IT202100005999A1 - APPARATUS AND METHOD FOR THE DETECTION, RECOGNITION AND INTERVENTION TO CONTRAST PHYSICAL VIOLENCE TO THE DAMAGE OF PEOPLE. - Google Patents

Description

DESCRIPTION of the patent application for industrial invention having object and title: ?Apparatus and method for detecting, recognizing and intervening to contrast physical violence against people?

Technical field

The found object of the invention pertains to the field of personal security systems for the prevention and intervention against violent crimes perpetrated, both in the domestic and in the public sphere, against the most vulnerable people. weak and unable to defend themselves and particularly pertains to wearable devices for protection and intervention in the event of physical violence against the user even if he is temporarily or permanently inhibited or unable, on a physical and/or psychological level, to defend himself or ask for help. Pi? precisely, the invention concerns an innovative and proactive system for the direct collection of kinetic signals and specific biometric data, capable of autonomously distinguishing situations of danger/violence against the user, and intervening by forwarding an alarm message , through channels and with different communication protocols, to a plurality? of entity? and users who have the possibility? to intervene to stop the violence and/or identify the person who perpetrated it. Purely by way of example, the proposed invention can be conveniently used to counter violent actions against all types of minorities (ethnic, religious, etc.), as an intervention tool against bullying in schools and public places of aggregation, to counter violence against the elderly at home or in structures of hospitalization and as a survey and intervention tool against violence against women. The system ? based on a specific wearable technology, integrated into specially designed clothing, which exploits conductive fabrics on which flexible sensors are made capable of detecting a multitude of different kinetic and biometric data, purely by way of non-binding example, nor? limiting factor, heart rate, perspiration, temperature, Galvanic Skin Response (GSR) and in particular the level of pressure exerted on the wearer's body. The choice of this technology? dictated by necessity? Of:

? Have an autonomous system that intervenes independently of the will?/possibility? of the user to do so;

? Having an invisible system that the attacker does not become aware of and therefore does not try to disable;

? Having a protection system without having to equip yourself with special devices: clothing and mobile devices are part of every person's daily life.

The above system? equipped with a removable electronic control unit, powered by a rechargeable battery, which collects said signals and processes them, normalizes them and forwards them to a combined electronic device, via wireless communication. Purely indicative, not binding, n? limiting, the electronic control unit ? equipped with communication modules using Bluetooth, Wi-fi, and NFC communication protocols. In the invention , a matched electronic device, ? a mobile electronic device, purely by way of non-binding example, a smartphone or tablet on which ? It is possible to install and run a data analysis and action execution application consistent with the analyzed context. The system, and in particular the mobile app, implements machine learning techniques and methods, based on supervised models, with a subjectivist probabilistic classification approach, capable of adapting and optimizing its output on the type of input data. The neural structure used? strictly dependent on the type of event and the data entered in the input layer: the identification of a situation of violence/danger implies the monitoring of heterogeneous variables such as, by way of example only, not limiting, biometric signals of the user (beat heart rate, respiratory rate, temperature, perspiration, GSR), kinetic signals (accelerations, rotations, position in space) and stochastic, impulsive, pressure signals detected by special sensors. In similar contexts, the use of common frequency statistical methods would be impossible for two reasons: frequency methods require datasets, i.e. they require very high data volumes, which multiple contexts do not allow, and they require a priori knowledge of the phenomenon, which is impossible to get in a scope where every ?event? it has its own inherent characteristics and differs from all others. By way of example, every violent action, aggression or malicious act implies its own boundary conditions, such as the biometric parameters of the victim which vary from person to person under the influence of emotions, moods, character, etc., preventing the use of objective statistical techniques.

The invention aims to overcome the limits of current systems dedicated to personal safety which have two major constraints: do they presuppose the ability? of voluntary activation of the alarm signal and are based on the use of an additional object to always have at hand, with needs of easy and immediate availability? at the time of need and operation ensured, or need? electric charging and accidental damage protection. Furthermore, it totally eliminates by design the problem of having to voluntarily activate the alarm, thanks to the use of flexible sensors integrated in the fabrics of the garments and a virtual application that exploits the electronic components present in modern smartphones to process the data acquired in real time. The invention, in its software component? can be integrated with devices on the market, without the need? replacement or modification.

State of art

To date, there are numerous alarm devices on the market for the management and signaling of dangerous situations or criminal events, and many others are presented daily in the form of concepts to be developed on internet platforms in search of funding for development/production. However, all of the systems described above are based on a button, or equivalent device, to be voluntarily pressed to activate a transmission system for an alarm signal. As an example we mention:

? Devices with reed contact button, designed for silent signaling of criminal events; pressing the button causes the closure of a microswitch, and the consequent activation of the alarm. These devices are generally equipped with self-protection against accidental activation.

? Push-button portable acoustic deterrents, in which pressing by the user activates the alarm, a loud sound generated by the device itself in order to disturb the aggressor and to attract the attention of people who can intervene. These devices are often battery powered and can be carried in pockets as well as being portable. as in bags or hooked to a set of keys.

? Silent listening devices, in which pressing a button generates contact with an operations center where an operator listens to the environment, alerting a mobile patrol in the event of a positive match for a criminal or violent act.

? GPS anti-aggression devices, which interface directly with the GPS system of smartphones via Bluetooth connection and a special app. Pressing the button activates the alarm signal via text message, email or telephone call, containing the user?s geographical position.

The most popular versions recent models of the latter type of device have an integrated GPS geolocation system and are able to directly transmit the link of the Google Maps service, containing the satellite coordinates of the device. even more recent are also the smart buttons linked to the? IoT, connected to a cloud platform and programmable to perform simple services, linked to pressing the button itself.

In the field of wearable devices, the anti-aggression solutions adopted are once again based on the generation of very loud sounds designed to stun the attacker and attract the attention of possible rescuers and on the physical impediment, through actual mechanical blocks, to perpetrate the violence. As an example:

? Project ?awese, a wereable security?, promoted by Fbk and Centro Moda Canossa in 2016, in which garments equipped with sensors that can be activated by the victims and capable of reproducing an alarm at 130 decibels to disorient the aggressor were presented.

? A line of underwear conceived by AR wear, equipped with a series of belts and blocks to be placed on the thighs and waist and which can only be unlocked by means of a specific combination; the fabrics used are cut and tear resistant.

? Underwear made by a German designer based on the use of a special combination lock which makes it highly difficult to remove.

The solutions currently available therefore require the purchase of dedicated devices (button systems) or are based on cumbersome and potentially discriminating solutions (clothing equipped with mechanical blocks) which can slow down, if not prevent, their spread. However, over 80% of people living in economically developed countries own and use mobile devices on a daily basis, including the ?weak? and ?fragile?. The result is a significant advantage: the possibility? to forward, through pi? channels, a request for help to the single emergency numbers both for crimes or violent actions, and for health emergencies. In dangerous conditions, for example during an attack, or in the event of illness, it is not, however, always possible for the user to request help by pressing a button or making a call. On the other hand, the technological evolution has enriched mobile devices with functionalities? accessories that go beyond the simple possibility? to make calls: current smartphones, in fact, are comparable to small computers, for capacity? and computing power, being able to process data, even complex ones, send and manage emails, documents and files of different types; they are also equipped with different types of sensors, global positioning systems, GPS, and GSM/UMTS data transmission systems, Bluetooth protocols and proximity data sharing.

Phone operating systems are therefore capable of running a multitude of programs, ?Applications?, that provide functionality additional, using the resources of smartphones such as, for example, the use of CPU and GPU for processing complex data in real time, the use of GPS to record the position of the device or short-range communication (Bluetooth ) between different telephones within a common action area to create a mesh-type network in which a precoded message can pass through all the connected devices.

Furthermore, in recent years, research applied to sport, both amateur and competitive, and to computer graphics (3D animations, special effects, AR, VR, etc.) has led to the miniaturization of many types of sensors (accelerometers, gyroscopes, magnetometers , temperature, respiratory rate, etc.), useful for monitoring movement and activity? physics, making them more? comfortable and easy to wear. By way of example, we cite:

? BEAST SENSOR, device with 3 accelerometers, 3 gyroscopes and 3 magnetometers, easy to wear at the wrist and upper back. Free to use during training in the gym, it connects via Bluetooth to Android and iOS smartphones.

? Xiaomi Zenph Smart Fitness Sportswear, smart shirt with Intexar sensor, made with DuPont Graphite Technology, constantly monitors heart rate, pressure, respiratory rate and body movements, activating an alarm in case of need.

? Xsens suit for recording body movements, made in two variants dedicated to 3D animation and the development of man-machine interfaces, kinesiological studies, etc. Recently, new studies have led to the creation of smart fabrics that incorporate flexible sensors into the yarn, eliminating the need for smart fabrics. to integrate a rigid electronic device into the garment.

Although there is therefore the possibility? technique of creating an anti-aggression alarm system made up of an item of clothing that detects biometric parameters and kinetic data in real time thanks to the presence of appropriate sensors and a mobile device that acts as an actuator of the signal, thanks to a code that can be executed with an algorithm of recognition of situations of violence or malaise, none found so far ? was designed to activate an alarm independently and without requiring the will? and the active intervention of the user.

The procedure that implements the biometric and kinetic data interpretation algorithm and activates the multi-channel forwarding of the alarm message to a plurality of of users, can? be trained on the specific biometric characteristics of the user and calibrated on the danger thresholds of the kinetic data detected and therefore made extremely specific and reliable, in order to avoid ?false positives?.

Purposes of the invention

To overcome the aforementioned limits and for obvious reasons of usability, feasibility? and comfort? of use with respect to the solutions mentioned above, we want to create a system that is able to:

? Autonomously and automatically detect a series of data and signals correlated to specific risk factors and attributable, directly or indirectly, to episodes of violence. By way of non-binding example, n? limiting, these signals and data relate to body position, accelerations suffered, precise or widespread pressure, instantaneous and prolonged over time, exerted on one or more? parts of the body, and biometric parameters (heart rate, ECG, respiratory rate, temperature, sweating, galvanic epithelial response, etc.). The detection must take place in every situation in which the user finds himself, whether it is an internal or external environment, both in static conditions of calm, and in conditions of agitation and motion;

? Autonomously and automatically recognize situations of danger/violence affecting the user; it must be able to process the aforementioned signals on site, cross-referencing data relating to different fields, to reconstruct what is happening, and compare them with specially created and memorized matching datasets. Purely by way of non-limiting example, it must be able to understand whether the user? victim of beatings and/or beatings, of an attempted sexual assault, if he is involved in scuffles or brawls, if ? been pushed and/or ? fell to the ground and, indirectly, if he suffered injuries or injuries;

? Intervene autonomously and automatically to activate the necessary assistance. Victims of violence are often physically and/or mentally unable, temporarily or permanently, to defend themselves and ask for help. A system that goes beyond the limits of current technology must be able to forward an alarm signal if a situation attributable to danger and/or violence is recognized. In order to maximize the forwarding of the alarm message and reach as many possible recipients, this system must be able to communicate wirelessly with other devices on which the aforementioned message can be displayed. By way of example, non-binding, n? limiting, the system must be able to communicate with Bluetooth, Wi-Fi, NFC, LoRA WAN, GSM/LTE protocol etc.;

? Create a network between systems of the same type to overcome the impossibility? to transmit the alarm signal: a system according to the present invention, based on wireless communications to maximize its flexibility? and feasibility, it must also be able to make up for any problems related to wireless communications, to always guarantee the forwarding of the alarm.

? Providing evidence of the crime and/or violence: in addition to the intervention to block the criminal act, ? useful to be able to collect evidence and documentation against the events that occurred, to be used, subsequently, in the appropriate offices. For this purpose, a prevention system according to the present invention must be capable, once the violence has been ascertained, of recording audio and video material, of transferring it in a safe and protected manner to a storage environment and of keeping it for as long as necessary, protecting it from tampering, theft and/or deletion attempts;

? Be easy to use by the user: a system that exceeds the limits described above must be activated and usable by the user without too much complexity, preliminary actions or installation/startup processes; in other words, a system according to the present invention must always accompany the user, without him realizing it, without him having to search for it or voluntarily activate it or having to frequently check its operating and/or operational status;

? Be concealable: the attacker or attacker who becomes aware of the presence of a safety device or of the attempt to activate it by the user, will attempt to to prevent its use. A system according to the present invention must be totally invisible to the eye and its use must not imply any action on the part of the user, which explicitly indicates its presence and activation, in order to that it is effective and can fully fulfill its purpose. A hidden system operates continuously with less risk of being disabled/blocked;

A system like this? done allows you to obtain the following advantages:

? ? highly scalable: a wearable and ?transparent? for the customer, i.e. totally autonomous and automatic, ? usable and applicable to a plurality? of users and different situations, being able to intervene on a?wide variety? of violence and gravity crimes different;

? ? difficult to deactivate: a system that is not seen can not? be identified and neutralized by the aggressor, therefore it always remains operational in case of emergency;

? ? factual: the invention, for construction and design, ? able to intervene in almost all? of the conditions under which it is activated and violence is detected. This aspect places it above any device and system currently in circulation since? guarantees intervention where current systems could not, extending and maximizing its effectiveness;

? It guarantees the forwarding of the alarm: a system according to the present invention bases the forwarding of the request for help on wireless communications between a paired device and the addressee; however, at the basis of the finding c?? also the creation of a mesh network between devices on which ? installed system management software. Creating a network between devices minimizes the risk of blocking the alarm signal, since ? able to compensate for any malfunctions, breaks or breakdowns of a single device.

Brief exposition of the invention

The solution to the above problems is achieved through a system made up of an item of clothing which:

? Acquire autonomously and automatically heterogeneous signals; by way of non-limiting example, n? binding, said signals can be biometric, kinetic (state of movement/stationarity, changes of direction of motion, accelerations, impacts and localized pressures, spatial (relative and absolute position and its variations)). The acquisition and sampling of each individual signal must take place net of electromagnetic interference on the same frequencies of use, attributable to other electromagnetic devices and/or to the user's own behaviour. By way of example, the item of clothing will be? capable of distinguishing localized impacts and pressures on the user's body, filtering the signal of muscle movement.

? Once these data have been edited, on site, standardizing them according to a single structure, to make them usable by a single program in order to process them, analyze them and extract behaviours, similarities? and patterns that allow recognition and identification.

? Acquire and edit the aforementioned data in a totally anonymous and invisible way so as not to be directly and clearly identifiable as a personal security system. The item of clothing will be made like any other passive item of clothing and, consequently, should not? bear evidence of sensors, electronic processing and communication devices or other apparatus that may indicate a purpose different from that typical of any dress, the covering of the human body. In this way the system will be more effective and effective since? cannot be neutralized by the aggressor.

? Be comfortable: the data acquisition, processing and forwarding systems must be minimally invasive and must not disturb the user; the garment must therefore be comfortably wearable in all circumstances and without limitations. For this purpose, the electronic subsystems will be totally integrated, flexible and miniaturized so as not to interfere with the normal and daily use of the garment.

A program that:

? ? able to be performed on the pi? common mobile devices on the market, by way of example not limiting tablets and smartphones, without any of its functionality? is slowed down, stopped or not performed why? computationally too demanding for the device itself;

? ? based on a data analysis method that ? effective even with a small number of input data, i.e. managed by a logic capable of:

o Understand, with few resources, the nature of the data;

o Extrapolate useful information (characteristics, similarities, trends, etc.);

o Put them in relation with those of the other input data;

o Obtain analyzes and classifications.

? ? able to distinguish and identify contexts, scenarios and what? that happens to the user by correlating the results of the analyzes performed on data of a heterogeneous nature, but all underlying the same situation. In other words, by analyzing and comparing the data mentioned above by way of example, the program ? able to identify the state of motion or stillness of the user (standing, sitting, walking, running, dancing, etc.), his emotional state (agitation, nervous, calm, fear, excitement, etc.), activity ? impulsive (jumps, falls, blows or traumas), discerning a safe situation from a dangerous one;

? In the event of conditions compatible with a dangerous situation, purely by way of non-limiting example, nor? binding, beatings, assaults, falls, accidents, illnesses, etc., ? able to forward an alarm signal to alert people who may, for various reasons, intervene to change the situation and provide assistance; ? Must be programmed with the data of one or more? recipients to forward the alarm signal to; said signal must be sent by means of different forwarding methods, purely by way of non-binding example, nor? limiting, pre-populated text messages, pre-recorded voice calls, mail with pre-set text, etc.

? ? able to forward the? alarm simultaneously to more? recipients through the use of the pi? common wireless communication protocols; by way of example, BLE, Wi-Fi, LTE, NFC.

? ? capable of forming a distributed mesh network between different devices that are located within the same range, in order to overcome any transmission problems of the device that generates the alarm. In particular, if malfunctions and / or breakdowns or, more? in general, impeding circumstances block direct forwarding of the alarm from the device that generates it, the program must search for other devices, within its range, that have copies of the program itself (that belong to the same ?community?) and exploit the their hardware resources for forwarding said signal. The search and connection with similar devices within the range of the one from which the alarm starts, contextually generate an alert in each device reached so that other people can detect the aggression and intervene promptly, further reducing rescue times. In other words, a program according to the following invention shares, with encrypted and protected communication, the data, logs and information necessary to forward the alarm and make anyone intervene, at the place of the attack itself or inside a circumscribed perimeter around it, has the same program.

? ? able, if the alarm is activated, to start audio and video recordings of what is happening, using the hardware of the device on which it is? installed. ? ? able to forward the recorded material, with protected and encrypted communication, to a cloud server where this material will be sent? archived awaiting possible subsequent use as evidence of the crime perpetrated. If there are impediments to direct forwarding, a program according to the present invention ? able to create a mesh network with other compatible devices in order to complete the upload of the acquired material.

? ? able to perform the tasks, analyzes and procedures of the previous points in total autonomy, without requiring the intervention of the user who suffers the aggression, since? it is common that the victim of violence is, temporarily or permanently, on a psychological and/or physical level, prevented from requesting help and assistance directly. ? ? able to refine their analyzes and to recognize and discern situations with ever greater accuracy, thanks to self-learning algorithms that exploit data sharing between compatible devices, i.e. devices on which ? installed a copy of the program.

A cloud platform that acts as a repository of the acquired recordings, also allowing to obtain an updated history of the logs and of the clothing items detections.

A system according to the present invention implements the following steps:

? Initial automatic calibration: during the dressing phase, the electronics inserted inside the clothes will acquire the user?s parameters from the biometric sensors and the status of the pressure and movement sensors, setting these values as the ?initial system setting?. Purely as a non-limiting example, n? binding, the state of the sensors ? referred to the percentage of pre-stretching and/or compression that the sensors themselves undergo when the garment is worn by a person. What? doing the system? always self-referenced and ? capable of recognizing anomalous parameters more effectively than an objective calibration performed during the assembly phase, with the sensors totally inactive;

? Launching the program installed on the mobile device: why? the system works once the garment has been worn, the user must voluntarily start the program to activate continuous monitoring;

? Pairing: once started, the program will execute? the pairing with the item of clothing, verifying the association keys assigned during the construction and assembly phase. A pairing completed, the program will acquire? from the garment the initial parameters and settings that will allow the system calibration and subsequent optimized use. At this stage the machine learning model ? engaged in the recognition and classification of the initial parameters of the dress, clustering them and using them as a dynamic dataset for subsequent analyses;

? Data acquisition: the sensors distributed in the garment record the signals from the user's body in real time and without interruption and the electronic control unit standardizes them and prepares them for forwarding to the mobile program. In this phase, the firmware on board the garment analyzes and maps the signals in order to eliminate interference; the approach differs according to the type of signal: for the biometric ones, the method ? based on high pass and low pass filters (purely by way of example not limiting, nor binding, the 40 and 50 Hz filters to exclude muscle movement) and on the extrapolation of signal formants, subject to significant variations based on the state psychophysical of the user. As for the kinetic signals, the method ? based on? differential analysis of pi? channels in parallel and with respect to the initial calibration threshold; this method allows, on the basis of the sensor and the type of signal to be sampled, to exclude non-useful deformations (purely by way of non-limiting example, for pressure signals if the differential of the channels of the x and y axes is not null, its value will be discarded);

? Sending standardized data: the data processed and edited in the previous step are sent, with a private protected and encrypted connection, to the device on which it is stored. installed the mobile program, which runs in the background, to limit energy consumption and bandwidth occupation;

? Reading of standardized data: the mobile program, having received the data packets, processes them to allow the analysis algorithms to read them and perform a matching with its own ?experience?, acquired during the initial training phase, and updated periodically, in search of similarities useful for the classification of the event;

? Classification of events and generation of expected outputs: the new input data are compared with the internal dataset of the program to classify the context and define the current situation; purely indicative, not limiting, n? binding, the biometric data are compared with sets of typical situations (state of agitation, fear, nervous, excitement, etc.), tabulated and loaded into the program, in order to find similarities. The pressure data on predefined areas of the body are compared with the preconfigured alert thresholds and related to other kinetic data (accelerations, falls, sudden changes in relative position, etc.). The combination of the results of these operations defines, probabilistically, the outline of the analyzed situation. The output of this step ? therefore a classification of the contingent situation, purely indicative, ?normal? situation, of ?potential danger? or of ?violence?, with indication more? detailed of what? that happens; purely by way of non-binding example, n? limiting, the? output can? indicate a situation of normality? and tranquillity? (physical activity or dancing), a risk situation (fear, frantic running, agitation) or aggression (beating, attempted rape, etc.). ? Forwarding of the alarm: in the event of a positive output due to a dangerous situation, the program activates a request for confirmation from the user to check his state of health and verify that there is no violence in progress). If the user does not respond to this request, the system activates the alarm by sending the alarm message to the previously selected contacts using the methods and means previously described. In case of consistency of the output with a condition of overt violence, the alarm is sent automatically without requesting confirmation from the user.

? Starting audio/video recording: in case of non-verification by the user or output compatible with violence, a program according to the present invention immediately activates the recording of audio and video files, exploiting the hardware of the device on which it is recorded. installed and run, in order to collect useful material to document, in the appropriate locations, the violence suffered by the user.

? Creation of a mesh network: if there are hardware impediments to forwarding the alarm, purely indicative, non-binding, nor? limiting, lack of connection to a data network, malfunction of one of the communication modules of the device on which the program is? running, etc. the program itself exploits the other communication protocols configured to search, within range, for other devices on which other copies of the same software are installed. If identified, will started the connection and pairing procedure between the devices, with asynchronous security keys; the devices will form a private communication mesh network and the alarm signal, the victim's reference contacts and all relevant data will be transmitted from the generating device to all the matched ones in such a way as to exploit their hardware for forwarding the request of help. At the same time, on the paired devices it will come sent a notification requesting the intervention of their owners, in order to reduce reaction times and avoid, if possible, serious consequences for the victim. ? Sending of the recorded files and the generated outputs to a dedicated repository on the cloud platform, to be used later as a history and as a source of information that proves and demonstrates the violence in the appropriate locations, if any.

Brief description of the attached drawings

Further characteristics and advantages of the proposed technical solution will be more evident in the following description of a preferred, but not exclusive, embodiment, represented, by way of non-limiting example, in the 12 attached tables of drawings, in which:

Fig.1 shows a simplified scheme of operation of the entire system with the representation of the connections between the item of clothing, the mobile device, the cloud platform and the devices of the registered contacts.

Figures 2 and 3 show the front and back of an item of clothing or garments according to the present invention with indication of the areas which can be used by way of example for inserting the sensors and the control electronics.

The Fig.4 and 5 represent in a more detailed the same views of the garment of the previous figures, with indications of the type and possible positioning of the sensors and with indication of the contacts for fixing the control electronics.

Fig.6 shows the detail of a preferred embodiment of the pressure sensors.

Figs. 7 and 8 schematically represent the wiring of the sensors, integrated in the garment.

Figs. 9, 10, 11, 12, 13 schematically represent the control unit with control electronics and its housing/protective case with annexed connections for recharging and anchoring to the garment.

Figs. 14, 15, 16 and 17 show in detail two possible configurations for fixing the electronic control unit to the garment.

Fig.18 shows a simplified diagram, by blocks, of the electronic card.

Fig. 19 represents the block diagram of the data acquisition and forwarding method according to the present invention.

Fig. 20 represents the block diagram of an automatic learning method for data analysis and for forwarding any alarm signal.

Fig.21 represents the data analysis logics employed according to the proposed embodiment. Fig.22 shows a simplified scheme for creating a mesh network between compatible devices, said network being usable for forwarding alarm messages.

Best way to implement the invention

With reference to the attached drawings, and in particular to Fig.1 thereof, ? represented the preferential form of implementation of a system for the detection, recognition and intervention to combat physical violence against people (100) in which the data pertinent to the classification and recognition of an act of violence, an illness or of a dangerous situation are collected in a discreet and confidential manner by means of an item of clothing or garment (101); said item of clothing (101) being characterized by sensored areas (102) equipped with sensors of different kinetic and/or biometric types whose data are collected and pre-processed by a control unit (103) and, after a first standardization, sent through a communication channel (104) to a mobile device (105) equipped with automatic learning software, said software being installable on commonly used mobile devices (105) and available without limitations on the market. Said software being usable to analyze and classify the signals collected in the sensorized areas (102) of the garment (101) in order to recognize the context, the situation and the type of interaction taking place. If the results of the classification are compatible with violence, malaise or a dangerous situation, the software starts recording audio-visual material that certifies what is happening and sends an alarm signal (106), with various methodologies and communication protocols , to contacts and numbers preset by the user during the first start-up. If, due to failures, malfunctions and/or impediments of a technical nature, direct forwarding of the alarm (106) is not possible, a system according to the present invention searches, within the range of the communication modules of the mobile device (105), further mobile devices compatible with it, i.e. devices bearing another installation of the same software, through which to alert entities? and the pre-registered contacts; once the alarm (106) has been sent, the recorded audio-video material is forwarded to a cloud server (107) for storage and subsequent recovery as evidence of the violent action or what happened. The communication (104) between the item of clothing (101) and the compatible devices carrying a copy of the associated software takes place through various protocols:

? BLUETOOTH LOW ENERGY (BTLE 4.0);

? WIFI (IEEE 802.11xx);

? NFC (13.56MHz);

In a non-limiting preferential way, n? binding, the communication between the item of clothing (101) and the combined mobile device (103) takes place, in the first instance, via BLUETOOTH protocol. If this type of communication is not possible, the garment tries to start a connection and consequent communication via Wi-Fi protocol and NFC again. If none of the attempts leads to pairing, the control electronics (103) located on the garment (101) alerts the user to allow him to intervene and restore at least one of the connections. The alert to the user takes place, in a non-binding preferential way, nor? limiting, by means of haptic signaling, i.e. vibration of a control unit (103) placed on the garment (101).

The forwarding of the alarm by the mobile device (105) which has identified a situation of violence takes place, in a preferential but not limiting way, using the following communication methods and protocols, in accordance with what is set by the user in the first phase launch of the software and compatibly with the possibilities? techniques of the mobile device (105) on which the program ? active:

? Sending a preset text message (108) to a phone number via GSM/LTE protocol;

? Activation of a voice call with pre-recorded message (109) via the GSM/LTE network;

? Sending an email with a pre-set text message (110) via GSM/LTE protocol or Wi-Fi via public or private internet networks.

All forwarded messages, regardless of the type and protocol used, will transmit the data of the person suffering the violence, the request for intervention and the geolocation of the mobile device (105) at the time the alarm signal is sent (106). Communication with the data cloud storage platform (107) takes place through the same channels and protocols, with preference for the GSM/LTE protocol over the internet one.

With reference to Figs. 2 and 3, the preferred embodiment of the present invention provides for an item of clothing (101) represented by way of example by a specially designed and manufactured short-sleeved bodysuit, which integrates, as anticipated, suitable and convenient areas sensors (102) equipped with kinetic and biometric sensors whose signals are collected through a distributed circuitry which conveys the signals to a collection center (111) to be subsequently processed by the control unit (103). The choice to create a short-sleeved bodysuit as a non-limiting but primary example? dictated by the will? to create a garment that is easy to wear and usability: in the preferred embodiment and use of the present invention, in fact, a short-sleeved bodysuit can be conveniently worn daily under normal clothes, in every season and occasion, thus providing a secure, continuous and invisible monitoring system to the eyes of a possible attacker. In alternate embodiments of the present invention, the garment (101) may be made up of a body made with long or sleeveless sleeves, combined with evening or formal wear. In at least one other alternative embodiment said garment (101) will be able to be a complete fitness format, purely by way of non-binding example, nor? limiting, from a sleeveless top and a pair of long leggings, to wear for the activity? indoor and outdoor physics.

The aforementioned sensorized areas (102) are intended to house sensors used to acquire signals of the kinetic and/or biometric type, said signals being subsequently collected, analysed, processed and used for monitoring and identifying situations of potential danger. Said sensors being distributed within said areas (112) according to specific technical requests aimed at obtaining:

? Biometric reference points: the biometric signals collected by the system require precise and unambiguous collection points that guarantee the acquisition of a useful, clean and usable signal.

? Kinetic reference points: the points for the collection of kinetic and pressure signals have been chosen on the basis of a study of the dynamics of the violence, of the interactions between the aggressor and the victim, of the most exposed parts of the body and those which are instinctively used for protection.

As a consequence of what has been described in the preferred embodiment of the present invention, the acquisition zones (102) equipped with sensors able to detect said kinetic and biometric signals are the following:

? Sternum-pectoral area, with the main localization on the under-pectoral and axillary girdle; ? Inguinal area, with the insertion areas of the joint between the leg and hip;

? Buttocks;

? Shoulder girdle and upper portion of the spine up to the base of the neck;

? In the preferred embodiment of the present invention, the upper arm band (brachial biceps and triceps);

? In at least one of the alternative embodiments of the present invention, the forearm area is added, instinctively used to protect the head and body in case of aggression.

With reference to the attached drawings and particularly to Figs. 4 and 5, they are represented in a more detailed the positioning and type of sensors used to acquire the aforementioned kinetic and biometric signals, said sensors being positioned for example according to the following preferential scheme in which:

Kinetic signals are collected by pressure or kinetic sensors (112) distributed on the garment (101) as follows:

o N. 8 sensors positioned along the lower edge of the pectorals in the lower area of the breast, 4 on each side;

o N. 6 sensors positioned in the groin area along the pelvis/leg junction line of the hip joint, 3 on each side;

o N. 4 sensors positioned on the humeral part of the arm, 2 anterior and 2 posterior; o N. 6 sensors positioned in the buttock area along the lower/rear edge of the body, in correspondence with the midline of the gluteus maximus. 3 per side;

o N. 6 sensors positioned along the lower scapular line, 3 on each side;

o N. 2 sensors positioned along the vertebral column, starting from the midline of the scapular up towards the neck.

Further kinetic signals are collected by inertial platforms (113) positioned on the garment (101) as follows:

o N. 1 inertial platform (113), IMU, composed of a 3-axis accelerometer, a gyroscope and a magnetometer, positioned in correspondence with the housing and connection area of the electronic control unit (107). In at least one of the alternative embodiments the inertial platform is inserted directly inside the electronic control card (107);

o N. 2 inertial platforms (113) positioned on the arms, in correspondence with the pressure sensors (112);

? Biometric signals:

o ECG signal (echocardiogram): collected by 10 sensors/acquisition points (114) positioned in the sensorized areas (102) corresponding to the abdominal loops, on the clavicular midline, on the sides of the sternal third and along the lower border of the left pectoral, from sternum to armpit;

o Respiratory frequency signal: the respiratory frequency and amplitude are read via a resistive strain gauge inserted into the body: a band of conductive fabric (115), integrated at chest height, and more? in particular in the lower insertion band of the pectorals, just above the diaphragm, it undergoes deformations as a result of respiratory acts; these deformations, which occur both on the longitudinal axis (increase in length) and on the transversal one (reduction in width) involve a quadratic increase in the resistance of the band itself. In the preferred embodiment of the present invention the conductive band is made using a metallic technical fabric, obtained, purely by way of non-binding indication, n? limiting, with the use of a mix of synthetic fibers (polyamide, polyester, elastomer, etc.) coated with silver and insulated with a TPU coating. In other alternative embodiments, yarns made of silver or other sufficiently ductile conductive metal can be used;

o Temperature: in the preferred embodiment of the present invention, the temperature sensor (116) is located in the central part of the chest strap, close to the of the xiphoid process; in an alternative form, said sensor is located on one of the sleeves of the body, near? of the inner portion of the wrist; in at least one other alternative embodiment, the sensor is located on a side, at side level;

o GSR (Galvanic Skin Response) : the galvanic response of the skin is used to measure the user's stress levels, since? the conductivity? of the skin varies under the action of the sympathetic nervous system. The GSR sensor (117), in the preferred embodiment of the invention, is positioned on the lower edge of the right pectoral, close to the of the axillary cable.

With reference to the preceding figures, an item of clothing (101) according to the present invention ? equipped with a small electronic/control unit (103), comprising:

? A signal pre-amplification system;

? A signal conditioning system;

? An A/D converter;

? a microprocessor;

? A memory module;

? A wireless communication module;

? A power battery;

? A micro USB port for device charging and data transmission.

Said system? assigned to carry out the following tasks:

? Management and power supply of the sensors integrated in the suit;

? Preamplification of the relative signals;

? Conditioning of said signals;

? Converting said signals from analog to digital;

? Pre-analysis of these signals with comparison to the threshold thanks to datasets and self-filling tables. ? Sending of signals above the threshold to the software on the mobile device (105) for the classification and identification of the event and possible sending of an alarm (106).

Both the kinetic and biometric sensors described above are totally integrated in the fabric of the garment and themselves made of conductive fabric, requiring external power supply, which they receive from the aforementioned electronic control unit (103). In the preferred embodiment of the invention presented, the sensors used are of the piezoresistive type, as regards the detection of kinetic data, and capacitive for biometric signals, have an average consumption of about 2.5 mW each and are powered at 5V by a lithium-ion rechargeable battery integrated in the electronic control unit (103). In its preferred embodiment, in a system according to the present invention, the overall consumption under normal operating conditions ? of 1 Wh, with peaks of 2Wh when searching for paired devices and sending data; in order to ensure adequate operation? and protection to the user, i.e. at least 4 hours of continuous operation, the battery has a capacity? of 1,200mAh. In at least one alternative embodiment, the battery can be of reduced dimensions, to the benefit of the miniaturization of the system and the comfort of use, providing, however, a more? limited in time.

With reference to the accompanying drawings and particularly to Fig.6 ? shown is a preferred embodiment for a kinetic pressure sensor (112) used in the sensor areas (102) of the garment (101). According to the proposed invention, said kinetic pressure sensors (112) are made of conductive fabric, to guarantee their flexibility. and maximum adherence to the body during all uses; said fabric? made of synthetic fiber coated with silver or other conductive and ductile material and isolated from the external environment by means of a micrometric coating of thermoplastic polyurethane (TPU). On the fabric? the circuit for capturing the pressure signal has been created: in the preferred embodiment of the present invention, the capturing circuit is created by sewing, according to the geometry shown in Fig.6, an insulated conductive thread (118), with respect to the fabric and the external environment, through a micrometric coating of non-conductive polymer. By applying an electric current (I) to said conductive wire (118), it will return, in the inoperative phase, and preload, a given resistance; this value is recorded as a calibration threshold each time a garment according to the invention is used. When the sensor (112) is subjected to a pressure which deforms it, in particular along the z axis, perpendicular to the construction plane of the circuit with the conductive wire (118), its resistance varies with proportionality quadratic with respect to the deviation from the initial condition; the initial values and those of subsequent measurements, compared with each other and with a table of reference values, with appropriate integrations over time, determine the nature of the event that generated the pressure on one or more? of the kinetic pressure sensors (112) energized.

In at least one alternative embodiment of the present invention, the pressure sensor (112), is not? integrated in the fabric as a conductive thread (118), but is applied to it: in this variant, the sensor is formed by a flexible and highly deformable substrate such as, purely by way of a non-limiting example, nor? binding, a polymeric rubber or a silicone rubber, inside which there is the picking circuit, made with different ductile and flexible conductive material. In at least another alternative embodiment, the sensors, both pressure (112) and biometric (114), are made using silver nanowires in a micrometric polymeric matrix.

With reference to the attached drawings and to Figures 7 and 8 ? the wiring system integrated in the garment is illustrated in detail: in the preferred embodiment of the present invention, the wiring and connections of the kinetic (112), (113) and biometric (114), (115), (116) sensors , (117) with the control unit (103) occur through the use of conductive threads (119) sewn inside the fabric. These wires convey all the signals to a collection center (111) where the signals are centralized and are further transmitted with further conductive wires (119) to the electronic control unit (103). Said conductive wires (119) are made, purely by way of non-binding example, n? limiting, in synthetic, elastomeric fiber, added with metal powders, ductile metals or other similar material, and are covered with an insulating coating made of polymeric material, TPU or similar. The electrical signals taken on the surface of the body are subject to multiple interferences and noises dictated not only by the state of motion of the body itself, but also by the surrounding environment with sources of electromagnetic radiation, wireless data transmissions, etc. The wiring therefore needs? to be isolated to ensure a? High quality? and signal cleaning; at the same time, however, the insulation must not be impactful and invasive so as not to affect fit? and comfort of the garment. Purely by way of example, a sheathed cable, a pi? tunics, normally used in electronics in similar applications, would be inconvenient in the invention. To overcome this limitation, in its preferred embodiment, the wiring insulation system exploits the external insulating layer of the conductive wires (119) used to make the connections and the conductive fabric itself: the manufacturing structure of the garment provides for the use of materials typical of the textile industry (tulle, cotton+elastomer, etc.) with inserts in conductive fabric. In the data acquisition areas, is the conductive fabric ? in direct contact with the user?s skin, while in the rest of the garment, which includes the wiring areas, it? placed on the outside, an insert sewn on top of the normally used material. What? by doing the conductive fabric it takes on the function of electromagnetic shielding for most of the emissions found in common environments frequented daily and associated with the insulating layer that surrounds the cable guarantees adequate shielding of the wiring.

In the preferred embodiment, the connection between the wiring wires (119) and the kinetic and/or biometric sensors takes place, in the case of synthetic fiber wires, by means of micro-crimping of the wiring itself on the sensor terminal, typically a wire in conductive material (119); in the case of insulated metal wires, the connection takes place by micro-welding of the sensor terminals on the cable. In at least one alternative embodiment both the crimping and the soldering can be replaced by a two-layer bonding in which the first layer, more? internal, ? made with conductive material, purely by way of non-binding example, n? limiting, a conductive silicone adhesive, and the second, external,? made of an insulating material, for example, a silicone adhesive. The choice of silicone adhesives ? dictated by necessity? to have excellent adhesion between the components, insulation from the external environment and adequate flexibility? and elasticity? of the junction which must follow the movements of the body without affecting the electrical contact.

With reference to the accompanying drawings and particularly to Figs. 9, 10, 11, 12, 13? represented the preferred embodiment of the invention, in which the control electronics (103) ? inserted in a housing (120) comprising a lower portion (121) and an upper portion (122) made of soft and flexible material, purely by way of non-binding indication, nor? limiting element, silicone, silicone rubber or similar material and equipped with two poussoir type electrical pressure contacts. Said electric pressure contacts will therefore comprise a part of the ?female? type? (123) positioned on the lower portion of the silicone housing (121), called backcover. while the corresponding contacts ?male? (124) are positioned on the item of clothing (101). Again with reference to the same aforementioned figures, the housing (120) containing the control electronics (103) provided with a and the rechargeable battery via connection (125) which allows connection to power banks or other commonly used electronic charging devices normally supplied with any mobile electronic device.

To remove the electronics (107) ? it is sufficient to grasp the housing (120) with the hands and exert a slight force in a perpendicular and coaxial direction with respect to the median connection axes. In this way the pressure contacts (123) will be released from the corresponding male contacts (124) positioned on the garment (101) and the electronics (107) and relative housing (120) removed, and be easily connected to a charging device . The removal of the control board (107) in a system according to the present invention? also necessary to wash the item of clothing (101): by separating the sensitive and usable electronic components from the mechanical and chemical action of washing, the item of clothing (101), totally inert and passive, ? washable without any problem with the typical methods used daily. To reconnect the electronic card (103) to the garment (101) ? it is sufficient to grip the casing (120) of the electronics, bring the female pressure contacts (123) close to the male ones (124) present on the item of clothing (107) and exert a force in a coaxial direction with respect to the contacts, until you hear the noise that certifies that hooking has taken place.

With reference to the attached drawings and particularly to Figs. 15 and 16 ? presented a preferable form of positioning and hooking said housing (120) to the item of clothing (101). According to this embodiment, the male contacts (124) are located on their side, supero posterior to the iliac crest. The positioning of the housing (120) of the electronics (103) in this area gives greater comfort and usability. daily found, increasing its potential diffusion and convenience? employment among the population.

With reference to the enclosed tables of drawings and particularly to Figs. 16 and 17 ? shown is one of the alternative forms of positioning and coupling of said housing (120) of the control electronics (107). According to this embodiment, the housing (120) is positioned at the height of the xiphoid process, i.e. at the end? lower sternum, near of the abdominal midline. The male contacts (124) are therefore anchored in this position. In the end ? it is appropriate to mention a further alternative embodiment of the present invention where the electronic card (107) is not? removable as in the cases shown and, consequently, the pressure contacts (123) present on the backcover (121) are replaced by connectors typical of electronic production; in this embodiment the terminals of the wiring incorporated in the fabric are replaced by flex pcb and flat connectors which mate with those present on the electronic card; in at least another alternative embodiment, the flex PBCs are replaced by flex buses, with related connectors.

The preamplifier subsystem according to the present invention ? capable of acquiring and managing both kinetic/pressure signals and biometric signals, with voltage values on the sensor itself typically between 0.5 and 4.0 mV. In the preferred embodiment ? a preamplifier is provided which according to the present invention has the following characteristics:

? CMRR > of 80dB;

? Other input impedance (>10M?);

? Earnings between 100 and 10,000;

? Differential inputs;

? Low bandwidth;

? Isolation.

With reference to Fig. 18 ? shown is a simplified diagram, by blocks (126), of the electronic card (103). According to said scheme, the signals relating to the kinetic (112), (113) and biometric (114), (115), (116), (117) sensors are acquired under ?everyday? conditions? of activity? of the user which, consequently, may include state of wakefulness and activity, state of sleep, etc. These conditions determine the possible overlap of more? sources of interference on the acquisition of the signal itself. By way of non-binding example, n? limiting, the muscle movement pi? or less accentuated, it creates disturbances in the reading both of the pressure kinetic signals (112), deforming the sensors on two of the three axes, and in the acquisition of the biometric signals, and in particular on the ECG signal (114); to overcome this limit and make the readings effective, the electronic board (103) integrates physical and digital filters which mitigate these effects, allowing to acquire a clean, stable and usable signal. In particular, an electronic card according to the present invention has the following filters:

? High pass filter with cutoff frequency at 0.1 Hz;

? Low-pass anti-aliasing filter with cutoff frequency of 100Hz;

? Muscle filters at 40 and 50 Hz.

The A/D converter used in the invention ? a standard subsystem, commonly included in electronic devices; purely by way of non-binding example, n? limiting the converter ? an integrated to n. 12 input channels, 8 single and 4 differential, with 12-bit resolution, and 2 output channels, with 11-bit resolution, with a maximum sampling frequency of 10,000 samples/sec.

With reference to Fig.19 of the attached drawings ? represented the system for acquiring and forwarding the signals relating to the kinetic (112), (113) and biometric (114), (115), (116), (117) sensors installed on the garment (101). The system according to the invention described up to now in its preferred form provides that the user, after putting on the item of clothing (101), must also activate the associated software which will be run in the background on the mobile device (105) on which ? been installed. Said software will go? to acquire (if necessary) the sensor data according to the following methodology:

? Every time the item of clothing (101) is worn and activated, the kinetic and biometric sensors start detecting the signals directly from the body and the electronic control unit (103) processes them;

? If the item of clothing (101) ? been just worn, the signals are read and recorded in the memory of the electronic control unit (103) as system calibration values and operating threshold;

? If the operating threshold ? already been set, the input signals from the sensors are compared with the tabled values:

o With input value ? to the threshold values, the data are discarded, why? considered useless for the purposes of the analysis as attributable to a standard condition of safety;

o With input value ? at threshold values, the data is forwarded to the software running on the paired mobile device (105) for subsequent analysis.

This strategy has two purposes:

o Energy saving: the wireless transmission module ? the consumption bottleneck of the electronic control unit (103) of the garment (101); its consumptions are those pi? high among all the types of components on board, therefore the? Optimization of its work allows to size the battery in an appropriate way to reduce weight and dimensions to a minimum for the benefit of wearability? and comfort of use. For this reason, reduce transmissions to a minimum, avoiding the forwarding of superfluous data and evidently not attributable to situations of violence or danger, ? a key optimization step. The identification of a calibration threshold at rest, recorded immediately after wearing the garment (101) and therefore in a situation of calm? and in a security context, and the subsequent comparison of the new values acquired with said threshold go in this direction: all values less than or equal to it can be ascribed to a situation of absolute security, therefore their forwarding to the software for analysis thorough, it is totally superfluous. On the contrary, the values greater than the threshold are those that can potentially lead to dangerous situations and therefore ? necessary to contextualize and analyze them in a more? thorough.

o Reduction of the computational load on the mobile device (105): the application or software associated with the garment (101) is installed on commercial mobile devices (105) whose primary purpose is? to make and receive wireless calls with LTE or similar protocol. On these devices are installed a multitude of other applications with load and priority? different computations. In order not to weigh heavily on the performance of the mobile device (105) and therefore on the performance of the security system, ? It is necessary that the processor is engaged as little as possible in data analysis and only when the data can be attributable to situations of danger or violence. Also in this case the purpose? fully achieved with the pre-analysis and forwarding strategy implemented in the control electronics.

With reference to the attached drawings and particularly to Fig.20 ? illustrated the flow chart (128) relating to the operation of the program that can be installed on the mobile device (105) and in particular:

? When the program is started for the first time, in the presale configuration phase, the thresholds used by the analysis algorithm for classification are set; these values are:

o Kinetic thresholds:

? acceleration/deceleration, in terms of force detected with respect to the mass of the object;

? inclination of the user?s body with respect to an ideal axis (standing position);

? Position of the user in relation to a magnetic field.

These thresholds are configured as a deviation value with respect to the optimal static condition and have three increasing levels (Sc1, Sc2 and Sc3);

o Biometric thresholds:

? Heart rate range;

? Respiration rate range;

? body temperature;

? EDA/GSR;

In the case of biometric signals, the thresholds indicate the limit value beyond which the signal is no longer? physiological and referable to a condition of safety and tranquillity. They have only one level (Sb);

o Time thresholds: indicate the duration of the recorded events and signals (range threshold) and the number of their iterations (single value threshold) in the time frame, t, taken into consideration for signal integration. In other words, these thresholds (St) indicate the duration of the events and how many times they are repeated during a given observation period.

? Upon first activation, the program starts the search for compatible clothing items. Search and connection are made via Bluetooth Low Energy, BLE wireless communication protocol. The connection (104) between the mobile device (105) and the control unit (103) of the item of clothing (101) within its reception range occurs through automatic exchange of asymmetric pairing keys. In the preferred embodiment, the program performs the pairing at each start by searching, in the first instance, for the first item of clothing (101) which ? been matched. In case of failed connection (104), the program activates a notification to the user with a request for support. If within the reception range of the communication module there is another compatible device (101), never paired with the program, a connection request to the new item of clothing (101) is activated.

In at least one alternative embodiment of the present invention, at each start of the program the pairing takes place freely with the first compatible item of clothing within detection range, without permanently memorizing its pairing keys.

? Once a connection has been established with a compatible item of clothing (101), the program works in the background on the mobile device (105) where ? installed, listening for any incoming data. When a data packet is acquired, it is opened and the contained signals classified according to their nature, ie: kinetic or biometric corresponding to the sensors described above.

? Biometric data comparison:

o If the data are lower than the threshold values, with the program open on the screen, they are displayed and subsequently stored as a log file to be used in a training dataset aimed at improving the recognition and classification efficiency of the system. In case the active program works in the background, the visualization does not take place, the data are saved in the memory of the app running on the mobile device (105) and used for the creation and expansion of the training dataset;

o If the data is above the threshold, they pass to the next analysis task which concerns the duration and the number of iterations over time;

? Kinetic data comparison:

o Comparison with Sc1: if they are minors, with an active program on the screen, they are displayed and subsequently memorized and used as a training dataset; with the program active in the background, they are directly saved and included in the training database; if greater than Sc1, they are compared with Sc2;

o Comparison with Sc2: if minor, they pass to the next task for the temporal analysis; if greater, they are compared with Sc3;

o Comparison with Sc3: if the data are less, they pass to the next time analysis task; if the data is greater, it goes directly to the classification task performed by the self-learning subroutine; ? Data duration comparison: in this task the time markers of the recorded data are compared with those of the presets set by the programmer;

o Biometric data: in case of detection above the threshold, a time stamp, a marker, is added to the raw data, which indicates how long said signal is for? remained above the threshold. The marker is compared with tabular values which serve to cluster the data for subsequent classification. By way of non-binding example, n? limiting, an increase above the threshold of the value of the heartbeat and breathing of a few seconds pu? be clustered in sporadic events (a reading error, a fright, etc.); the same increase protracted over time, in the order of minutes, is clustered as a possible activity? physical or agitated/fearful;

o Kinetic data: Kinetic data is accompanied by two time markers:

? Duration of the event, expressed in milliseconds;

? Number of events, expressed in n/min, events per minute.

Said markers are compared with a cross-analysis between the two markers and the tabulated and consequently clustered values; purely by way of non-binding example, n? limiting, an acceleration above the threshold of a few milliseconds which is not repeated, associated with a deviation of the gyroscope axis with similar temporal characteristics, is clustered as a single accidental event (accidental impact, random thrust, etc.); events of very low duration and high amplitude, repeated over time, can be clustered as possible traumatic events, activities physics, etc. The output of this analysis serves to provide data already? conform to the machine learning model for the final classification of events. ? Event Classification: Classification is done using a machine learning algorithm.

With reference to the accompanying drawings and particularly to Fig. 21 ? shown is a diagram (129) of the data analysis logics used. Said logics based on (mainly on an approximation method of (Bayesian) inferences subjective statistics, and more particularly the method of variational inferences, since they guarantee better scalability with large models, especially for large scale neural networks in applications of deep learning. In addition to the speed of calculation, the variational inference method of the invention is able to efficiently represent the weights (probability distributions) with a limited number of parameters. The presented method is similar to the category of FNNs (Feedforward Neural Networks) and approximates the posterior distributions of the weights of the network as Gaussian distributions associated with variational parameters.In the preferred embodiment of the present invention, the training of the model takes place by exploiting CPU/GPU systems before the parameters for the distributions of weights, for which, in the invention, two types of parallel random number generators have been used:

o A generator with binomial distribution and linear response logic (RLF-GRNG);

o A Wallace Gaussian Random Number Generator, oriented to Bayesian Neural Networks, with recursive method.

In the invention, the outputs of the comparisons enter the inferential engine where they are classified by exploiting the ?knowledge? a priori provided by the dataset, constantly updated with the log files generated by the program itself. The output of the classification? a probability value? which indicates whether, given the signals relating to a given event, it can be indexed as a violent/dangerous event, a malaise, a normal situation, etc.

The first indexing that the inferential engine performs on the input data concerns their inclusion in a category to which they belong:

o Physical discomfort of the user;

or Violence.

This step takes place by comparing the clustering carried out in the previous steps with the dataset inside the program: purely by way of example, it is not binding, nor? limiting, an event in which there are above-threshold alterations of the biometric signals, for a certain period of time without any type of threshold exceeding of the kinetic data, is indexed as a potential illness and classified by comparing it with samples belonging to this category.

Following indexing, the actual classification takes place, the outputs of which arise in the system reactions to the event:

o Malaise: an event indexed as such has two possible classification outputs such as confirmation of malaise or ?false positive?. In the event of a ?false positive?, with the program active on the screen, the data are displayed and subsequently saved as an event log in the system memory; while the program is active in the background, nothing is displayed and the log is saved immediately. If the classification of the event is confirmed, with the program active on the screen, the relative notifications are displayed and the system prepares for the subsequent steps;

o Violence: an event indexed in this category is classified according to three severity thresholds:

? The first threshold concerns random events, limited in time by duration, number of repetitions and amplitude, purely by way of non-binding example, nor? limiting, accidental falls or pushes, ?friendly? (a pat on the back, cheerfulness among friends?), etc. In this case, the system output ? the display of data on the screen, if the program ? active, with consequent storage of event logs. In the case of an application that works in the background, the data are not displayed but directly saved in the system memory for subsequent use;

? The second, intermediate threshold defines events of intensity moderate / high that can be attributable to situations of danger / violence, so? as in situations of activity? of the user (contact sport, high impact and intensity physical activity, dance, etc.). In this case the program, via haptic and sound signaling, requests the user?s intervention to clarify the type of event. In the event of a positive response from the user, i.e. confirmation of being in a situation of safety and tranquillity, the data, displayed on the screen, are recorded in an event log and memorized for subsequent use; in the event of a negative response from the user, i.e. confirmation of the violence, or no response from the user, the system prepares for the subsequent steps.

? The third threshold, pi? high, classifies the event as directly dangerous for the user and directly starts the alarm activation procedures.

? Start recording material relevant to the event: in the event of a full-blown illness, failure to respond to system requests or exceeding the third severity threshold, the system, in its preferred embodiment, starts recording audio/visual material, using the? hardware of the device on which ? installed, in order to document the event in progress. Purely by way of non-limiting example, n? binding, the program accesses and activates the device's microphones to record what is happening; in the same way it accesses and activates the system camera/s to record a video in real time. According to the invention , the recording of the audio/visual material lasts, in its preferred embodiment, for 4 minutes; at the end of the registration the files are uploaded to a dedicated repository, on a cloud server, accessible only by the user. In at least one alternative embodiment of the present invention, the audio/video material can be registered for a longer period of time. In at least one other alternative embodiment, the audio/visual material may not be registered.

The purpose of the registration ? that of providing evidence of what is happening, useful in the moments following the event to acquire useful information to intervene and, subsequently, to document what happened in the appropriate locations.

? Activation of the alarm: in a system according to the present invention, on first start-up and for all subsequent ones until the action is completed, the mobile program requests that contacts be indicated to which any alarm signal can be forwarded; purely indicative, not limiting, n? binding, these contacts can be telephone numbers, email addresses, social accounts, etc. The program then asks for the text of the message to be forwarded to the recipients. In its preferred embodiment, when dangerous conditions arise, the program sends the message indicated by the user, accompanied by his personal details. and from the geolocation of the device at the time of sending, to the recipients, consistently with the type of contact set, using one or more of the following protocols:

o BLE communication;

o LTE communication;

o Wi-fi communication;

o NFC communication.

With reference to the accompanying drawings and to Fig. 22, ? represented a simplified scheme (130) of creating a mesh network between compatible devices; since said network can be used for the forwarding of alarm messages in particular conditions and in particular in the case in which the alarm signal cannot be forwarded due to system malfunctions or breakdowns to one or more? communication modules, the program implements alternative sending strategies. In this situation, the program installed on the mobile device (105) will search, using the proximity communication modules? (NFC) and reduced range (BTLE, Wi-fi), of devices compatible with the mobile device (105) within the coverage range of the device on which it is? installed and which generated the alarm. By compatible device we mean a mobile device, purely indicative, a smartphone or a tablet, on which another iteration of the same program is installed. Once you have identified one or more compatible devices, a program according to the present invention establishes a connection and pairing with them in order to start a secure and encrypted communication; is obtained cos? a mesh network in which each device, or network node, operates both as a data forwarding gateway to the outside, and as a bridge to reach other compatible devices and share data with them. When pairing is completed, the program will forward? in encrypted form, anonymous and not viewable by the owners of said devices, all the data relating to the emergency contacts and the alarm message (106) of the mobile device (105) generating. The iterations of the program present in the paired devices will start attempts to forward the alarm to the newly imported contacts until one of them is successful. When forwarding is complete, it will come generated a command that will block? attempts on all other devices.

In the preferred embodiment of the present invention, together with the forwarding of all the information useful for sending the alarm, a haptic and acoustic notification is generated on the coupled devices, to allow their owners to intervene suddenly in support of the alarm. user.

Applicability? industry and implementation alternatives

The invention is applied to the sector of personal monitoring and safety and to vulnerable groups and workers who carry out tasks with a high risk of aggression, and in particular to the sector of devices designed to prevent domestic and public violence: in the field of prevention and intervention, a system according to the present invention ? able to continuously, silently and completely anonymously monitor the user's situation, to autonomously recognize a situation of danger, violence and/or discomfort and to intervene automatically, bypassing the voluntary limits? of activation typical of the prior art.

At the end of this description ? It is useful to reiterate how the invention can be easily implemented at the industrial production level and it should be noted that in the previous description the elements have been neglected for obvious reasons of synthesis and brevity. all those components and trivial operations and in the availability? of the person skilled in the art which concern the practical realization and the typical functional finalization of a device and method such as those claimed herein.

Particularly and, regardless of the example of embodiment given above, ? protection is required for any implementation alternatives that can be reasonably deduced and applied to the proposed solution in the following additional areas: e-health sector, and particularly in the field of monitoring and prevention of pathologies affecting various organs and systems, with further focus on cardiovascular pathologies.

Furthermore, instead of the subjectivist statistical method mentioned, it will be? can you use more methods? traditional methods of data analysis, which do not involve the use of probability distributions. Hardware-wise? is it possible to replace the textile sensors with pi? traditional piezoelectric foil sensors fixed in special ?fabric pockets? made on the garment. Finally, how already? mentioned several times, but it is necessary to repeat it once again, the invention just described lends itself to numerous and further variations which can offer further advantages with respect to those previously mentioned. These further variations could reasonably be made by a man skilled in the art without thereby departing from the invention as it results from the present description and from the claims attached hereto. Furthermore, the invention itself could be realized in a partial way cos? like all the various details described can be replaced by technically equivalent elements or solutions.

Claims (10)

1. System (100) for detection, recognition and intervention to combat acts of physical violence or physical damage to the detriment of people; said system (100) being capable of automatically monitoring and recognizing physical damage suffered by a person provided with a mobile device (105); said system (100) being capable of sending, autonomously and without the intervention of the monitored person, an alarm signal (106) by means of said mobile device (105); said system including: ? an item of clothing (101) provided with signal acquisition areas (102); said areas being equipped with kinetic sensors and biometric sensors; ? a circuit of conductive threads (119) integrated in said article of clothing (101); said circuit being used to connect said kinetic and biometric sensors with a collection node (111); ? a control unit (103) connected by means of said conductive wires (109) to said collection node (111); said control unit being further connected via a wireless communication channel (104) to said mobile device (105); ? means for using said central control unit (103) to acquire, pre-process and select the signals provided by said kinetic and biometric sensors and to transmit the corresponding data to said mobile device (105); ? means for using said mobile device (105) to analyze and classify with artificial intelligence techniques the data provided by said control unit (103); said means being employed to identify data correlated with possible acts of violence or physical danger; ? means for sending said alarm signal (106) via said mobile device (105). The system (100) according to claim 1 wherein said kinetic and biometric sensors comprise: ? kinetic pressure sensors (112); ? kinetic sensors in the form of inertial platforms (113); ? biometric sensors to detect ECG signal - electrocardiogram (114); ? biometric sensors to detect amplitude and frequency of respiration (115); ? biometric temperature sensors (116); ? GSR biometric sensors - Galvanic Skin Response (117). 3. System (100) according to claim 1 wherein said kinetic pressure sensors (112) comprise a circuit for capturing the pressure signal, said circuit being obtained by sewing a conductive thread (118) onto the garment (101); said conductive wire (118) being insulated by means of a micrometric coating of non-conductive polymer. 4. System (100) according to claim 1 wherein said wireless communication channel (104) uses at least one of the following transmission protocols: ? BLUETOOTH LOW ENERGY (BTLE 4.0); ? WIFI (IEEE 802.11xx); ? NFC (13.56MHz). The system (100) according to claim 1 wherein said central control unit (103) is equipped with a battery and a connector (125) for recharging; said control unit (103), battery and connector (125) being housed within a case (120) equipped with electric pressure contacts (123), (124) used to fasten and connect said control unit (103) with said item of clothing (101 ). 6. System (100) according to claim 1 wherein said means for forwarding said alarm signal (106) via said mobile device (105) are used to send to predetermined contacts information comprising at least the identification data of said person and the data of geolocation of said mobile device (105); said information being sent by at least one of the following methods: ? a pre-set text message (108); ? a voice call with a pre-recorded message (109); ? an e-mail with a pre-set text message (110). The system (100) according to claims 1 and 6 wherein said messages can be transmitted through the following networks: ? GSM/LTE or similar networks; ? Wi-Fi networks; ? Public or private Internet networks; ? Mesh networks (130) obtained by means of proximity communication modules? NFC and low-field communication BTLE and Wi-fi installed on said mobile device (105); said modules being used to transmit said messages (108), (109), (110) in cooperation with further adjacent mobile devices compatible with the system (100). 8. System (100) according to claim 1 wherein said means for forwarding said alarm signal (106) comprises audio video recordings started autonomously on the mobile device (105) and forwarding to a cloud server (107). The system (100) according to claims 1 and 2 wherein said means for using said central control unit (103) to acquire, pre-process and select the signals provided by said kinetic and biometric sensors and to transmit the data corresponding to said mobile device (105) include: ? power supply, pre-amplification, conditioning and digital conversion of the signals provided by said kinetic sensors (112), (113) and of the signals provided by said biometric sensors (114), (115), (116), (117); ? selection and transmission of data corresponding to signals exceeding originally pre-defined and progressively updated kinetic and biometric thresholds; these thresholds being memorized in self-updating tables. The system (100) according to claim 1, wherein said means for using said mobile device (105) to analyze and classify with artificial intelligence techniques the data provided by said control unit (103) comprises a machine learning system and classification (129) based on techniques of variational inference; said automatic learning and classification system (129) comprising an inferential statistical engine capable of processing a dataset of the data progressively acquired by the kinetic and biometric sensors and a dataset used as a reference model; said inferential statistical engine being used to classify the data obtained from said garment and inferential statistic (101) according to different levels of attention and said inferential statistical engine being used to update said self-updating tables on the unit? control (103).