec 1


Decided to move Gro Bot diary from my website to the forum in view of the imminent release of the V2 software.

When the first builds began to appear I was unable to get the V1 software running so decide that I would wait for the V2 to be developed but continue, albeit at a slower pace , to construct a, ’ Food Bot’ using the ubiquitous dead fridge. Partly as a nod and a wink to the extremely popular movement of, ’ Up cycling ’ , here in Britain and also as I have previously constructed a 3D printer from up-cycled printer and scanner parts which was exhibited as an 'Art Work ’ a la Marcel Duchamp Readymades.

At the start I came across this, TED talk, by Caleb Harper, and was inspired.

Just what was needed, but having checked through the OPENAGriculture site, , Forum, Wiki and Blog it was obvious an insulated box was required.

Which mass produced insulated box came to mind ? Yes, A Fridge. Just bought a new one and the old one went so no fridge!! to work on. Tried the local fridge seller but was told I would have to wait until one came in. As luck would have it, driving the country lane route back home, there appeared around a bend, in a passing place, the fridge you see here, so it was rescued to be resurrected as a OpenAGriculture Personal Food Computer.

It’s title, The ‘ WILDCARD ‘ Gro-Bot Organeec. is for when it is Exhibited as an Art Work.

!!! Safety alert !!! If you decide to use a fridge.

Older fridges using environmentally harmful refrigerant gases need to be de-commissioned by a registered fitter. In the UK it is illegal to vent harmful gases to atmosphere. Check your local laws for information regarding the scrapping of fridges.

This fridge is labeled as using R600a refrigerant which is, isobutane, an inflammable gas, lighter fuel. It is not harmful to the atmosphere but will ignite easily. At first I thought of keeping the cooling system in case the nutrient solution needed cooling but, as this was an unknown fridge I didn’t want to run the risk of a leak which may cause a fire so removed the cooling system.

Cleaned and painted ready for internal re-fit.

Good clean with bleach, then coated in white anti-rust paint = good as new ! ( except for the lack of cooling power. )

The freezer compartment door is air tight so the idea at the moment is to put the electronics inside and create cable access either through the door or through the back. This will be possible as the cooling pipes run around the top and sides. It might also be possible to have the vent fan at the back and connected by a flexible pipe to a hole in the door.

Keep finding out snippets of info from the OPEN AG Forum, turns out plants like water temperature around 10 degrees cooler than the air temperature.Blue sky thought ! Maybe it’s possible to cool the nutrient water in the fridge freezer and collect the heat of extraction from the fridge radiator. Then feed it back into the fridge to heat the air ? If the freezer section was open to atmosphere it would have a heat source so only the section cooling the water would need sealing from the atmosphere.

Inside I have removed the temperature adjustment sensor from the actuator so as not to compromise the, what I assume is part of, the sealed system.That leaves a useful space which housed the lamp, ( It could be used to mount the camera, and other sensors ), a switch actuated by the door, and access to the outside for cabling. There is also the drain hole which can be used for access. The original container for the lower part of the fridge was missing so this is one that I kept from the fridge that was taken. It is smaller than the original so if I find a larger one then I will use that.

In hindsight it would probably be better to use a freezer rather than a fridge freezer as this would give a bigger grow volume height and therefore a bigger range of plants that could be grown. I am already thinking about converting the most popular greenhouse sizes into grow bots, mostly 6’ X 6’ , 6’ X 8’ and 6’ X 10’, with removable insulation that is only added during the colder months so that running costs are kept to a minimum.

Anyone calculated the average grow volume for the most popular plants ??? Guess current greenhouse sizes are already sized around that.

Sensors and Actuators.

In order to control how a closed system works you need to sense chosen changes in the system and modify those changes to your requirements. Sensors and actuators provide that control coupled with computer applications connecting the two together according to predetermined rules. Here are some of the practical goodies, Temperature / Humidity sensor, light sensor, Ph sensor, Base shield, for the attachment of actuators to the microprocessor control, grove connectors and a small prototyping board. until the system is working I’m holding fire on getting the Co2 sensor as it’s quite expensive. Could be that the v2 version of the Gro-bot has a superior sensor ???

Proposed power supply. If not beefy enough will add another !!


Getting a fan

Finally decided that it would be OK to put the circulation fan in the door of the Freezer compartment. As you will notice this required slicing a chunk off the shelf that is part of the door to allow the louvres to rotate open. I also removed the refrigeration pipes now that I know that venting the coolant in this system will not harm the environment. Makes it easier to route the power and digital cabling.

Below you can see the inside of the freezer compartment showing the rear of the circulation fan and the air intake / cable duct opening at the back of the compartment. I’ve also decided to go ahead with housing all of the electronics including the air heater inside the freezer compartment so if this proves too small for whatever reason then another hole can easily be made now that the refrigeration system has been removed.

Extra Arduino bit

Above is a pic of several Grove connectors. As I have decided to create my own expansion board using veroboard I will need to bend the outer two pins in order to solder the Grove connectors to the board as the pin pitch of the connectors is not standard.

In the picture below you can see the extra nine connectors mounted on the veroboard.

The above picture shows the board with the jumper wires and Arduino connections.

Below is a close-up of through pins attached to a piece of veroboard used to connect to the Arduino.

Heat and Sight

I decided to go for this 1200W, two setting, £5.00 hair dryer, from UK WalMart / Asda, to act as the air heater for the Gro-Bot it comes with a thermal cut-out so it should cut out before melting. It’s going to be housed in the freezer compartment of the fridge so will not be affected by the humidity levels in the grow chamber.

Below you can just see the thermal cut-outs in this close-up of the air intake of the dryer as well as the heater coils.

In the image below you can see the original cover for the fridge light and temperature control. The bit sticking out on the right is a switch for the light which is actuated when the door is opened and shut thus switching the light on and off.This will be connected as the door open/shut sensor.

As you also see I have mounted a USB camera in the space which used to house the temperature control switch. It is a Playstation 3 USB camera costing £2.00 at the local games exchange shop. I have connected this to the version 2 software and it functions OK some of the parameters will need tweeking to get the best image though.


Can you share the settings in the DB for interfacing the PS3eye camera? Are you able to see the feeds in the latest openAg_ui?


Hi Vishwanathanlyer, with the latest builds from the repo I have no feed showing in the database and also nothing from V2 Openag_ui both the hosted version and my own local build.

The only setting that I changed when it was working was for the autofocus , it needs to be, False, but that is the default setting, or was.

There hasn’t been much changing in the repos over the last few days but Serein noticed that the usb_cam Ros package is not installed by the openag_brain_install shell script. So maybe that has something to do with it.
Just looked on Git Hub and Gordon is busy working on the UI.

At the moment I’m brushing up on my knowledge of Ros until the script is updated, and re-nstalling the latest code from the repo.

Have you installed the , openag_brain_docker_rpi, there was some mention that with that running parts of the system worked, but not sure whether or not the ui is linked to the camera yet. I don’t get much when running the docker brain as I have not got the right sensors for the V2, I’m trying to help with getting the V1 sensors working with the V2 brain. V2 is the priority.


Hi John B,

I have installed using the same shell script. (openag_brain_install). I have installed the openag_brain_docker_rpi. I read in the forum that the camera part of the code is in heavy development and hence not stable. I also do not have all the V2 sensors. I got the air temperature and humidity and water temperature sensors working along with the plots on the openag_ui. I am still figuring out how to interface actuators. Actively waiting for the stable release of V2.


Just add to your firmware_module something like this:

"_id": “light_actuator_1”,
“environment”: “environment_1”,
“type”: “binary_actuator”,
“arguments”: [

Argument (6 in my case) is Arduino pin number you have connected your actuator to. This will let you to control it manually by publishing messages through rostopic command like this:

rostopic pub /actuators/light_actuator_1/cmd std_msgs/Bool True -1

Yet there is no official code published to control specific variable with actuators (i.e. control loop through PID controller). I created my own glue code for PID and was able to turn on/off lights following the recipe needs. But IMHO this part is also rather raw and subject to future change, so decided to wait until it stabilizes. Here is relevant thread on GitHub


Hi, Vishwanathanlyer, just today got a system functioning and showing data from the dht22 air and humidity sensor in the hosted UI.
I downloaded from the repos Thursday 13th Oct, debian jessie with the pixel UI and the latest software from openag repo.
Will be trying to add the TSL2561 light sensor code and playing with the actuators following the instructs from Serein in the coming days.



Into the MISTS OF TIME !


How to get the nutrient solution to the roots of the growing plants ? I’ve decided to try a hybrid approach where :

  1. The nutrient is pumped from the container and drip fed to the plant roots and returning back to the container.

  2. At the same time the container will be filled with mist from the humidifier in order to keep the roots from drying out and possibly to encourage the dripping nutrient to cover most of the growing root system.

That’s why this BIg Baby was purchased, half price at £15.00. Available in drug stores and baby product suppliers here in England being used to keep the centrally heated bedrooms of babies at the optimum humidity levels for the baby to grow / sleep in.

The image below shows the base of the humidifier with the piezo ceramic ultrasonic generator in the centre of the metal ring. This is where the solution is vibrated and separates into 200 nm – 1500 nm sized particles. which then rise up the central tube and will be guided through a flexible pipe into the gro chamber.



Thanks! That worked! One change was that the id to be used in the one which automatically generated in couchDB.


I was able to interface BS1750 digital light sensor with the system. I have moved from the docker based image into the locally installing the software in the RPi. It seems that the docker image has the light illuminance variable missing which has been fixed the github repo.


Gro-Bot Organeec 9 : Window into the BRAIN

Finally decided that access to the BRAIN wiring in case of a fault or re-design would be needed without opening the door and carefully performed a fridge brain lobotomy using that piece of trusty specialized equipment the unmounted metal saw blade. The saw blade cut through the insulating outer layer with ease and slowed down only when the refrigerant circulation tubes were encountered. Now it is time to tidy the wiring, insert the arduino, relay board and the Raspberry Pi2 Brain. After that building a, nutrient module seems the best way to go as there isn’t room to mount the humidifier on or inside the fridge. By the side or on top ? Next model might be better in an upright fridge/ freezer then the fridge section can house the, electronics and nutrient modules, and the freezer section the growing section.



Gro-Bot Organeec 10. Brain Surgery.

Since the last post I have concentrated on getting the V1 sensors to connect to the UI so that there is some data logging happening and being recorded to the Couch database. Slowly I am
beginning to understand how the different parts of software work together.

This screen shot from the, Brain, is showing the data traces from the V1 sensors connected to the arduino, A TSL2561, dht22, dfr0161 and a simple thermistor based home made temperature sensor which is intended to be used to show water temperature.

This is only part of the story as the data from the sensors has to be connected to the actuators so that when the system parameters change they can be acted on to bring them back to prescribed values, following instructions from the plant recipe.


@JohnB, great job! I have a FC v1 and I would like to update the software to v2, so I can get the data along time. I already did an attempt to do that, I was able to install the brain and log from my laptop, but I saw nothing. The Arduino wasn’t working and my Pi wasn’t getting any data from the sensors.

Could you please share with us how you have made this work?


@JohnB can you share your progress? I’m on your six and super interested in how you got the sensors to read. I’m assuming you put data manually into the couchDB? How did you modify the ROS topics? There is another thread actively moving on this too. It would be great to create a wiki page for this specifically as there are many of us who are struggling bringing PFC1 up on openag_brain v2.

Thanks for your help!

Help needed with Loading the default fixture

Hi away on hols at the moment using tablet and unsure of interface I will do my best to answer your question. The data is real time from the conected sensors but intermittently as the serial connection to the arduino through ROS was always giving errors. I could not work out whether this was hardware or software issue ???I decided to wait until the software became more stable and at the present time watching developments on GitHub as Gorden and the team are busy changing the software to work with the raspberry pi gpio and dispensing with the arduino altogether. At present will arrive home week beginning 23rd April and will resume working on project then. The software changes will require a complete rewire of the sensor/actuator system for the v1 sensors that I am using as well as reworking the software drivers. Catch up with you next week, John B


Hmm. I’d like to know more about using the GPIO. There’s a reason a MCU is superior to a CPU for that type of work. I get that the USB integration is tough but I’d prefer it to work reliably when it works.

I’m currently looking for any and all Firmware_module JSON code you may have. I eeked the DHT22 from your posts but I want them all :slight_smile:

Have a great holiday. I look forward to hearing from you.

Also, check this out:

Feedback desired.




The wiki page looks great. I just read through the instructions without trying them on hardware–based on that, your procedure sounded clear.


Hi, juhnke, try searching, misterburnley on GitHub, You should get a list of links to all the posts that have been made about getting the v1 sensors working with the v2 software. Also the stand alone brain box might be a suitable route if the sensor code for v1 could be added or swapped in.


Will do John!

I’m currently re-writing the gc0011 driver using the original gro-microcontroller code because the current gc0012 driver included with the PFC2 configuration uses a serial port and not the hardwired pins in the original Arduino Shield. The original pins were 11 and 12, reversed rx/tx and are NOT one of the three serial ports that are available on the Mega 2560 so without some clever re-soldering, PFC1 HW users will not be able to use them.

I’ll post the driver when I get it working.




Hi, just had a read through of the instructions and they are straightforward and easy to follow. I used the docker image last year and as the software developed the docker image did not get updated so I have stuck with the code updates and recompiled locally when necessary. Same with the UI. At the moment I am in the process of working out the wiring in the “fridge” so don’t have a working system at the moment.
As to the firmware modules ( sensor driver code ) they are hosted on my Git Hub account, , the DHT22 has always been included in the openag initiative V2 sensors firmware_module_type database list, so would be available to be added into the code when put into the firmware_module database.
I think you can put any number of sensors into the, firmware_module_type , database. They are only used when you add them to the, firmware_module database which adds them to the code that platformio compiles for the Arduino.
I am at #137 22nd Feb so have some catching up to do.
When I had the Ui working in November, the workaround was to issue the command, platformio serialports monitor , from a second terminal.
It might take a week or two to get up and running again, will let you know How I get on.