The article was based on the text of the author’s speech at the Project pitch-session DataFest6 , which was held on May 10-11 in Moscow.
We were all once children, many of us already have our own children, or at least have experience with small children. Therefore, we can imagine how a baby who is unable to speak tells adults what is happening to him. First, an adult pays attention to the baby, sees his appearance, complexion, perhaps touches his forehead and, based on the results of such actions, makes a decision: what needs to be done to make the child feel good.
We formalize the description of the process:
- remove data
- we substitute them into the model (which is in our head),
- Based on the results of the simulation, we take action.
What is important: the child himself does not give orders to anyone, adults only look at biometric data and as a result the child receives actions that return him to the comfort zone.
This is an illustration of the statement: a person can control automata not only through movements or through thought (through reading of neural impulses), but also through bio-parameters. Or more mundane: a life support system can be built on the basis of a cycle with biometric feedback .
First, remember what is Smart Homeor smart home. Wikipedia writes that this is an automation system for a residential or commercial building. For example, leaving the house, you close the front door, and at the same time, the house turns off the electricity, water and the alarm is connected. Now all the elements of the Smart Home are working in a tough scenario. That is, so that the air conditioner included in the Smart Home system ceases to heat, and begins to cool the air, it must be reprogrammed. If an adult, capable person is engaged in reprogramming, then for him it will simply not be very convenient. But imagine that we are talking, for example, about small children who alone play in the room, or about an incapacitated person who, in principle, is not able to do something on his own, or simply about a sleeping person to whom the same air conditioner directs its air flow…
Sometimes the actions of “human assistants” working in a tough scenario can more likely harm than help a person. It is clear that a system that itself reads the needs of the owner would be much more convenient.
A variant of such a system is shown in the figure. This is a business idea, the material elements of which are already largely developed.
In the diagram at the beginning of the chain is an expert. He sets the rules of work, after which the server gives the order to the climate system to start working. The climate system regulates air humidity, temperature, air flow rate, light exposure. As a result of the operation of this system, the parameters of the home climate change, which affects the person. Further along the chain, data from a person and from the external environment (both from home and from the street) go to the server, are processed and the server gives commands based on new data. The cycle is closed.
Sensors on a person provide biometric feedback and allow you to get from the climate system such home climate parameters that allow a person to feel comfortable.
Key points of the scheme
The key points of the circuit are the sensors and the control unit .
What sensors are convenient to use for a person? It seems that for this role one should try to use the modified analogues of fitness bracelets that are worn on the wrist to measure bio-data during sports.
Now ordinary fitness bracelets are measured (this list is most likely not complete):
- breathing rhythm
- perspiration rate
- blood sugar
- heart rate
- the proportion of adipose tissue
- arterial pressure,
- blood oxygen saturation,
- body water balance,
- and know how to send this data over Wi-Fi.
The second key point of the circuit is the control block. The task of managing the system is reduced to modeling the relationship of human parameters and environmental parameters. Man, if considered as a generator of biological parameters, is a system that is stable in the range. Stability is the ability of a system to maintain its current state under the influence of external influences. Each person has his own individual set of combinations of parameters that maintain stability in a certain range. Presumably, a set of points from either the parameter space or phase space pointsassociated with these parameters (phase trajectory), describes the dynamic state of comfort, its own for each individual person. That is, such a model, by definition, should take into account the individual preferences of people.
It seems that it is most expedient to create a model of the relationship between human parameters and environmental parameters using machine learning .
Now the main problems associated with machine learning in the project are:
- how will the model be trained,
- how to reduce training time.
A separate consideration deserves the search for model training options that will depend, in particular, on the nature of human parameters that have the most serious impact in the framework of feedback.
The question of reducing the model training time is most likely solved by using a pre- trained system that will be retrained by the user during operation.
Simpler issues, for example, related to the simultaneous stay of several people in the room (whose preferences to choose) and other issues of the “managerial” level are easily resolved by manually configuring the system with the user.
Similar systems with biometrics can be used widely: they are suitable for many rooms where a person is for a long time. These are private houses, pensions, enclosed spaces, cars, cabins of water vessels.
Why is the commercialization of this business idea interesting? There are several related answers to this question.
There is a lot of writing in the business literature about the aspects of digitalization. Old ways to promote products are becoming ineffective and the business is looking for new approaches. The world of business speaks of customization - the transition from a group approach to individual production. With an individual approach, products are not made “for-all” ( fit-for-all ), they are made for each individual person. Customization is cost-effective for manufacturers, as it provides a competitive advantage by creating higher value for the consumer.
Development of a system with biometric feedback will allow us to customize our living space: an apartment, a house, a car, a watercraft.
An important factor in any development is its degree of readiness and cost . And here the project also looks advantageous: the material components of the presented scheme - the climate system and sensors - do not need to be developed from scratch, in fact, it will be necessary to adapt existing samples to specific needs. Significant development requires only soft: creating a model on which the system will work.
We give an example. I recall a graph from one study of the main trends in the field of Artificial Intelligence. The graph shows that some of the AI trends are already widely used by business (High Level of Business Adoption). But attention is attracted by other areas of research, which are less interesting to business and are conditionally called Experimental. Considering these trends, you are surprised to understand that they have been heard for a long time, but business still does not use them actively. From the point of view of the author, this fact clearly shows the importance of choosing a research topic for subsequent business success. It is the correct topic that to a large extent ensures a high speed of transition from the “experimental” stage of the project to the “production” one. Choice for theme development,
And finally: when considering the economy of a product, it must be understood that the person who develops this project receives not one product, but a ready-made platform for a number of applications.
It seems interesting to the author to model human-related systems. In particular, it is of interest that with the mass use of systems with biometrics, over time, information will accumulate on the relationship of human bio-data with environmental parameters. Such information will be valuable material for further work.