Most Read This Week
From the Blogosphere
[video] @CHarrold303 Is Building a Personal #IoT | @ThingsExpo #M2M #AI
The birth of an IoT DIY project from beginning to finish
Mar. 6, 2017 09:45 PM
About 6 months ago I had this "brilliant" idea that I would like to monitor the water level of my plants living on my balcony. The goal was to keep them healthy and happy (and producing vegetables for me in my urban garden). After looking for a commercial solution to aid me in this project, I discovered that there wasn't anything out there that would do what I need. The primary reason why nothing worked for me was that none of the products out there are capable of monitoring multiple plants at once, leaving me with the need to buy multiple units. At 10-15 containers for plants, this was immediately recognized as unrealistic, due to their high individual cost. Being the enterprising person I am, my immediate thought was, "I will just build one! How hard can it be?"
Six months have passed and I am starting to grasp the reality of it. It isn't hard, in the sense that I am not treading lots of new ground and making a lot of things up from scratch. It is definitely not simple either. I have a working prototype of a single instance of a soil moisture sensor, temperature and humidity sensor, and a light level sensor in a "sensor pack." Emphasis being on "prototype" and "single." It has taken me 6 months of time to buy, assemble, trial, error, and otherwise jigger around with bits and wires to get a working prototype that is extremely fragile and not at all ready for commercial use. Now to be fair, I have a full-ime job so I didn't do this all day every day; however, I did spend serious late night hours on it, and it was not a trivial time or monetary investment (still cheaper than the commercial option though). The net outcome is that at achieving phase 1 completion, I have a working prototype. I can monitor a single plant, indoors, relatively close to a power outlet, and with a bunch of wires and bits and bobs all over that wiggle loose if you look at them too hard; and that's kind of the big problem. Not all IoT projects are supposed to be indoors with "plug-in" power supplies and exposed wires; it just isn't feasible. The reality of "build it yourself" IoT is that unless you already like technology and making things, and don't mind that charming "cobbled together industrial mess" aesthetic, it is incredibly unlikely you would ever really get into it. If you do, it is like running a race where the finish line moves every time you reach it. I intend to continue on, but if you are considering your own IoT initiative you should know what you are signing up for up front!
Ultimately the end game for any device is that I will want to operationalize my prototype into something more consumer friendly for someone with potentially limited skill and technical know-how. The benefit to me for my little project is that the end product becomes something I don't have to babysit every day, but something I can turn on and forget until it reminds me it is there.
No big deal, right?
At this point it hit me why the whole concept of the IoT is both amazing in its potential to be something incredible and something incredibly destructive. You'll notice that I footnoted a couple of points above, relating to things I genuinely don't know how to do (yet). There are resources out there for a lot of those things, maybe even all of it. The best part about our modern world is that you are very rarely the first to experiment with something. There is usually a life line out there, unless you are performing bleeding edge innovation. This is why there is so much potential in IoT. You are taking work done by others and enhancing, improving, or simply modifying it to match your need. This is the heart of the open source movement and IoT is no different. Nothing really starts at "zero" and you don't have to figure out every single detail. I didn't have to build the sensors themselves. I certainly didn't design the Raspberry Pi, or create a language to use to talk to the sensors. I am using things already discovered in a new(ish) way for my specific project. That is an amazing thing to be able to do, and more importantly it means that innovation in IoT can happen sometimes frighteningly fast.
As I went through this project, it was easy for me to see why defaults work their way into commercial IoT devices so easily; no one wants to have to hassle with multi-factor security or cumbersome passwords during the prototype phase when things are constantly changing. Unfortunately I know the demands to get things "out the door" all too well and in that rush is when defaults, bad code, and "it's working, ship it" intersect is when bad security happens. This is where consumer design and security clash, and IoT is one of those rare places that we are collectively going to need to get right, because the stakes are getting really high. In my case, my plants might not get watered if my rig is compromised. No big deal for me, I will hand water them again, but if it is my car, my fire and smoke alarm, my medical equipment, the controls for my elevator, or any number of other things that are connected to the Internet, we could be talking about serious or even life threatening risks.
The second is that IoT is not a simple exercise in "connecting things together" and linking it to the Internet. The amount of knowledge and skills required to make a true commercial IoT device functional and successful is a legitimate challenge. The barriers to entry are high, even for someone like me that enjoys learning and discovering and has decent technical background.
Can we accelerate the ability of IoT technologies, but keep it engaging and approachable without making them so simplistic and unsafe, that we create a new army of devices that wield increasing power over our daily lives? I truly believe we can, and I hope that in a small way my own IoT project is one of many small bridges leading to plan to figuring out the how of the technology, so that others will take my work and make it that much better.
Watch out for the next part of this blog Do It Yourself IoT blog series. As we go forward I will keep updating and sharing my learnings as I move into phase 2 and try to make my vision come to life. The video for the final prototype is over on YouTube, here.
* - I toyed with "MacGuyvering" the housing out of things like outdoor electrical boxes and such, but then I also started investigating things like 3D printing options which could allow me to print a case that fits my package of sensors and connections perfectly because it is built for it. If I wanted to get really serious about it and sell them, I could invest in my own printer; for 4-500$ you can get a simple but reliable 3d printer that would do what I needed. As I was researching I also found options for "on demand" 3d printing services too. A great resource list is here: https://all3dp.com/best-online-3d-printing-service/
** - I have zero experience designing PCB boards. I have some EE friends, but I want to learn on my own for the full experience. In doing some research I found a number of places that will "print" you a PCB that you design, and a couple even have an online tool for designing the PCB itself. I am using EasyEDA (https://easyeda.com/) which has a nice library of user-submitted designs and "templates" that help a total n00b like me get started. (I could, were I so inclined, go so far as to print my own Raspberry Pi motherboard in effect and truly slim down my design, but that seems like a LOT of extra work for a "fun" project, but it could be done - just not by me!)
*** - It turns out that while making things talk over distance is pretty standard tech, miniaturizing it and making it work the way you want is not. I have started delving into the things available for this and found a few options. I have NOT played with any of them yet, but this one I found on Amazon might do the trick: https://www.amazon.com/Arduino-NRF24L01-2-4GHz-Wireless-Transceiver
I *could* assume the end consumer has a usable wifi network, but again, if we are talking about potential uses outside in a big yard or garden, their house-based wifi may not cut it. Also I want to be conscious of security so having my own send and receive function that does not rely on something external means that I don't expose anything outside of the devices themselves.
**** Something that keeps coming up for this project is that an Arduino is likely to be a much better option for running the sensors and sending the information to the central station (which could be the Raspberry Pi). The Arduino has a significantly lower power draw (A solar panel to run a single RPi for 24 hours a day would cost upwards of 150$ - not realistic for a distributed monitoring system!) and it can do the analog sensor readings natively which removes cost from the sensor package in the form of the MCP3008 chip I need to do analog conversions for the digital-only RPi. It is definitely something I will have to explore as I go into the next phase of the project. A very nice writeup on running a solar Raspberry Pi is here: http://www.voltaicsystems.com/blog/powering-a-raspberry-pi-from-solar-power/
The same site has a number of other write ups and important power consumption info for solar-based projects.
-- Check out the companion video to this blog post: On YouTube, here, along with the rest of the video series for this project. --
Subscribe to the World's Most Powerful Newsletters
Today's Top Reads