Apply HN: Undercontrol.io – IoT platform powered by affordable hardware and AI | Hacker News Reader

Apply HN: Undercontrol.io – IoT platform powered by affordable hardware and AI

10 pointsdaraosn11 comments

Undercontrol.io is an IoT platform that provides an open-source SDK and plug & play hardware to easily connect sensors and actuators for monitoring data, controlling processes and triggering actions. Our aim is to optimize processes by implementing machine learning with the data gathered from the devices connected to our platform. To achieve this we want to offer affordable hardware and turn it into a commodity to reduce the barriers to entry of implementing IoT solutions.

We're excited to announce that we JUST launched out of beta and are on production, we saw this opportunity as a way to get feedback from the community.

Feel free to ask us anything and if you are looking to make your new IoT project with us, we are very happy to help guiding you on how the platform works. We currently support MQTT, HTTP GET, POST requests and email alarms too.

Also, we have a bunch of sensors and actuators, if you have any interesting idea let us know and we can develop it now as we speak :-) (can provide a video).

Website: http://undercontrol.io Platform: http://platform.undercontrol.io

> Our aim is to optimize processes by implementing machine learning with the data gathered from the devices connected to our platform.

What does this mean and why do I need it (assuming I'm an IoT manufacturer)?

Two aspects are fundamental regarding this:

1. As a platform, we can take advantage of the network effect. The data we gather from all the users using it, plus the interaction they do when we classify certain patterns can be used to improve our machine learning algorithm. Thus, you as an IoT manufacturer can take advantage of a trained algorithm instead of you developing one from scratch.

2. This network effect brings in another discussion which is privacy and security, we are very aware about this. The data is never shared with third parties and it's only used for training/improving our algorithms, we protect user data, because we know this is a big concern that will surge with IoT in the future. We are planning to implement end-to-end encryption and open-source the core code at some point so you can deploy your own local applications.

Thus, you as an IoT manufacturer can take advantage of a trained algorithm instead of you developing one from scratch.

This puts you at a strict disadvantage--without your own IP, you basically are just a hardware manufacturer, and that race to the bottom is already pretty well finished due to Shenzhen and whatnot.

We are not implementing the algorithm in the hardware, it's on our servers, hardware is lightweight logic-wise, we use it for sensing or actuating, but the brain is in-house, therefore no risks with IPs.

Will companies using this be able to: 1) implement their own algorithms? 2) retain exclusive ownership of their data? 3) host your platform locally?

Could you give an example of a process you can optimize using the data you gather?

Hey Anton, don't know if you remember me, we met in a hardware meetup in Berlin some time ago (Diego, drones). :-)

To your question, one example: - detecting anomalies in production lines of industries: we are talking with a potential client that needs to monitor the down-time in their plastic recycling machines. Using the input generated by different sensors, it's possible to detect failures and restart the machines if something fails in the process.

other applications: - monitoring store's product stock in real-time to optimize the restocking process. - energy saving in buildings, industries, homes. - security and health monitoring. - and of course, home automation by providing affordable hardware.

Oh hey Diego, didn't recognise your username! Great to see you applying to the fellowship, you definitely have the right 'hacker' mentality. (Just for the rest of HN's reference, Diego built a rig where you could watch yourself walk around in 3rd person like in a video game, using an Oculus rift and a camera drone).

Back to feedback:

- Linemetrics has been doing this sort of thing for industry for a little while, though more around energy usage. https://www.linemetrics.com/de/ If you already have a paying / letter of intent client for your system, you should have that front and center in your application.

- Where does the AI part come in? Those seem like fairly straightforward monitoring applications. What you wrote suggests that you would be able to tell customers what they should do about the anomalies/waste you detect, how do you plan to do that?

- Which of the industries you mentioned do you want to go after first?

Thanks for the props Anton!

- Yes, there are many actors in the market of smart building, we are currently talking with another one, cannot disclosure though.

- We want to have AI in the DNA of our platform, the first stage is collection of data, we can already do that with our platform, now we will focus in developing the first approach of pattern recognition for the data we receive from sensors, so we can identify clusters, then ask the users to tag data to train our classifier, and build from there our machine learning component.

- At the moment, we are focused on the platform. We are staying wide and open to developers and enthusiasts (horizontal approach), we want to be flexible, use open source technologies (currently supporting Arduino and ESP8266, and common sensors and actuators), we will explore some verticals on the way, we hope this approach will help us make the platform bigger.

Why would your potential customers choose your platform over others before you have the fancy A.I DNA?

We are making a huge effort in keeping things very simple, we want to lower the barriers to entry by offering affordable hardware in two aspects: price and ease to implement-use (plug n play).

All this started over a year ago, we worked on some IoT projects, we realized we needed a platform to make it real, but none of the existing were simple and flexible enough for our needs.

The one project that kickstarted Undercontrol.io was monitoring the temperature of the fish tank we have at home, after some fish died due to an issue with the heater we started monitoring the temperature and receiving alerts when something went wrong. Sure, you can buy fancy sensors, but they are expensive, we already had all the components and the technology from other applications at home. We believe IoT should be accessible for everyone, you have the example of Arduino, Raspberry Pi, NodeMCU and so many others. All of those are very useful developing tools, but we were missing a powerful workshop for creating real applications with all these tools.

From here we want to unleash the potential we believe our idea has, but baby steps first. :-)

Reply on news.ycombinator.com