Method and system for fault detection from audio streams in a public transport vehicle

Abstract

The invention relates to a method (100) for detecting an incident in a public transport vehicle, in particular in a railway vehicle, comprising at least one iteration of a phase (102), referred to as the detection phase, comprising at least the following steps: - capture (104) in said public transport vehicle, and in particular in a passenger area of ​​said public transport vehicle, of at least one audio stream, referred to as the first audio stream, by at least one device, referred to as the portable device, said device comprising at least one audio capture means and being intended to be worn by a person moving within said public transport vehicle, - transmission (106) of said at least one captured first audio stream to at least one device, referred to as the analysis device, separate from said at least one portable device and comprising at least one intelligent computing unit, and - analysis (108), by said at least one intelligent computing unit,said at least one initial audio stream for detecting an incident. It also relates to an incident detection system for implementing a method according to the invention, as well as a public transport vehicle equipped with such an incident detection system.

Description

[0001] The present invention relates to a method for detecting incidents in a public transport vehicle, particularly a railway vehicle, using audio streams. It also relates to a system for detecting incidents in a public transport vehicle, particularly a railway vehicle. Furthermore, it relates to a public transport vehicle equipped with such a system.

[0002] The field of the invention is the field of incident detection in public transport vehicles, in particular a railway vehicle. État de la technique

[0003] Numerous incident detection systems are available for public transportation vehicles. For many years, video surveillance systems have been the preferred choice. However, these systems are not without limitations. An incident can only be detected by video surveillance if it is visible. Therefore, if an incident occurs in a blind spot of the video surveillance system, or if the system's field of vision is obstructed, the incident cannot be detected.

[0004] To compensate for the shortcomings of existing video surveillance systems, audio surveillance is also used as a supplement. This audio surveillance involves analyzing audio content captured by microphones integrated into existing devices such as intercom alarms or CCTV cameras. These solutions thus only improve incident detection in specific areas that are generally already monitored by video surveillance systems.

[0005] EP 1 473 709 A1 discloses an incident detection device for use in a public transportation vehicle, employing fixed microphones distributed throughout the vehicle and transmitting captured sounds to a central computer. US 2014 / 0361886 A1 discloses a portable firearm detonation detection device that analyzes the captured sound.

[0006] One objective of the present invention is to remedy at least one of the aforementioned drawbacks.

[0007] Another objective of the present invention is to provide a solution enabling better incident detection in a public transport vehicle, and in particular in a passenger area of ​​said public transport vehicle. Exposé de l'invention

[0008] The invention makes it possible to achieve at least one of these goals by means of an incident detection method in a public transport vehicle, in particular in a railway vehicle, comprising at least one iteration of a phase, called the detection phase, comprising at least the following steps: capture in said public transport vehicle, and in particular in a passenger area of ​​said public transport vehicle, of at least one audio stream, referred to as first audio stream, by at least one device, referred to as portable, said device comprising at least one audio capture means and being intended to be worn by a person moving within said public transport vehicle, transmission of said at least first audio stream captured to at least one device, referred to as analysis, separate from said at least one portable device and comprising at least one intelligent computing unit, and analysis, by said at least one intelligent computing unit, of said at least first audio stream to detect an incident.

[0009] The invention thus proposes to capture an audio stream with a device worn by a person moving within the vehicle and which can be freely positioned within said vehicle.

[0010] Thus, the solution according to the invention is free from the positioning constraints of the audio capture means inherent in state-of-the-art audio surveillance solutions where the audio content is captured by one or more microphones equipping existing devices, such as an intercom alarm device or a video surveillance camera, whose positioning is chosen to best perform their main function.

[0011] Therefore, a portable device can be positioned closer to an incident than a fixed device. This proximity increases the likelihood that the captured audio stream will contain more information about the incident being detected, thus enabling its detection during analysis. Furthermore, a portable device can be positioned in locations where it is not possible to install a fixed device, such as in the middle of a crowd of people on public transportation.

[0012] Furthermore, the solution according to the invention provides better audio coverage than existing solutions, particularly when a plurality of first audio streams are captured by a plurality of portable devices.

[0013] By way of example, and by no means exhaustive, an incident can be: verbal and / or physical aggression, an accident during a mission of said public transport vehicle, damage to or destruction of material elements, such as furniture, of said public transport vehicle, intrusion into said public transport vehicle when it is not on a mission, an act of vandalism, such as the creation of graffiti, a terrorist attack and / or the use of a firearm, a material and / or mechanical failure of said public transport vehicle, etc.

[0014] As a non-limiting example, at least one portable device can be a user device such as a smartphone, smartwatch, tablet, computer, digital camera, etc...

[0015] Alternatively or in addition, at least one portable device may be a device, called public, local, accessible to any person moving within said place.

[0016] Advantageously, the steps of transmitting said at least one first audio stream and analyzing said at least one first audio stream can be carried out in real time.

[0017] Thus, the transmission of the first audio stream can begin before the recording is complete, and the analysis of the first audio stream can begin before the transmission is finished. Consequently, it is possible to detect an incident more quickly after it has started.

[0018] Advantageously, the detection phase can also include the following steps: capture in said public transport vehicle, and in particular in a passenger area of ​​said public transport vehicle, of at least one audio stream, referred to as second audio stream, by at least one device, referred to as fixed, said fixed device comprising at least one audio capture means and being intended to be fixed in said public transport vehicle, transmission of said at least one second audio stream captured to said at least one analysis device, and analysis, by said at least one intelligent computing unit, of said at least one second audio stream to detect an incident.

[0019] Thus, it is possible to detect an incident within a public transport vehicle using a second audio stream captured by a fixed device. Consequently, the method according to the invention allows for improved incident detection within a public transport vehicle. Indeed, in addition to analyzing a first audio stream captured from a free and potentially changing location during recording, it is possible to analyze a second audio stream captured from a fixed location within the vehicle.

[0020] At least one fixed device can be positioned in a specific area of ​​interest, such as the entrance of the vehicle, while the portable device can be positioned freely, particularly where a fixed device cannot be installed. Therefore, it is possible to better cover the interior of the vehicle for audio recording and incident detection.

[0021] By way of non-limiting example, at least one fixed device may be a dedicated device, installed in said public transport vehicle solely for incident detection.

[0022] Alternatively or in addition, by way of non-limiting example, at least one fixed device may be a microphone fitted to an existing device in said public transport vehicle, such as an intercom alarm device or a video surveillance camera.

[0023] Advantageously, the steps of transmitting said at least one second audio stream and analyzing said at least one second audio stream can be carried out in real time.

[0024] Thus, the transmission of the second audio stream can begin before the recording is complete, and the analysis of the second audio stream can begin before the transmission is finished. Consequently, it is possible to detect an incident more quickly after it has occurred.

[0025] Advantageously, at least one intelligent computing unit can include a neural network pre-trained for incident detection in audio streams, especially audio streams captured within public transport vehicles.

[0026] Thus, it is possible to detect an incident more accurately within a public transport vehicle.

[0027] The neural network can be a deep neural network, or a convolutional neural network, such as a CNN.

[0028] The neural network can be trained using a training dataset, each training dataset comprising a training audio stream corresponding to an incident and a data point indicating the nature of the incident, or at least whether or not an incident existed.

[0029] The training of the neural network can be carried out using a known training algorithm, such as, for example, the backpropagation of gradient method.

[0030] Advantageously, the process may further include at least one iteration of a phase, called the matching phase, prior to the first iteration of the detection phase, comprising at least the following steps: sending, by said at least one portable device, a connection request with said at least one analysis device, to a communication gateway, said request including a geolocation history of said portable device, verification of the presence of said portable device in said public transport vehicle based on said geolocation history and at least one geolocation data of said vehicle, and establishment, by said communication gateway, of a connection between said portable device and said analysis device, when the verification is satisfactory.

[0031] Thus, it is possible to verify the presence of a portable device in the vehicle before allowing that device to transmit an initial audio stream to an analysis device. Therefore, it is possible to verify, before the analysis, that the portable device is indeed located in the vehicle. This reduces the occurrence of false positives, for example, when an incident is detected by analyzing an audio stream captured and transmitted by a portable device located outside the public transport vehicle.

[0032] During the verification stage, the wearable device can, for example, be determined to be in the public transport vehicle when the last position in its geolocation history matches the current geolocation of said public transport vehicle.

[0033] Alternatively, a plurality of geographic positions included in the geolocation history of said portable device can be compared to a plurality of geolocation data of said vehicle, corresponding for example to the route taken by said vehicle.

[0034] As a non-limiting example, communications between the communication gateway and at least one portable device can be wireless communications, more specifically using LiFi, Wifi and / or Bluetooth protocols.

[0035] Advantageously, the pairing phase can include, prior to the step of sending a connection request, a step of powering on the communication gateway according to a state data relating to said public transport vehicle.

[0036] Thus, the implementation of the matching phase can be controlled so that a matching can only be performed under specific conditions, such as a specific configuration or a moment identified as important.

[0037] At least one state data point relating to the vehicle can be captured by a sensor located in or near said vehicle. The state data can then be transmitted to a system implementing the present invention in the form of at least one signal, either analog or digital.

[0038] By way of non-limiting examples, a state data item relating to the vehicle may be, or include: a time-based piece of data, such as the start or end time of the mission of said public transport vehicle, data relating to the locked or unlocked status of at least one, in particular each, access to said vehicle, data relating to a geographical position of said public transport vehicle, data relating to the speed of movement of said public transport vehicle, etc...

[0039] According to another aspect of the invention, an incident detection system is proposed for use in a public transport vehicle, particularly a railway vehicle, for implementing a method according to the invention, comprising: at least one device, referred to as portable, intended to be worn by a person moving within said public transport vehicle, said device comprising: at least one audio capture means for capturing an audio stream, referred to as first audio stream, and at least one means for transmitting said first audio stream to a separate device, at least one analysis device, separate from said at least one portable device and comprising at least one intelligent computing unit for analyzing at least one first audio stream to detect an incident.

[0040] Thus, the incident detection system according to the invention allows a portable device to be freely positioned within a public transport vehicle to capture an initial audio stream. The system according to the invention further allows the transmission of this initial captured audio stream to an analysis device comprising an intelligent computing unit to analyze said initial audio stream in order to detect an incident.

[0041] Therefore, the system according to the invention allows for more flexible and adaptable audio surveillance coverage, depending on the needs. Indeed, a portable device can be positioned freely, particularly where a fixed device cannot be installed. An audio surveillance device can thus be positioned as close as possible to an incident. This proximity increases the likelihood that the captured audio stream will allow for the detection of an incident during its analysis.

[0042] Advantageously, the system can also include: at least one device, called fixed, intended to be fixed in said public transport vehicle and comprising at least one audio capture means to capture at least one audio stream, called second audio stream, and at least one device, called analysis, comprising at least one intelligent computing unit to analyze at least said second audio stream to detect an incident.

[0043] Thus, it is possible to detect an incident within a public transport vehicle using a second audio stream captured by a fixed device. Consequently, the system according to the invention allows for improved incident detection within a transport vehicle. Indeed, in addition to analyzing a first audio stream captured from a free and potentially changing location during recording, it is possible to analyze a second audio stream captured from a fixed location within the vehicle.

[0044] Thus, at least one fixed device can be positioned in a specific area of ​​interest, such as the entrance of the vehicle, while the portable device can be positioned freely, particularly where a fixed device cannot be installed. Consequently, it is possible to better cover the interior of the vehicle for audio recording and incident detection.

[0045] By way of non-limiting example, at least one fixed device may be a dedicated device, installed in said public transport vehicle solely for incident detection.

[0046] By way of non-limiting example, at least one fixed device may be a microphone equipping an existing device in said public transport vehicle, such as an intercom alarm device or a video surveillance camera.

[0047] By way of non-limiting example, the system according to the invention may include a single analysis device configured to analyze at least one first audio stream captured by a portable device and at least one second audio stream captured by a fixed device.

[0048] Alternatively, the system according to the invention may include an analysis device dedicated to the analysis of the first audio stream and another analysis device dedicated to the analysis of the second audio stream.

[0049] As a non-limiting example, at least one analytical device and one fixed device may be included in the same device.

[0050] As a non-limiting example, at least one fixed device may be connected by a wired link to said at least analysis device.

[0051] Advantageously, said at least one portable device can be configured to transmit in real time to said at least one analysis device a first captured audio stream and in that said analysis device is configured to analyze in real time the first transmitted audio stream.

[0052] Thus, the transmission of the first audio stream can begin before the recording is complete, and the analysis of the first audio stream can begin before the transmission is finished. Consequently, it is possible to detect an incident more quickly after it has occurred.

[0053] Advantageously, when the system includes at least one fixed device, said at least one fixed device may be configured to transmit in real time to said at least one analysis device a second captured audio stream and in that said analysis device is configured to analyze in real time the second transmitted audio stream.

[0054] Thus, the transmission of the second audio stream can begin before the recording is complete, and the analysis of the second audio stream can begin before the transmission is finished. Consequently, it is possible to detect an incident more quickly after it has occurred.

[0055] Advantageously, the system may further include at least one communication gateway to enable a connection between at least one portable device and at least one analysis device.

[0056] By way of non-limiting example, communications between said communication gateway and said at least one portable device may be wireless communications, more specifically using LiFi, Wifi and / or Bluetooth protocols.

[0057] As a non-limiting example, the communication gateway can be connected by a wired link to the analysis device.

[0058] Advantageously, the presence of a portable device in said public transport vehicle can be verified during a pairing phase, based on a geolocation history of said portable device and at least one geolocation data of said vehicle, so that the gateway allows a connection between said portable device and said analysis device only once the verification is satisfactory.

[0059] As a non-limiting example, when the system includes at least one fixed device, at least one communication gateway can allow a connection between at least one fixed device and at least one analysis device.

[0060] According to another aspect of the invention, a public transport vehicle is proposed equipped in at least one passenger area with a system according to the invention.

[0061] Advantageously, the vehicle can be: a wagon, a locomotive, a train, a tram, etc... Description des figures et modes de réalisation

[0062] Other advantages and features will become apparent upon examination of the detailed description of a non-limiting embodiment, and the accompanying drawings on which: there FIGURE 1 is a schematic representation of a non-limiting example of a method according to the invention; the FIGURE 2 is a schematic representation of another non-limiting embodiment of a method according to the invention; the FIGURE 3 is a schematic representation of another non-limiting embodiment of a method according to the invention; the FIGURE 4 is a schematic representation of another non-limiting embodiment of a method according to the invention; the FIGURE 5 is a schematic representation of a non-limiting example embodiment of a system according to the invention; the FIGURE 6 is a schematic representation of another non-limiting embodiment of a system according to the invention; the FIGURE 7 is a schematic representation of another non-limiting embodiment of a system according to the invention, and the FIGURE 8 is a schematic representation of a non-limiting example embodiment of a vehicle according to the invention equipped with a system according to the invention.

[0063] It is understood that the embodiments described below are by no means exhaustive. In particular, variants of the invention may be conceived comprising only a selection of the features described below, isolated from the other features described, if this selection of features is sufficient to confer a technical advantage or to differentiate the invention from the prior art. This selection includes at least one preferably functional feature without structural details, or with only a portion of the structural details if this portion is sufficient solely to confer a technical advantage or to differentiate the invention from the prior art.

[0064] In the figures, elements common to several figures retain the same reference.

[0065] There FIGURE 1 is a schematic representation of a non-limiting example embodiment of a method according to the invention.

[0066] The 100 process illustrated in FIGURE 1 is a method of detecting incidents in a public transport vehicle, particularly in a railway vehicle.

[0067] By way of example, and by no means exhaustive, an incident can be: verbal and / or physical aggression, an accident during a mission of said public transport vehicle, damage to or destruction of material elements, such as furniture, included in said public transport vehicle, intrusion into said public transport vehicle when it is not on a mission, an act of vandalism, such as the creation of graffiti, a terrorist attack and / or the use of a firearm, a material and / or mechanical failure of said public transport vehicle, etc.

[0068] The person detection process 100 includes at least one iteration of a phase 102, called the detection phase.

[0069] Phase 102 of detection includes a step 104 of capturing in said public transport vehicle at least one audio stream, referred to as first audio stream, by at least one device, referred to as portable, said device comprising at least one means of audio capture and being intended to be worn by a person moving within said public transport vehicle.

[0070] During this data capture step 104, the portable device can be freely positioned within the public transport vehicle. Furthermore, the position of the portable device within the vehicle can be changed during the data capture process.

[0071] The detection phase 102 further includes a step 106 of transmission of said at least one first captured audio stream to at least one device, called an analysis device, separate from said at least one portable device and comprising at least one intelligent computing unit.

[0072] This transmission step 106 can, for example, be performed once the capture of a first audio stream is complete. Preferably, the transmission of said first audio stream can be performed in real time. Thus, step 106, which involves transmitting at least a second audio stream, begins before the completion of step 104, which involves capturing at least a first audio stream.

[0073] Furthermore, phase 102 of detection includes a step 108 of analysis, by said at least one intelligent computing unit, of said at least one first audio stream to detect an incident.

[0074] This analysis step 108 can, for example, be performed once the transmission of a first audio stream is complete. Preferably, the analysis of said first audio stream can be performed in real time. Thus, step 108, which analyzes at least one first audio stream, begins before the end of step 106, which transmits at least one first audio stream, preferably as soon as the analysis device begins receiving the first audio stream.

[0075] The intelligent computing unit may, for example, include a neural network previously trained for incident detection in audio streams, particularly audio streams captured within the public transport vehicle.

[0076] There FIGURE 2 is a schematic representation of another non-limiting example of an embodiment of a method according to the invention.

[0077] The 200 process illustrated in FIGURE 2 includes all the steps of process 100 described in relation to the FIGURE 1 .

[0078] Phase 102 of detection of process 200 further comprising a step 202 of capturing in said public transport vehicle at least one audio stream, referred to as second audio stream, said fixed device comprising at least one audio capture means and being intended to be fixed in said public transport vehicle.

[0079] Although illustrated after step 108 of analyzing a first audio stream in FIGURE 2 , step 202 of capturing at least a second audio stream can be carried out in parallel with step 104 of capturing at least a first audio stream.

[0080] Phase 102 of the detection of process 200 further includes a step 204 of transmitting said at least a second audio stream captured to said at least one analysis device.

[0081] This transmission step 204 can, for example, be performed once the capture of a second audio stream is complete. Preferably, the transmission of a second audio stream can be performed in real time. Thus, step 204, the transmission of at least one second audio stream, begins before the completion of step 202, the capture of at least one second audio stream.

[0082] Furthermore, phase 102 of detection includes a step 206 of analysis, by said at least one intelligent computing unit, of said at least one second audio stream to detect an incident.

[0083] This analysis step 206 can, for example, be performed once the transmission of a second audio stream is complete. Preferably, the analysis of said first audio stream can be performed in real time. Thus, step 206 of analyzing at least one second audio stream is started before the end of step 204 of transmitting at least one second audio stream, preferably as soon as the analysis device begins receiving a second audio stream.

[0084] There FIGURE 3 is a schematic representation of another non-limiting example of an embodiment of a method according to the invention.

[0085] The 300 process illustrated in FIGURE 3 includes a phase 102 detection as described previously in relation to the FIGURE 1 or with the FIGURE 2 .

[0086] The process 300 further includes, prior to the first iteration of the detection phase 102, a pairing phase 302 of at least one portable device with a communication gateway.

[0087] The 302 pairing phase includes a 304 step of sending, by at least one portable device, a connection request with at least one analysis device, to a communication gateway, said request including a geolocation history of said portable device.

[0088] Phase 302 further includes a step 306 of verification of the presence of said portable device in said public transport vehicle based on said geolocation history and at least one geolocation data of said vehicle.

[0089] During verification step 306, the wearable device can, for example, be determined to be in the public transport vehicle when the last position in its geolocation history matches the current geolocation of said public transport vehicle.

[0090] Alternatively, during verification step 306, a plurality of geographic positions included in the geolocation history of said portable device may be compared to a plurality of geolocation data of said vehicle, corresponding for example to the route taken by said vehicle.

[0091] Phase 302 further includes a step 308 of establishing, by said communication gateway, a connection between said portable device and said analysis device, when the verification is satisfactory.

[0092] This pairing phase 302 verifies whether a portable device is present in the vehicle before allowing it to transmit an initial audio stream to an analysis device. Therefore, a device located outside the vehicle cannot transmit an audio stream captured outside the vehicle.

[0093] There FIGURE 4 is a schematic representation of another non-limiting example of an embodiment of a method according to the invention.

[0094] The 400 process illustrated in FIGURE 4 includes all the steps of process 300 described in relation to the FIGURE 3 .

[0095] Phase 302 of the pairing of process 400 includes, prior to step 304 of sending a connection request, a step 402 of energizing said communication gateway according to a state data relating to said public transport vehicle.

[0096] Thus, the implementation of the matching phase can be controlled so that a matching can only be performed under specific conditions, such as a specific configuration or a moment identified as important.

[0097] By way of non-limiting examples, a state data item relating to the vehicle may be, or include: a time-based piece of data, such as the start or end time of the mission of said public transport vehicle, data relating to the locked or unlocked status of at least one, in particular each, access to said vehicle, data relating to a geographical position of said public transport vehicle, data relating to the speed of movement of said public transport vehicle, etc...

[0098] There FIGURE 5 is a schematic representation of a non-limiting example embodiment of a system according to the invention.

[0099] System 500 is an incident detection system in a public transport vehicle, particularly in a railway vehicle, for the implementation of a process 100 described above.

[0100] The example illustrated in FIGURE 5 The 500 system for incident detection in a public transport vehicle comprises a number n, greater than or equal to 1, of 5021-502n wearable devices, each capable of capturing a first audio stream and intended to be worn by a person 504. Hereafter, the reference 502i may refer to any of the 5021-502n wearable devices.

[0101] In the example shown, each 502i wearable device is a smartphone-type user device, carried by a person, designated respectively 5041-504n, also referred to as 504i hereafter. Alternatively, at least one 502i wearable device may be another wearable device, for example, a smartwatch, tablet, computer, digital camera, etc. Alternatively, or in addition, at least one 502i wearable device may be public, local, and accessible to anyone within the location. In this latter case, each 504i person may use a wearable device at a vending machine or designated distribution point.

[0102] The example illustrated in FIGURE 5 The 500 system for incident detection in a public transport vehicle includes an analysis device 506, separate from each of the portable devices 502 i and comprising at least one intelligent computing unit to analyze at least one initial audio stream to detect an incident.

[0103] Each 502 i portable device of the 500 system is in communication with the analysis device 506. Thus, the first audio stream captured by each 502 i portable device can be transmitted to the analysis device 506 for analysis in order to detect a possible incident.

[0104] The 506 analysis device can, for example, be configured to analyze the various initial audio streams captured sequentially. Preferably, the 506 analysis device can analyze a plurality of initial audio streams in parallel and simultaneously.

[0105] For example, communications between the 506 analysis device and each 502 i portable device can be wireless.

[0106] There FIGURE 6 is a schematic representation of another non-limiting example of an embodiment of a system according to the invention.

[0107] The 600 incident detection system illustrated in FIGURE 6 includes all elements of the 500 incident detection system illustrated in FIGURE 5 .

[0108] The 600 system is an incident detection system in a public transport vehicle, in particular in a railway vehicle, for the implementation of a method 100, 200, 300 or 400 described above and generally any method according to the invention.

[0109] The 600 incident detection system further comprises a number n, greater than or equal to 1, of fixed devices 6021-602n, each capable of capturing a second audio stream and intended to be fixed in said public transport vehicle. Hereafter, the reference 602i may designate any one of the fixed devices 6021-602n.

[0110] At least one fixed 602 i device may be a dedicated device, installed in said public transport vehicle solely for incident detection.

[0111] Alternatively or in addition, at least one fixed 602 i device may be a microphone fitted to an existing device in said public transport vehicle, such as an intercom alarm device or a CCTV camera.

[0112] In the example shown, each fixed device 602 i is schematically illustrated by a microphone and is connected by a wired link to the analysis device 506. Thus, each fixed device 602 i is in communication with the analysis device 506. Thus, the second audio stream captured by each fixed device 602 i can be transmitted to the analysis device 506 for analysis in order to detect a possible incident.

[0113] In the example shown, a single 506 analysis device is intended to analyze at least one first audio stream and at least one second audio stream.

[0114] Alternatively, a system according to the invention may include an analysis device for analyzing first audio streams and another analysis device for analyzing second audio streams.

[0115] There FIGURE 7 is a schematic representation of another non-limiting example of an embodiment of a system according to the invention.

[0116] The 700 incident detection system illustrated in FIGURE 7 includes all elements of the 600 incident detection system illustrated in FIGURE 6 .

[0117] The 700 incident detection system also includes a 702 communication gateway connected by a wired connection to the 506 analysis device.

[0118] In the example shown, each fixed 602i device is connected via a wired connection to the communication gateway 702. Thus, the communication gateway 702 interfaces each fixed 602i device with the analysis device 506.

[0119] Alternatively, at least one, in particular each, fixed 602 i device can be connected via a wired connection to the analysis device without going through the 702 communication platform.

[0120] In the example shown, the 702 communication gateway is in wireless communication, for example using LiFi, Wifi and / or Bluetooth protocols, with each 502 i portable device. Thus, the 702 communication platform interfaces each 502 i portable device with the 506 analysis device.

[0121] The 702 communication platform can, for example, establish a connection between a 502 i portable device and the 506 analysis device only once the presence of the 502 i portable device in the public transport vehicle has been verified.

[0122] This verification can, for example, be carried out during a matching phase as described above in relation to the FIGURES 3 et 4 .

[0123] There FIGURE 8 is a schematic representation of a non-limiting example embodiment of a vehicle according to the invention equipped with a system according to the invention.

[0124] The 800 public transport vehicle illustrated very schematically in FIGURE 8 is a railway vehicle, or a road vehicle.

[0125] The vehicle 800 includes an area, called the driver area 802, reserved for the driver and an area, called the passenger area 804, in which passengers are free to move around.

[0126] The 800 vehicle is also equipped with a 700 detection system as described previously in relation to the FIGURE 7 .

[0127] In the example shown, the communication platform 702 and the analysis device 506 are arranged in a ceiling 806 of said passenger area 804. In addition, the fixed devices 602 1 -602 n are located inside the passenger area 804 and fixed to the ceiling 806. Furthermore, the portable devices 502 1 -502 n are each carried by a person, referenced respectively 504 1 -504 n, moving freely in the passenger area 804.

[0128] Of course, the invention is not limited to the detailed examples above.

Claims

1. Method (100;200;300;400) for detecting an incident in a public transport vehicle (800), in particular in a rail vehicle, comprising at least one iteration of a phase (102), referred to as a detection phase, comprising at least the following steps: - capture (104) in said public transport vehicle (800), and in particular in a passenger zone (804) of said public transport vehicle (800), of at least one audio stream, referred to as a first audio stream, by at least one device (5021-502n), referred to as a portable device, said device (5021-502n) comprising at least one audio capture means and being intended to be carried by a person (5041-504n) moving within said public transport vehicle (800), - transmission (106) of said at least one captured first audio stream to at least one device (506), referred to as an analysis device, separate from said at least one portable device (5021-502n) and comprising at least one intelligent computing unit, and - analysis (108), by said at least one intelligent computing unit, of said at least one first audio stream to detect an incident.

2. Method (100;200;300;400) according to the preceding claim, characterized in that the steps of transmitting (106) said at least one first audio stream and analyzing (108) said at least one first audio stream are performed in real time.

3. Method (200) according to either of the preceding claims, characterized in that the detection phase (102) further comprises the following steps: - capture (202) in said public transport vehicle (800), and in particular in a passenger zone (804) of said public transport vehicle (800), of at least one audio stream, referred to as a second audio stream, by at least one device (6021-602n), referred to as a fixed device, said fixed device (6021-602n) comprising at least one audio capture means and being intended to be fixed in said public transport vehicle (800), - transmission (204) of said at least one captured second audio stream to said at least one analysis device (506), and - analysis (206), by said at least one intelligent computing unit, of said at least one second audio stream to detect an incident.

4. Method (200) according to the preceding claim, characterized in that the steps of transmitting (204) said at least one second audio stream and analyzing (206) said at least one second audio stream are performed in real time.

5. Method (100;200;300;400) according to any of the preceding claims, characterized in that at least one intelligent computing unit comprises a neural network previously trained for incident detection in audio streams, in particular audio streams captured within public transport vehicles.

6. Method (300;400) according to any of the preceding claims, characterized in that it further comprises at least one iteration of a phase (302), known as a pairing phase, prior to the first iteration of the detection phase (102), comprising at least the following steps: - sending (304), by said at least one portable device (5021-502n), a request for connection with said at least one analysis device (506), to a communication gateway (702), said request comprising a geolocation history of said portable device (5021-502n), - verifying (306) the presence of said portable device (5021-502n) in said public transport vehicle (800) on the basis of said geolocation history and at least one geolocation datum of said vehicle (800), and - establishing (308), by said communication gateway (702), a connection between said portable device (5021-502n) and said analysis device (506), when the verification is satisfactory.

7. Method (400) according to the preceding claim, characterized in that the pairing phase (302) comprises, prior to the step of sending (304) a connection request, a step (402) of powering up the communication gateway on the basis of a status datum relating to said public transport vehicle (800).

8. Incident detection system (500;600;700) in a public transport vehicle (800), in particular in a rail vehicle, for implementing a method (100;200;300;400) according to any of the preceding claims, comprising: - at least one device (5021-502n), referred to as a portable device, intended to be worn by a person (5041-504n) moving within said public transport vehicle (800), said device comprising: • at least one audio capture means for capturing an audio stream, referred to as a first audio stream, and • at least one means of transmitting said first audio stream to a separate device, - at least one device (506), referred to as an analysis device, separate from said at least one portable device (5021-502n) and comprising at least one intelligent computing unit for analyzing at least a first audio stream to detect an incident.

9. System (600;700) according to the preceding claim, characterized in that it further comprises: - at least one device (6021-602n), referred to as a fixed device, intended to be fixed in said public transport vehicle (800) and comprising at least one audio capture means for capturing at least one audio stream, referred to as a second audio stream, and - at least one device (506), referred to as an analysis device, comprising at least one intelligent computing unit for analyzing at least said second audio stream to detect an incident.

10. System (500;600;700) according to any of claims 8 or 9, characterized in that said at least one portable device (5021-502n) is configured to transmit in real time to said at least one analysis device (506) a first captured audio stream and in that said analysis device is configured to analyze in real time the first transmitted audio stream.

11. System (700) according to any of claims 8 to 10, characterized in that it further comprises at least one communication gateway (702) to enable a connection between at least one portable device (5021-502n) and at least one analysis device (506).

12. Public transport vehicle (800) equipped, in particular in an at least one passenger zone (804) of said vehicle (800), with a system (700) according to any of claims 8 to 11.

13. Vehicle (800) according to the preceding claim, characterized in that it is: - a railroad car, - a locomotive, - a train, or - a streetcar.

Images