Sometimes when we are running, our watch automatically identifies the activity. Other times it does it when we ride a bike, or when we swim, but the number of activities that can be recognized automatically is very limited.
It seems that this will change with a new sensor that is capable of recognizing practically any activity, from jumping to push-ups to, of course, those mentioned above.
It has been Bosch that has developed a new autonomous chip that can be taught to recognize any activity using Artificial Intelligence. In other words, you not only recognize activities, you can learn to recognize new ones, so that anyone could “train” you to detect and record each individual exercise in your routine.
How are the physical activity sensors
To recognize an activity, the sensors, which were at first just step counters, used technologies like GPS and sophisticated gyros to let runners and runners know how far and fast they were going.
As the processors improved, they began to recognize swimming and cycling movements, using the data collected by the sensors to analyze them in real time.
The problem is that there are many ways to swim, for example, which is why Bosch Sensortec has created this new motion sensor, to differentiate very specific movements and automatically recognize them later.
How this new sensor can be used
The self-learning AI sensor BHI260AP It could be included in the activity trackers of different manufacturers. It’s small enough to fit into portable devices and even wireless headphones.
Right out of the box it comes with 15 different previously learned fitness activities that can also be adjusted for technique or style given by a user.
The motion sensor has built-in artificial intelligence, so the data doesn’t have to be processed on a connected smartphone and uploaded to the cloud for analysis.
As they do not have to go to the cloud, the data will be more private, something essential in professional training.
On the other hand, the motion sensor is smart enough to recognize an activity without having to select it previously.