The term Artificial Intelligence (AI) was introduced in 1956, and since then we can clearly see the growth curve of popularity of this field of science. But why does it seem to have become a trend of the last few years? Well, simply because, before this, we didn't have as much data as is available now. Machines successfully understanding human speech, self-operating cars, transfers from an online shop, chatbots, and even newsletters - all involve AI.
AI contains many subfields; well known, for example, are:
Natural language processing: connects human speech (or writing) with computer language.
Deep learning: a function that imitates the activity of the human brain and helps devices to mend themselves.
Machine learning: helps a device improve itself through experience and accessible data. In CN Group we are working on an exciting project involving machine learning
Since 2006 we have been working in the agile way. Agile’s ability to bring you value in the very early stages of a project is one of the reasons we recommend it. We believe that thanks to fast data analysis and our sophisticated usage of Machine Learning tools, you can get the most of it.
What we are currently working on
CN Group was approached by a global company developing a medical aesthetic product, who were facing a problem with its positioning in space. This product contains an IMU sensor; a gyroscope and accelerometer provide information that can be used for tracking orientation and speed of the device.
How does it work? We all know and use GPS; however, it can only be used in places where a GPS signal can be received. It does not work in buildings, the subway or in a tunnel. As well as GPS, we also have IMU sensors in our smartphones where the IMU sensor helps users to navigate in space when GPS is not available.
Our goal is to assemble a device which can inform its user whether they are using the device correctly. There is the need to estimate the device's position and movement over a very short distance, centimetres in this particular case. So, we used ‘state of the art’ Machine Learning algorithms and neural networks to achieve the goal. Our next step was orientation tracking, processing data from the IMU sensor. Orientation is aligned to a camera-based system which provides ground truth motion signals in 3-dimensional space.
Speed estimation processing data from IMU sensors is achieved employing two different approaches:
This is partly a research activity in which you do a lot of testing, while exploring and looking for new approaches and ways to work. There is always one challenge in Machine Learning projects: to get the Ground truth data that help data scientists compare their results with reality. Despite this challenge, we believe that we are working on a unique project that we hope to present to you next year. We are currently testing the data, developing the application and we are really looking forward to telling you more once it all goes live. So, stay tuned.