Research and development
Passion for technology and innovation are our driving force. We are always at the forefront when it comes to testing new technologies and searching for solutions outside the box.
The research for us
Our customers high expectations when they come to us and our objective is to cover their needs through solutions that are "unthinkable".
But to propose solutions, apparently not ordinary, it is necessary to have already studied and tested them.
A certain Aristotle said that «What we must learn to do, we learn by doing».
At Alexide we do this, we experience everything on ourselves to bring safe results into the real world.
On this page you will find some examples of our research.
A building based on AI
Our headquarters is dominated by domotics and there are no physical switches. The most "frightening" part is that the Artificial Intelligence that we have developed makes the decisions.
As data experts we collect all data from the more than 300 sensors in our premises (temperature, CO2, humidity, brightness, ...). Then we give them to a neural net with the aim of figuring out something.
This neural network drives HVAC (Heating, Ventilation and Air Conditioning). For example, in the middle of winter, it decides the moment when to turn on the heating to have at 9 o'clock in the morning 22 Âş C (in 2022/2023 I think we will train it on the 20 ÂşS). It does this based on time series, current situation, weather forecast, etc. The network has understood that the weekend we are closed and does not need the heating on, that on Monday you have to turn on a little earlier since for 2 days everything was turned off, If it is damp you have to anticipate a little ignition, etc.
Machine learning
Another neural network with a simpler task decides the ideal light for each person. In fact, those who are younger need less light than an old man and there are, in general, different habits. The network understands that after 18, and after the inevitable play at football table, there is an event that turns off all the lights (for safety reasons we keep the reserve on the type of event).
These are some examples of applying machine learning techniques to our daily lives. We confess that Artificial Intelligence still loses a few shots every now and then but we have plan B. For not being masochists we can always decide whether to switch to the manual. old school model!
Recognition of people via neural network
We decided to count ourselves, to know how many people there are in a room, how many in the academy space, how many in the open space, how many in the relax area... and to do so through anonymous data collected by cameras.
By using commonly used machine learning techniques (tensorflow docet) we count each other and the system also works well.
What’s the point? Well, you’ll soon find out because this technique will also find a place within SolidRules.
Object recognition and augmented reality
It is a subject on which we have done many experiments. Some were in the direction of making 3D objects appear in a video stream taken from a normal smartphone. Let’s just say that we were doing this long before it was cleared by Pokémon Go game.
But our goal is to do everything without even using an App and do it directly from browser, but on this the Web technology is not ready yet for our "visions".
We then had other experiences on neural networks and object recognition. We found that it is easy to distinguish a person from a cat and on that we had great successes. It is much more difficult to recognize a M10x16 screw from an M10x25 but we do not give up!
Recognition to find similar items
Finding similar objects is a fascinating and strategic theme for us (but not only for us). You can do it by geometry, you can do it by descriptions and characteristics and so on.
Here are two cases we are applying it:
- When a report of a customer arrives, can you categorize or assign automatically a priority based on the content?
- When a report arrives, is it already possible to show tickets that contain similar things to facilitate the response process?
If you don’t see them yet as a feature of SolidRules it means we haven’t reached the acceptable level yet but we’re not that far away!
Blockchain for smart contracts
Blockchain is a word to use in a job interview, a bit like machine learning. It’s "cool," and it looks like it can solve the problems of humanity. To find out if it is all mouth and no trousers we studied it and tried to use it to manage commercial contracts.
The blockchain is unchangeable and you can’t play games, you can’t "whiten" an error, and you can’t deny it either. In some cases it can be really useful. From contracts to software licenses and maybe even for an approval workflow.
But we’re pretty sure it’s not going to solve the problems of humanity, and it’s too slow and energy-intensive for now. Two things that aren’t going too well today.
Questions? We are here for you.
Fill out the form or send us an email to info@alexide.com. We will contact you to provide you all the solutions.