System and method for detecting the presence of hidden persons in freight / passenger transport vehicles

Abstract

The invention relates to a system and method for detecting people in means of freight / person transport intended for the identification and real-time monitoring of human trafficking at border crossing points, penitentiaries and other facilities. The system according to the invention uses an adaptive filtering module (47) for applying a first level of adaptive filtering of signals using a Recursive Least Squares – Finite Impulse Response RLS-FIR filtering type, which attenuates the noise induced by the ambient environment acquired by the ground vibration sensor (6), from the signal acquired from the people detection sensors (2, 3, 4, 5) and for applying a second level of adaptive filtering of the RLS-FIR type using a Recursive Least Squares – Finite Impulse Response RLS-FIR filtering type for adaptive attenuation of the vibrations generated by the wind present on the Y axis from the signal acquired from the people detection sensors (2, 3, 4, 5). The method according to the invention consists in analyzing the relevant characteristics of the vibration produced by human physiological tremor from the processed and adaptively filtered signals of the person detection sensors (2, 3, 4, 5), namely: the envelope, the autocorrelation, the Fourier transform of the signal on the Z axis with three neural networks: the convolutional neural network (51), the recurrent neural network (52) and the Gated Recurrent Unit neural network (53) to determine the pattern of vibrations generated by human physiological tremor.

Description

System and method for detecting the presence of hidden persons in freight / passenger transport vehicles

[0001] The invention relates to a system and a method for detecting persons in freight / passenger transport vehicles. It is intended for the identification and real-time monitoring of human trafficking at border crossing points, prisons, and other facilities that require rigorous control against the illegal presence of persons in freight / passenger transport vehicles.

[0002] It is known from patent applications [1]US 4415979 A, published on 1983-11-15 (Method and apparatus for detecting the presence of an animate body in an inanimate mobile structure) and [2]US 6370481 B, published on 2002-04-09, "Apparatus and method for human presence detection in vehicles", the fact that the person hidden in freight / passenger transport vehicles will have an impact on them by inducing mechanical vibrations. These mechanical vibrations, originating from the heartbeat, breathing, muscle reflexes and other involuntary movements of the person's body, generate low-level mechanical vibrations in a frequency band between 2Hz and 20Hz, which can be transformed into an electrical signal using geophones placed externally in contact with the vehicle structure and then amplified, analogically processed and converted into discrete signals for processing and analysis.

[0003] FromUS 6370481 B, it is known as a system that determines the presence of hidden persons in freight / passenger transport vehicles that have in their composition geophone type sensors, amplification modules of the signals of the geophone type sensors, analog-digital converters to transform the analog signal into a discrete signal, a computer for processing and analyzing the signals, respectively a display and a keyboard for ensuring the operation by the user. The main shortcomings of this solution are related to the acquisition of omnidirectional vibration signals in which both the signal induced by the hidden persons and the disturbing noises from the air or ground, generated from the outside such as the wind or neighbouring vibrations produced by other devices or vehicles are additively found.

[0004] From the international patent application [4]WO 2008101725 A2,"Method for detecting movement,"2008, it is known as a method for detecting movement in a vehicle such as trailers, trains, or containers that have spring suspension. The main disadvantage of this method is that prior knowledge of the spring system and the vehicle load is required to make a decision regarding the detection of a person's presence in it.

[0005] From the patent [5]US 7353088 B2, "System and method fordetecting the presence of human in a vehicle ", published in 2008, a system and method for detecting human presence in a vehicle are known that analyzes vibrations coming from sensors with a processor running an algorithm based on a neural network with a series of nodes, at least some of which are recurrent. The disadvantage of this method is that the accuracy of the neural network is low when disturbing vibrations coming from the air or ground are superimposed on the vibrations generated by human presence due to direct vibration analysis and a network with a relatively small number of nodes.

[0006] The invention solves the problem of eliminating disturbing vibrations originating from the ground and the air by detecting people in freight / passenger transport vehicles.

[0007] The system for detecting the presence of hidden persons in freight / passenger transport vehicles solves the technical problem by being composed of:

[0008] four people detection sensors mounted on the chassis of the vehicle under test using magnets, each consisting of a first metal casing resistant to shocks and water or dust penetration in which a first microcontroller is placed acquiringsignals from geophone-type sensors incorporated in the casing,on three axes: X - horizontal transverse to the vehicle -, Y - horizontal longitudinal to the vehicle - and Z - vertical to the vehicle - which are impedance-adapted and pre-amplified with signal conditioning modules that programmably amplify and transform the signal from analog to digital through a first programmable amplification and analog-digital conversion module;

[0009] a ground vibration detection sensor placed on the ground near the vehicle under test,containing of a second metal casing resistant to shocks and water or dust penetration in which a second microcontroller is placed acquiring signals from another built-in sensor of the geophone type on the vertical Z axis, signals that are impedance-adapted and pre-amplified with another signal conditioning module, respectively amplified and transformed from analog to digital through a second programmable amplification and analog-to-digital conversion module;

[0010] an anemometer mounted at the base of a tripod near the vehicle under test, consisting of a third metal casing resistant to shocks and water or dust penetration, in which a third microcontroller is placed that takes the wind speed information from a wind speed measurement sensor placed on the same tripod at a height of 2m;

[0011] a central unit placed on the ground near the sensors at a distance of no more than 70m, consisting of a main metal casing resistant to shocks and water or dust penetration in which an embedded computing subsystem is placed that communicates with the people detection sensors, the ground vibration detection sensor and the anemometer through a communication module, acquiring data from them through a first data acquisition module from the people detection sensors and through a second acquisition module from the ground vibration sensor, controlling the amplification level of their signals and monitoring their status;

[0012] an adaptive filtering module for adaptive filtering of vibration signals corresponding to the three axes X, Y and Z from the people detection sensors and of disturbing ground vibration signals on the Z axis from the ground vibration detection sensor;

[0013] a signal feature extraction module for extracting signal features from person detection sensors;

[0014] a detection stop module for stopping detection if the wind speed exceeds a preset limit determined by the data acquisition module from the anemometer and calculates the wind speed;

[0015] a user interface display device based on WEB (World Wide Web) technology that displays the detection results.

[0016] According to one aspect of the invention, the power supply and charging of the central unit battery are carried out by a power supply voltage source from the electrical network, and the batteries of the detection sensors, the ground vibration detection sensor, and the anemometer are charged using electrical cables via the central unit.

[0017] According to another aspect of the invention, the communication module is of the WiFi type and uses the TCP / IP protocol.

[0018] According to another aspect of the invention, the communication module communicates with the person detection sensors, the ground vibration detection sensor, and the anemometer via electrical cable using the TCPIP communication protocol in the event of excessive electromagnetic load or radiofrequency interference.

[0019] According to another aspect of the invention, the electrical power supply of the central unit modules is achieved using a charging and power supply module directly from the power supply voltage source from the electrical network or an integrated battery.

[0020] The technical problem is also solved by a method for detecting persons in freight / passenger transport vehicles that uses the above system, which consists of the following steps:

[0021] acquisition and sampling of signals coming from the vibrations of means of transport / freight, on which are magnetically attached some people detection sensors and a ground vibration detection sensor with a frequency of 63Hz;

[0022] taking data from people detection sensors through a first data acquisition module andfromthe ground vibration sensor through a second data acquisition module,the mentioned acquisition modulesensuring data synchronization over time, because they come from sensors over WiFi communication using a TCP / IP protocol, which can delay data packets depending on the load of the radio communication channel;

[0023] applying a first level of adaptive filtering of signals through the adaptive filtering module using a Recursive Least Squares – Finite Impulse Response RLS-FIR type filtering which attenuates the noise induced by the ambient environment acquired by the ground vibration sensor, from the signal acquired from the people detection sensors;

[0024] applying a second level of adaptive RLS-FIR filtering through the adaptive filtering module using a Recursive Least Squares – Finite Impulse Response RLS-FIR filtering for adaptive attenuation of wind-generated vibrations on the Y axis from the signal acquired from the people detection sensors;

[0025] adaptive filtering of filtered signals from people detection sensors using an FIR-type bandpass filter with cutoff frequencies between 4Hz and 15Hz, respectively, and then decimated with the decimation factor M=2;

[0026] permanent reading of wind speed through the anemometer data acquisition module and calculating wind speed to stop detection if the wind is strong and the vibrations generated from the air have a much higher amplitude than those generated by the person through a detection stop module;

[0027] when detection occurs, providing the processed signals from the output of the adaptive filtering module to the input of the signal feature extraction module thatextracts the relevant characteristics of the vibration produced by human physiological tremor from the processed signals of the people detection sensors, namely: the envelope, the autocorrelation, the Fourier transform of the signal on the Z axis;

[0028] - feature analysis with three neural networks: convolutional neural network, recurrent neural network and Gated Recurrent Unit neural network for determining the vibration pattern generated by human physiological tremor, the three neural networks generating a numerical percentage value between 1% and 100% indicating the confidence whether a person is hidden in the freight / passenger transport vehicle;

[0029] - making the person detection decision based on the results obtained with the neural networks,andiftwo of the three algorithms implemented by the three neural networks give a percentage result greater than 50%, a person detection alarm is generated in the freight / passenger transport vehicles under control;

[0030] - displaying the results on the user interface device based on a WEB (World Wide Web) technology that displays the detection results.

[0031] According to one aspect of the invention, the signal generated by the person is obtained by performing adaptive filtering through the filter module, attenuating the disturbing vibrations in the air of the vibration signals corresponding to the three axes X, Y, and Z from the person detection sensors, and attenuating the disturbing vibrations in the ground on the Z axis from the ground vibration detection sensor.

[0032] By applying the invention, the following advantages are obtained:

[0033] -         The system for detecting person in freight / passenger transport vehicles is portable, easy to use and ensures a long operating autonomy without an external power supply;

[0034] -         The batteries can be charged from a 220Vac / 50Hz or 110Vac / 60Hz electrical network or from a 12Vcc / 24Vcc car socket;

[0035] -         The detection process, depending on the vehicle type selected, automatically configures the detection sensors and is done without any further operator intervention;

[0036] -         The system is operated through an easy-to-use user interface, based on web technology, on any display device such as a mobile or desktop computer, tablet or smartphone;

[0037] -         Elimination of disturbing vibrations in the air through adaptive processing of signals from people detection sensors on three axes X, Y and Z;

[0038] -         Elimination of disturbing ground vibrations through adaptive processing of signals from person detection sensors and ground vibration detection sensors;

[0039] -         Increasing detection accuracy by extracting features and analyzing signals from person detection sensors using three algorithms based on neural networks;

[0040] -         A limit on the amplitude of the signals is not required for the detection of a person hidden in the means of transport of goods / people;

[0041] -         It offers a high probability of detection in conditions of disruptive noise of external vibrations coming from the ground and air;

[0042] An embodiment of the invention is given below, about figures 1, 2, 3, 4, 5 and 6a, 6b, 6c, which represent:

[0043] -        – Block scheme of the system for detecting persons in freight / passenger transport vehicles, according to the invention;

[0044] -        – Block scheme of the central unit, according to the invention;

[0045] -        – Block scheme of the people detection sensors, according to the invention;

[0046] -        – Block scheme of the ground vibration detection sensors, according to the invention;

[0047] -        – Block scheme of the anemometer, according to the invention;

[0048] -        a, b, c – Flowchart of the method for detecting persons in freight / passenger transport vehicles, according to the invention;

[0049] The system that determines the presence of hidden people in the freight / personnel transport vehicle consists of a central unit1running a software application that communicates with four people detection sensors2,3,4,5, aground vibration detection sensor6, an anemometer7.The results' configuration / operation and presentation is done on a user interface display device9based on a WEB (World Wide Web) technology. The power supply of central unit1and the charging of battery14is carried out through a power supply voltage source from the electrical network8in which the electrical energy can be provided both from a 220Vac / 50Hz, respectively 110Vac / 60Hz electrical network and from the 12Vcc / 24Vcc car socket. The charging of the batteries of the people detection sensors2, 3, 4, 5,the ground vibration sensor6and the anemometer7are carried out with electrical cables via the central unit1.

[0050] The central unit1, as shown in, consists of a metal casing15resistant to shocks and water or dust penetration in which the embedded computing system10is placed, which communicates with all sensors2, 3, 4, 5, 6and7through the wireless communication module12. Ifan excessive electromagnetic load, radiofrequency interference, or the sensors' batteries are charged during operation, the embedded computing system10can communicate with the sensors through the wired communication module11. The power supply and the provision of all voltage levels of the central unit modules are carried out through the charging and power module13directly from the power supply voltage source from the electrical network8or the battery14in the component.

[0051] The central unit1, through wired or wireless communication, acquires data from the people detection sensors2,3,4,and 5, controls the signal amplification level and monitors their status, acquires data from the ground vibration detection sensor6and the anemometer7,monitors the charge level of the batteries27,36,43respectivelyof sensors and the of anemometer, and the level of the radio frequency signal reception power from the wireless communication modules24, 33and40of the sensors and the anemometer7.

[0052] When receiving electrical energy from the power supply voltage source from the electric10 al network8,the charging and power module 13 can also power the sensors2, 3, 4, 5, 6, and 7to charge their batteries14,27,36,and 43when they are connected to the central unit1with the electrical power and communication cables via the wired communication module11.

[0053] The people detection sensors2,3,4, and 5,as shown in the block scheme in, comprise a metal housing28resistant to shocks and water or dust penetration in which the microcontroller16is placed. It acquires the signals from the geophone type sensors21,22,and 23on three axes X (horizontal transverse to the vehicle), Y (horizontal along the vehicle) and Z (vertical) which are impedance-adapted and pre-amplified with signal conditioning modules18,19,20that programmable amplify and transform the signal from analog to digital through a first programmable amplification and analog-digital conversion module17.

[0054] The ground vibration detection sensor6, as shown in the block scheme in, consists of a metal housing37resistant to shocks and water or dust penetration in which the microcontroller29is placed,which acquires the signals from the geophone sensor32on the Z (vertical) axis, which are impedance-adapted and pre-amplified with the signal conditioning module31,respectively amplified and converted from analog to digital by the programmable amplification and analog-to-digital conversion module30.

[0055] The anemometer6,as shown in the block scheme in, consists of a metal housing44resistant to shocks and water or dust penetration in which the microcontroller38is placed,which receives wind speed information from the wind speed measurement sensor39.

[0056] The wireless communication modules 12, 24, and 40, which are included in the central unit 1, can wirelessly communicate the people detection sensors2,3,4,5,the ground vibration detection sensor6,and the anemometer7with each other. The modules also encrypt and determine the level of the radio frequency signal upon reception.

[0057] If the radio frequency band is jammed, loaded with other radio frequency communication devices, or the sensors' batteries are charged during operation, the communication of the people detection sensors2,3,4,5,the ground vibration detection sensor6and the anemometer7with the central unit1can be achieved with electrical communication cables through the wired communication modules25,36and41included in them.

[0058] The person detection sensors2,3,4,5,the ground vibration detection sensor6and the anemometer7can operate autonomously, powered by the batteries27,36,and 43from their composition. Charging and power supply modules26,35,and 42,in this case, monitor and charge the batteries, respectively ensuring all the electrical voltages necessary for the sensor components.

[0059] In the situation where the people detection sensors2,3,4,5,the ground vibration detection sensor6,and the anemometer7are connected via the electrical cable to the central unit1,they are powered and communicate with the central unit simultaneously via the wired communication modules25,34,and41.

[0060] The detection method is based on the flowchart in. According to an embodiment of the invention, the signals coming from the vibrations of the freight / personnel transport vehicle, by which the people detection sensors are magnetically attached, are acquired and sampled within the people detection sensors2, 3, 4, 5and the ground vibration detection sensor6with a frequency of 63Hz. The data acquisition module45from the people detection sensors takes the data from the people detection sensors. The data acquisition module46from the ground vibration sensors takes the data from the vibration detection sensor.Modules45and46also have the role of synchronizing the data in time, because they come from sensors over WiFi communication using a TCP / IP protocol, which can delay the data packets depending on the load of the radio communication channel. The adaptive filtering module47uses an RLS-FIR (Recursive Least Squares – Finite Impulse Response) type filtering through which it attenuates the noise induced by the ambient environment acquired by the ground vibration detection sensor6, from the signal acquired from the sensors2, 3, 4, 5.It was also observed that the signal from the human physiological tremor is predominantly on the Z axis. Predominant vibrations, generated by the wind, are found on the Y axis for which module47applies the second level of adaptive filtering of the RLS-FIR type for each sensor2, 3, 4, 5for the adaptive attenuation of the vibrations generated by the wind present on the Y axis from the Z axis. The adaptively filtered signals from sensors2, 3, 4, and 5are filtered with a FIR (Finite Impulse Response) bandpass filter with cutoff frequencies between 4Hz and 15Hz, respectively, and then decimated with the decimation factor M=2. The data acquisition module49from the anemometer reads the permanent wind speed. It calculates the wind speed to stop the detection if the wind is strong and the vibrations generated from the air have a much higher amplitude than those generated by the person by the detection stop module48.When the detection occurs, the signals processed from the output of module47are provided by module48to the input of the signal characteristics extraction module50.Module50extracts the relevant characteristics of the vibration produced by human physiological tremor from the processed signals of sensors2, 3, 4, and 5,namely: the envelope, the autocorrelation, the Fourier transform of the signal on the Z axis, but other characteristics relevant to the detection of people can be added through the signal characteristics extraction module50.To classify vibrations for detecting the person in the vehicle, the characteristics obtained at the output of module50constitute the input data of the convolutional neural network modules51,the recurrent neural network52and the Gated Recurrent Unit neural network53.The three neural networks50,51and52will generate a numerical percentage value between 1% and 100% indicating confidence whether a person is hidden in the freight / passenger transport vehicle. If the percentage is greater than 50%, then there is a probability that a hidden person is present. The person detection module54 makes the final decision,which analyzes the results of the three neural networks. If two of the three algorithms50,51and52give a percentage result greater than 50%, the system generates a person-detected alarm in the freight / passenger transport vehicle under control.

[0061] Obviously, the invention is not limited to the embodiments and operation described and shown above. It may be varied or modified without abandoning the fundamental principle mentioned above and claimed hereinafter.

Claims

System for detecting the presence of hidden persons in freight / passenger transport vehicles,characterized by the fact thatit consists of:four people detection sensors (2,3,4,5)mountedon the chassis of the vehicle under test using magnets, each containing a first metal housing (28)resistant to shocks and to the penetration of water or dust in which a microcontroller (16)is placed acquiringsignals from some geophone-type sensors (21,22,23)incorporated in the housing(1),on three axes X: - horizontal transverse to the vehicle -, Y - horizontal along the vehicle - and Z - vertical to the vehicle - which are impedance-adapted and pre-amplified with some signal conditioning modules (18,19,20),which programmable amplify and transform the signal from analog to digital through a programmable amplification and analog-digital conversion module (17);a ground vibration detection sensor (6) placed on the ground near the vehicle under test, containing a second metal casing (37)resistant to shocks and water or dust penetration in which another microcontroller (29)is placedacquiringsignals from another geophone type sensor (32)built-in on the vertical Z axis, signals that are impedance-adapted and pre-amplified with another signal conditioning module (31),respectively amplified and transformed from analog to digital by another programmable amplification and analog-to-digital conversion module (30);an anemometer (7)mounted at the base of a tripod in the vicinity of the vehicle under test and containing a third metal casing (44)resistant to shocks and water or dust penetration in which a third microcontroller (38)is placed,which takes wind speed information from a wind speed measurement sensor (39)placed on the same tripod at 2m height,a central unit (1)placed on the ground near the sensors at a distance of no more than 70m,containing a main metal casing (15)resistant to shocks and water or dust penetration in which is placed an embedded computing subsystem (10) that communicates with the people detection sensors (2,3,4,5), withthe ground vibration detection sensor (6)and withthe anemometer (7)through a communication module (12), acquiring data from them through a first data acquisition module (45)from the people detection sensors (2,3,4,5)and through a second acquisition module (46)from the ground vibration sensor, controlling the amplification level of their signals and monitoring their status;an adaptive filtering module (47)for adaptive filtering of vibration signals corresponding to the three axes X, Y and Z from the people detection sensors (2,3,4,5)and of the disturbing ground vibration signals on the Z axis from the ground vibration detection sensor (6);a module for extracting people's characteristics (50)for extracting the signal features from the person detection sensors (2,3,4,5);a detection stopping module (48)for stopping detection if the wind speed exceeds a preset limit determined by the data acquisition module (49) from the anemometer (7) and calculating the wind speed;a user interface display device (9)based on WEB (World Wide Web) technology that displays the detection results.    Detection system, according to claim 1,is characterized in thatthe power supply and charging of the battery of the central unit (1)is carried out by a power supply source (8)from the electrical network and the charging of the batteries of the detection sensors(2,3,4,5),of the ground vibration detection sensor (6)and ofthe anemometer (7)is connected via electrical cables via the central unit (1).The detection system, according to claim 1,is characterized in thatthe communication module (12)is wireless.The detection system, according to claim 1,is characterized in thatthe communication module (12), in the event of an excessive electromagnetic load or radio frequency interference, communicates with the people detection sensors (2,3,4,5),withthe ground vibration detection sensor (6)and withthe anemometer (7)via electric cable using the TCP / IP communication protocol.The detection system, according to claim 1,is characterized in that,according to claim 1, the power supply of the central unit modules is carried out by means of a charging and power supply module (13)directly from the power supply voltage source from the electrical network (8)or from a battery (14)in the component.Method for detecting the presence of hidden persons in freight / passenger transport vehicles using the system of claim 1,characterized in that itincludes the following steps:acquisition and sampling of signals coming from the vibrations of hidden persons in freight / passenger transport vehicles, on which people detection sensors are magnetically attached (2, 3, 4, 5)and a ground vibration detection sensor (6) with a frequency of 63Hz;taking data from the people detection sensors (2, 3, 4, 5)through a first data acquisition module (45)andfrom the ground vibration sensor (6)by a second data acquisition module (46),said acquisition modulesensuring data synchronization over time because they come from sensors over WiFi communication using a TCP / IP protocol, which can delay data packets depending on the load of the radio communication channel;applying a first level of adaptive filtering of signals through the adaptive filtering module (47)using a Recursive Least Squares – Finite Impulse Response RLS-FIR type filtering, which attenuates the noise induced by the ambient environment acquired by the ground vibration sensor (6),from the signal acquired from the people detection sensors (2, 3, 4, 5).applying a second level of adaptive RLS-FIR filtering through the adaptive filtering module (47)using a Recursive Least Squares – Finite Impulse Response RLS-FIR filtering for adaptive attenuation of wind-generated vibrations present on the Y axis from the signal acquired from the people detection sensors (2, 3, 4, 5);adaptive filtering of the filtered signals from the people detection sensors (2, 3, 4, 5)using an FIR-type bandpass filter with cutoff frequencies between 4Hz and 15Hz, respectively, and then decimated with the decimation factor M=2;permanent reading of the wind speed through the data acquisition module (49)from the anemometer (7)and calculating the wind speed to stop the detection if the wind is strong and the vibrations generated from the air have a much higher amplitude than those generated by the person through a detection stop module (48);when detection occurs, providing the processed signals from the output of the adaptive filtering module (47)to the input of the signal features extraction module (50),whichextracts the relevant characteristics of the vibration produced by human physiological tremor from the processed signals of the people detection sensors (2, 3, 4, 5),namely: the envelope, the autocorrelation, the Fourier transform of the signal on the Z axis; - feature analysis with three neural networks: the convolutional neural network (51),the recurrent neural network (52)and the Gated Recurrent Unit neural network (53)for determining the vibration pattern generated by human physiological tremor, the three neural networks (51,52and53)generating a numerical percentage value between 1% and 100% indicating the confidence whether a person is hidden in the means of transport of goods / people;- making the person detection decision based on the results obtained with the neural networks (51,52and53), andif two of the three algorithms implemented by the three neural networks (51,52and53)give a percentage result greater than 50%, a person detected alarm is generated in the means of passenger / freight transport under control.- displaying the results on the user interface display device (9)based on a WEB (World Wide Web) technology that displays the detection results.The detection method, according to claim 12,is characterized in thatto obtain the signal generated by the person, adaptive filtering is performed by the filtering module (47)by attenuating the disturbing vibrations in the air of the vibration signals corresponding to the three axes X, Y and Z from the person detection sensors (2,3,4,5)and by attenuating the disturbing vibrations in the ground on the Z axis from the ground vibration detection sensor (6).BIBLIOGRAPHY SHEET[1] US 4415979 A, Method and apparatus for detecting the presence of an animate body in an inanimate mobile structure, 1983[2] US 6370481 Bl, Apparatus and method for human presence detection in vehicles, 2008[3] US 2011051569 A1, System for detecting heartbeats, 2011[4] WO 2008101725 A2, Method for detecting movement, 2008[5] US 7353088 B2, System and method for detecting presence of human in a vehicle, 2008

Images