This disclosure relates to detection of surface and buried objects.
A large percentage of land mines contain some amount of metal. Many types of mines use metal for firing pins, shrapnel, and portions of the casing. If a mine has a sufficient quantity of a detectable metal, that mine can be found using a metal detector.
Techniques for detecting surface and buried objects, such as land mines, unexploded ordnances, pipes, reserves of liquid, and power lines are disclosed. In particular, a movable and controllable arm is attached to a platform such that a sensor head coupled to the arm may move independently of the platform. The sensor head includes sensors that are used to image the surface of the ground and/or a region beneath the surface of the ground. The sensor head may be used to image through or into other turbid, dense, or compacted media.
The platform may be, for example, an autonomous robotic vehicle or a platform that is movable from place-to-place but is stationary while the sensor head is collecting data. Thus, in the techniques discussed below, the motion of the sensor head is not dependent upon the motion of the platform. As compared to techniques in which the motion of the detection sensors corresponds to the motion of the platform, moving the sensor head independently of the platform may allow for an increased scan speed, improved detection rates and/or lower false alarm rates, and the ability to cover larger swaths of ground than a system in which the sensor head and platform move together.
As discussed below, the platform is stationary while the sensor head scans over a region. Keeping the platform stationary may improve the quality of the data collected by the sensor head because the collected data is not contaminated by noise and artifacts that may result from the motion of the platform. For example, if the platform moves while the data is collected, jitter resulting from the motion of the platform may appear as noise in the collected data. The jitter may be worse in situations in which the platform travels over rough terrain. Additionally, by moving the sensor head independently of the platform, the sensor head may be readily repositioned to re-scan a particular area. In contrast, in systems in which the sensor head moves with the platform, the entire platform is repositioned in order to rescan a particular area. Repositioning the entire platform may take longer than repositioning the sensor head. Additionally, the techniques discussed below may free up space on the platform for other items, such as, for example, a neutralization device and/or a marking device.
The sensor head may be an integrated sensor head that includes a continuous-wave metal detector (CWMD) and a ground-penetrating radar (GPR), both of which are located in the sensor head.
In one general aspect, a sensor head includes a ground penetrating radar (GPR) system and a continuous-wave metal detector (CWMD). The GPR system includes a transceiver configured to transmit radiation toward an object and to receive radiation from the object. The CWMD includes a transmission antenna configured to produce a first magnetic field in the vicinity of the object sufficient to generate a current in the object, and a receive antenna configured to sense a second magnetic field produced by the current generated in the object.
Implementations may include one or more of the following features. The outer surface of the sensor head is formed by rigid material that forms part of the GPR. The sensor head also may include a housing that holds the GPR and CWMD. The GPR system may include at least two transceivers and the receive antenna of the CWMD is positioned between the two transceivers. The GPR transceiver may include an antenna configured to transmit the radiation and an antenna configured to receive the radiation. The sensor head also may include a shell, the GPR transceiver may be mounted in an opening formed in the shell, and the receive antenna of the CWMD may be placed about the opening such that the receive antenna and the GPR transceiver are interleaved. The receive antenna of the CWMD may be wrapped around the opening and another opening in the shell such that the receive antenna forms a figure-eight shape. The GPR system includes multiple antennae configured to transmit and receive radiation. The GPR system and the CWMD may be co-located in the sensor head.
In another general aspect, a system includes a sensor head including a continuous-wave metal detector (CWMD) and a ground-penetrating radar (GPR). An articulating arm is coupled to the sensor head. The articulating arm configured to move the sensor head independently of a platform on which the articulating arm is mounted.
Implementations may include one or more of the following features. The system may include an electronic processor and an electronic storage. The electronic storage may include instructions, that when executed, cause the processor to receive data from the sensor head, and control the articulating arm to position the sensor head. The data may be received from the GPR and CWMD in parallel. The electronic processor and the electronic storage may be mounted on the articulating arm. The platform may be a movable platform. The articulating arm may include a non-metallic material. The system also may include a rotation plate coupled to the articulating arm and the sensor head. The rotation plate may be coupled to the articulating arm at a pivot point to allow the sensor head to rotate in all directions about the pivot point.
In another general aspect, a method of scanning a region for subsurface objects includes positioning a platform in the vicinity of a region having a surface and a subsurface, controlling an articulating arm coupled to the platform and holding a sensor head to position the sensor head at a first position above a first portion of the region, and controlling the articulating arm to move the sensor head through a swath to a second position above a second portion of the region. The motion of the sensor head is independent of a position of the platform.
Implementations may include one or more of the following features. The platform may be moved to a position in the vicinity of a second region having a surface. The platform may stationary while the articulating arm moves the sensor head from the first position to the second position. The sensor head may be activated while the sensor head moves from the first position to the second position. Data representative of the subsurface of the swath may be received from the sensor head. The data received from the sensor head may be analyzed, a region of interest may be identified in the data, a position of the region of interest may be determined, and the sensor head may be moved to the position of the region of interest while the platform is stationary. After moving the sensor head to the position of the region of interest, the sensor head may dwell over the region of interest to collect first data representative of the region of interest and second data representative of the region of interest. Receiving data from the sensor head may include receiving data from a GPR and CWMD that are included in the sensor head.
In another general aspect, an antenna includes a resistive element formed in a casing, and a feed line formed in the casing and electrically coupled to the resistive element. The positioning of the resistive element and the feeding is such that the feed line is approximately perpendicular to the resistive element.
Implementations may include one or more of the following features. The resistive element may be a resistive vee. The positioning of the feed line and the resistive element may result in the antenna being sized to allow integration of the antenna into a sensor head. A radar-absorbing material may surround the feed line.
In another general aspect, a method of operating an integrated sensor head includes receiving data collected by a continuous-wave metal detector (CWMD). The data includes a representation of a magnetic field produced by a current flowing in a metallic object that has a known position relative to the CWMD. The received data is analyzed to identify the representation of the magnetic field produced by the metallic object, and the received data is compensated to minimize the identified representation.
Implementations may include one or more of the following features. The metallic object may include one or more of a metallic portion of a ground penetrating radar (GPR) included in the sensor head with the CWMD and a metallic arm coupled to the sensor head. Compensating the received data may include removing the representation. A pre-determined constant level may be removed from the received data.
Other implementations are within the scope of the claims. Implementations may include a method or process, a system or apparatus, an antenna, a sensor head, or computer software stored on a computer-accessible medium.
Like reference numbers refer to like elements.
Referring to
In the example shown in
The sensor head 110 may include more than one sensor configured to detect surface and buried objects. Thus, the sensors included in the sensor head 110 may be considered to be integrated together in the sensor head 110. The height “h” of the sensor head 110 above ground 150 determines the height above ground 150 of each of the sensors included in the sensor head 110. Accordingly, integration of the multiple sensors in the sensor head 110 may eliminate the need to determine and/or control the height of the individual sensors, which may lead to a simplified design and improved performance.
The sensor head 110 may include, for example, a ground penetrating radar sensor, such as a GPR transceiver 940 shown in
The electronics module 130 includes a processor 132 and an electronic storage 134 that stores instructions for processing data received from the sensor head 110 and for performing autonomous object detection on the data received from the sensor head 110. The electronics module 130 may format the data received from the sensor head 110 into images that are automatically analyzed for the presence of surface or buried objects. The analysis may include anomaly detection that identifies regions of interest within the images. The regions of interest correspond to physical locations that may include surface or buried objects. The analysis also may include additional processing of the regions of interest with, for example, parallel sets of neural network classifiers. The additional processing may be performed on data collected in an initial scan made by the sensor head 110 and/or the additional processing may be performed on data collected in an additional scan. The additional scan may be made by moving the sensor head 110 with the arm 120 (independently of the platform 115) to a physical region associated with the region of interest identified by the anomaly detector.
In the example shown, the platform 115 is wheeled. However, in other examples, the platform 115 may be a tracked robotic vehicle. The forearm 122 may be made from a plastic or other lightweight material.
Referring to
Referring to
Referring to
By allowing the sensor head 110 to move independently of the platform 115, the techniques discussed above and in the following description may help address challenges presented by “plow” or “push broom” systems in which the sensors (such as ground penetrating radars and metal detectors) are fixed on the platform (e.g., a vehicle) such that the sensors move with the platform. For example, the sensors of “plow” and “push broom” systems typically collect data over a swath having a size that is determined by the size of the sensor. For example, a “plow” system may include a radar that is one-half-meter wide, and, thus, the “plow system” collects data over a one-half-meter wide swath as the vehicle moves through the swath. In contrast, as a result of being mounted on the arm 120, the sensor head 110 may be swept over a wider area as compared to the width of the platform and the sensor head. Additionally, the data collected by the “plow” system may be contaminated by noise that occurs as a result of the vehicle moving as the system collects data.
Accordingly, mounting the sensor head 110 on the arm 120 allows the sensor head 110 to move independently of the platform 115 and may result in improved performance as compared to systems in which the motion of the sensor head is determined by the motion of the platform on which the sensor head is mounted.
Referring to
As shown in
The rotation plate 410 couples to the sensor head 110 through bearings 413 mounted to the rotation plate 410 through tracks 414a and 414b. The bearings 413 allow the sensor head 110 to rotate about the forearm 122 in the x-y plane. The bearings 413 are held in the tracks 414a and 414b, which, in the example of
Referring to
The process 700 may be performed by one or more processors in an electronics interface, such as processor 132 of the electronics module 130 discussed with respect to
Referring to
Referring also to
Referring also to
As shown in
Referring to
Referring to
Thus, the sensor head 110 may be scanned or moved independently of the platform 115.
As discussed above, in some implementations, the sensor head 110 includes multiple sensors. In general, a sensor produces a measure of a phenomenology detected by the sensor. For example, ground penetrating radar (GPR) measures dielectric contrast, metal detectors produce an indication of an amount of metal in an object of interest, and an X-ray detector produces an indication of a material's ability to absorb X-ray radiation. Thus, using multiple different types of sensors to detect how a particular object interacts with different types of radiation may provide more information about an object than a single sensor. For example, a metal detector and a GPR together may detect metallic objects, non-metallic objects, and objects that include both metallic and non-metallic components. However, a system that only includes the metal detector may only detect objects that include metal. Thus, the multi-sensor system may have better performance than a single-sensor system. Additionally, locating various complementary sensors in a single sensor head may further improve performance by reducing (perhaps eliminating) the need to register the data collected by the various sensors that view different aspects of the scene. Registration typically includes sampling a scene or region from different view points (such as sensors located in different positions relative to the region) and then transforming the collected data into a common coordinate system. Because multiple sensors are placed in the sensor head 110, the multiple sensors view the same portion, or nearly the same portion, of the region. As a result, the need to register the data from the various sensors included in the sensor head 110 may be reduced.
In some implementations, the sensor head 110 includes a continuous wave metal detector (CWMD) and a ground penetrating radar (GPR), such as CWMD 905 and GPR 940 of
Some prior systems employed a GPR and a CWMD, but, in these systems, the GPR and the CWMD were separated by a distance sufficient to prevent the background or fixed metallic objects (such as metallic components of the GPR or the arm 120) from being detected by the CWMD. The separation distance in such systems may be one meter or more. Detection of background or fixed metallic objects by the CWMD creates noise in the CWMD data that results in the CWMD data being less than optimal. In contrast to these prior systems, the sensor head 110 includes a GPR and a CWMD in a single integrated sensor head, such as head 900 shown in
Referring to
The exterior of the example integrated sensor head 900 is shown in
Referring to
A first magnetic field is produced in the vicinity of an object (1010). The object has an orientation relative to a direction of propagation of the first magnetic field and the first magnetic field induces a current in the object. In-phase and quadrature (“I&Q”) data representing the second magnetic field is sensed as a current arising in a coil of the sensor (1020). The sensed data is fit to a two-dimensional signature (1030). The two-dimensional signature may be a signature that represents the quadrature data as a function of the in-phase data.
A template of data that is independent of the orientation of the object relative to the first magnetic field is generated (1040). The template of data also may be independent of an orientation of the object relative to a direction of propagation of radiation produced by the sensor and directed toward the target. The template of data may be a template that represents a three-dimensional object associated with a two-dimensional signature that matches, or closely matches, the two-dimensional signature found in (1030). The three-dimensional object may be found from among multiple candidate three-dimensional object by iterating through the potential three-dimensional space of I&Q data that could project into the two-dimensional signature found in (1030). The number of candidate objects may be reduced by removing non-logical values (non-positive values) until the iteration converges to a unique candidate three-dimensional model that projects the two-dimensional I&Q signature found in (1030) in real (positive) values.
In the model, the shape and material of each of the metallic objects is described using vectors representing amplitude and frequency, where frequency is the relaxation rate of the signature measured after being influenced by the electromagnetic field produced by the sensor. Because the three-dimensional model is a close approximation to the detected object, the orientation of the detected object relative to the sensor may be accounted for, and the vectors are independent of the relative orientation of the detected object and the sensor.
A feature of the object is extracted from the three-dimensional template (1050). The feature of the object is extracted from data that is derived from, or produced by, the three-dimensional template, such as the amplitude and frequency vectors discussed above.
Extracting a feature of the object may include determining an amplitude of the second magnetic field and determining a frequency of the second magnetic field or the relaxation rate of the detected object after being influenced by the electromagnetic field produced by the sensor. Extracting a feature of the object may include identifying, from the frequency vector, a first frequency value and a second frequency value. Extracting a feature of the object may include identifying, from the amplitude vector, a first amplitude value and a second amplitude value. In some examples, the feature may include a ratio of the first frequency value and the second frequency value and a ratio of the first amplitude value and the second amplitude value. Using the ratio instead of the raw frequency and amplitude values as the extracted feature values may remove noise from the value of the feature, particularly if the noise is common to all frequency values and/or all amplitude values. The first and second frequency values may be the two highest frequency values, and the first and second amplitude values may be the two highest amplitude values. The first and second amplitudes may be the amplitudes respectively associated with the first and second frequencies.
In some examples, a distance between the detected object and the sensor may be estimated. The estimated distance between the detected object and the sensor may be used to normalize the data collected by the sensor to a constant, arbitrary distance before extracting the feature values of the amplitude and frequency. Determining the distance between the detected object and the sensor allows the extraction and/or use of additional features. For example, the distance itself may be used as a feature.
Whether the object is an object of interest is determined based on the extracted features (1060). To determine whether the object is an object of interest, the extracted feature values may be input into one or more classifiers that are configured to produce a confidence value that may assume a range of numerical values, each of which indicates whether the object is more likely to be a target object or a clutter object. In some examples, the classifier is configured to produce a confidence value that is one of a discrete number of numerical values, each of which indicate whether the object is an object of interest (a target) or an object not of interest (clutter).
Although the example process 1000 includes determining the template of data that is independent of orientation (such as the three-dimensional object), this is not necessarily the case. In some implementations, data produced by the three-dimensional object is received by the processor from a pre-generated or separately generated template of data.
Techniques such as those discussed in
The process 1100 may be performed on one or more processors in communication with the sensor head 900. The processors may be part of the electronics module 130 discussed above with respect to
Data collected by a continuous-wave metal detector (CWMD) is received (1110). The data includes a representation of a magnetic field produced by a current flowing in a metallic object that has a known position relative to the CWMD. The metallic object may be a metallic portion of a GPR (such as an antenna) that is in close proximity to the CWMD and included in the sensor head 900. The metallic object also may be an articulating arm that is in a fixed location relative to the CWMD. The object having a known position relative to the CWMD may be any piece of fixed metallic clutter that is part of a system that includes the CWMD or the object may be a portion of the environment in which the CWMD operates. For example, the object may be a portion of ground that includes metal. The received data is analyzed to identify the representation of the metallic object (1120). The received data is compensated to account for the representation of the metallic object (1130). Compensating for the representation of the metallic object may include removing the representation from the received data. In some implementations, compensating for the metallic object includes minimizing the representation.
The analysis techniques discussed with respect to
In addition, as compared to CWMDs, in general, pulsed metal detectors transmit a pulse and detect an amplitude of a corresponding response signal, whereas a CWMD alternates between generating a fixed frequency signal and an amplitude signal. Changes in amplitude and phase of a corresponding received signal indicate the presence of a metallic object. Because the CWMD senses I&Q data, which includes amplitude and phase, rather than just the amplitude data that is detected by a pulsed metal detector, the effects of metallic clutter (such as soil and metal structural components of a detection system) are more apparent in data collected by the CWMD than in data from a pulsed metal detector. As discussed above, by identifying the metallic clutter as a non-target object, analysis such as that discussed with respect to
Although some systems may have integrated a pulsed metal detector with a GPR, because of the nature of the data collected by a CWMD, the integration of a CWMD with a GPR is different from the integration of a pulsed metal detector and a GPR. Like data collected by a CWMD, data collected by a pulsed metal detector reflects the presence of metallic clutter. However, because data from a pulsed metal detector is amplitude-only, rather than I&Q, the effects of the metallic clutter appears relatively constant in the data collected by the pulsed metal detector. Thus, in systems employing a pulsed metal detector, the presence of metallic clutter may be removed (or otherwise compensated for) by performing an analysis that, for example, removes a constant level representing the metallic clutter from metal parts of the device from the signal.
However, removing a constant level from the I&Q data collected by a CWMD may introduce inaccuracy such that the CWMD data may be unusable to accurately discriminate between different types of objects.
The effects of the metallic clutter on the I&Q data collected by a CWMD may vary based on metallic clutter in the sample being searched. For example, the metallic clutter may be a metallic soil in which an underground pipe (the target) is buried. The amount and distribution of the metal in the soil may vary slightly over a region scanned by the CWMD. The I&Q data from the CWMD reflects the variation more than amplitude data collected from a pulsed metal detector scanned over the same region. Thus, use of an analysis such as that discussed with respect to
Additionally, the techniques discussed with respect to
Referring to
The CWMD 905 produces a magnetic field sufficient to induce a current in a metallic object in the vicinity of the CWMD 905. The induced current flows in the metallic object and creates a second magnetic field. The parameters of the second magnetic field depend on characteristics of the object such as the amount of metal in the object. The CWMD 905 senses the second magnetic field and generates a representation of the metallic object based on the sensed second magnetic field. The CWMD 905 may transmit a magnetic field signal that has twenty-one frequencies that are logarithmically spaced and within a frequency band of about 330 Hz to 90 kHz.
The GPR transceiver 940 transmits and receives radio-wave (or microwave) signals. The GPR 940 may operate in a frequency range of 700 MHz to 4 GHz. Operation in this frequency band allows for potentially greater depth penetration and improved imaging performance for applications in which the sensor head 900 is used to image a region beneath the surface of the ground. The GPR 940 may be a continuous-wave radar that transmits radio-wave signals having linear polarization.
In some implementations, the sensor head 900 also includes a single-board computer (SBC) 960. The SBC 960 may include the electronic components of the electronics module 130, electronics for interfacing with the GPR 940 and the CWMD 905, and electronic storage for storing instructions to cause a processor to perform data processing such as discussed in
The sensor head 900 also may communicate with an inertial measurement unit (IMU) 975 that tracks the position of the sensor head 900. In the example shown, the IMU 975 receives position measurements from the GPR 940. In other implementations, the IMU 975 may receive position measurements from the CWMD 910, the computer 970, and/or the SBC 960.
Referring to
The antenna 1500 shown in
To achieve the folded over feed design, a radar-absorbing material (RAM), such as C-RAM MT available from Cumming Microwave of Avon, Mass., is placed around the feed-line 1510. The presence of the radar-absorbing material 1520 allows the feed-line 1510 to operated in the folded over position by eliminating or minimizing the effects of currents flowing in the feed line 1510 on the operation of the resistive element 1515. In some implementations, the feed-line 1510 is placed adjacent to a hollow 1520 that is formed between the top ridges 1505 on the first and second antenna halves 1501, 1503 when the halves are butted together. The hollow 1520 is padded, filled, or otherwise includes the radar-absorbing material.
Referring to
To improve the mechanical reliability of the antenna 1500, the resistive element 1515 may be sandwiched between two blocks of a non-conductive material, such as polystyrene foam, and/or the resistive element 1515 may be attached to a dielectric substrate 1525. The foam may be encased in a heat-sealable plastic. Other components of the antenna 1500, such as the feed-line 1510, may be placed between the non-conductive material and/or attached to the substrate. In implementations in which the resistive element 1515 is attached to the substrate 1525, the material of the substrate 1525 has a relative permittivity, or dielectric constant, (∈R) close to 1 to minimize performance degradation caused by the presence of the substrate 1525, the substrate 1525 is relatively thin, and the arms 1516a and 1516b are relatively wide. To further minimize the effects of the substrate 1525, the substrate may be cut out around the resistive element 1515.
The feed line 1510 is coupled to a connector 1530 that allows the signals from the resistive element 1515 and carried by the feed line 1510 to be coupled out of the antenna 1500 for further analysis or coupled into the antenna 1500 to, for example, excite a region with a particular signal.
Referring to
The sensor head 900 includes a rigid exterior. The rigid exterior may be formed from a rigid portion of the antenna 1500. The rigid exterior of the sensor head 900 may include a rigid Styrofoam. The rigid Styrofoam protects the GPR and CWMD sensors while also keeping the weight of the sensor head 900 relatively low. In other examples, the sensor head 900 is placed within a housing that holds the CWMD and the GPR. The housing may be a plastic housing.
The sensor head 900 includes a GPR and a CWMD. In this example, the sensor head 900 includes eighteen v-dipole GPR antennas, three CWMD receive antennas, and one CWMD transmit antenna. The transmit and receive CWMD antennas may be formed as a coil that is sufficiently thin to fit between two GPR antennas. As discussed above, a CWMD transmit antenna emits a magnetic field having sufficient strength to induce current in metallic portions of items in the vicinity of the transmit antenna. The induced current generates a second magnetic field, and the CWMD receive antenna detects that second magnetic field and produces a representation of the strength of the magnetic field. Upon further processing, the representation may be used to identify and/or categorize the object.
In the example of
In operation, the sensor head 900 is swept along a direction 1610 and collects data while scanning. When operated in the direction 1610, the sensor head 900 collects data that may be analyzed or otherwise manipulated to produce an image of the region directly below the sensor head 900. For example, the sensor head 900 may be placed over the ground, and the image produced by the data from the sensor head 900 may include features of the region that are beneath the surface of the ground. The sensor head 900 also may be moved in a direction other than the direction 1610. For example, the sensor head 900 may be moved approximately parallel to the surface of the ground in a direction “O” that is orthogonal to the direction 1610. Because of the arrangement of the GPR antennas along the “W” direction, scanning the sensor head 900 in the direction “O” results in multiple GPR and CWMD antennas examining the same portion of ground, thus resulting in the collection of redundant data. The redundant data may be used in applications that benefit from multiple looks at the same region.
The sensor head 900 also includes supports 955 and 960. The supports 955 and 960 help to support the sensor head 900 and hold the sensor head 900 together. The supports 955 and 960 also may be used to attach the sensor head 900 to the plate 410 and/or to the arm 120. The supports 955 and 960 are coupled to the sensor head 900 with bolts or other fasteners. The sensor head 900 also includes other supports that are internal to the rigid exterior of the sensor head 900.
Referring to
Other implementations are within the scope of the claims. For example, the GPR 940 may operate up to 8 GHz.
This application claims priority to U.S. Provisional Application Ser. No. 61/157,098, titled DETECTION OF SURFACE AND BURIED OBJECTS, and filed on Mar. 3, 2009 and to U.S. Provisional Application Ser. No. 61/243,814, titled CONTINUOUS WAVE METAL DETECTOR, and filed on Sep. 18, 2009. Both of these applications are incorporated by reference in their entirety.
Number | Date | Country | |
---|---|---|---|
61157098 | Mar 2009 | US | |
61243814 | Sep 2009 | US |