Patent Application
20230154329
The present disclosure relates to vehicles, and more specifically to a method and system for obtaining and communicating vehicular information.
Parking space availability is a major problem in crowded areas, particularly urban areas. The importance of better parking systems in urban areas has been recognized recently as one of the most important avenues for betterment of urban infrastructure. One study estimated a loss of $78 billion in one year in the form of 4.2 billion lost hours and 2.9 billion gallons of wasted gasoline in the United States alone. Several projects recently have sought to address this issue through the design of mobile systems that collect traffic congestion information to improve route fmding and trip planning. Unfortunately, a significant portion of traffic congestion and travel delays are experienced in downtown areas where it is not always possible to reroute a driver. In these densely populated urban areas, congestion and travel delays also are due to parking. In one study, researchers found in one small business district of Los Angeles that, over the course of a year, vehicles looking for parking created the equivalent of 38 trips around the world, burning 47,000 gallons of gasoline and producing 730 tons of carbon dioxide. Clearly, addressing the problems associated with parking in downtown areas would have significant societal impact, both economically and ecologically.
Prior attempts to solve this problem often have focused on monitoring the presence or absence of a vehicle over each parking spot using a dedicated sensor. These attempts typically rely on fixed sensors installed by municipalities in the ground or on parking meters. This results in a large fixed cost for installation and operation in order to cover parking spaces at a city-wide level (e.g., millions of dollars to cover a small percentage of the total number of parking spots).
As a result, there remains a need to better address problems associated with parking space availability.
Addressing this problem does not necessarily require real-time identification of individual available parking spots. Instead there is also great value in collecting approximate parking statistics, for example aggregate counts of available parking spots on one road or historical averages of parking spot usage.
For example, such spatio-temporal statistics on parking availability is typically valuable to municipal governments to make better decisions about how to set prices for street-parking, setting time-limits, and where to install parking meters. Beyond adjusting road-side parking prices, detailed parking availability statistics could be widely disseminated on web-based maps or navigation systems which would incur the following further benefits:
In one aspect, a vehicular information system and method includes receiving, by a server computer, sensor information from each of a plurality of sensors. Each sensor in the plurality is associated with a vehicle. The sensor information includes location coordinates of each vehicle in the plurality. The server computer translates the sensor information associated with each vehicle in the plurality to parking statistics information (e.g., counts of spaces per road, fraction of roadway available for parking, historical averages for a road, etc.). In one embodiment, the translation is based on an aggregate of sensor information corresponding to the plurality of vehicles. In one embodiment, the server computer communicates the parking statistics information to the vehicle.
In one embodiment, the receiving of the sensor information includes receiving a range of the sensor and/or receiving a speed of the sensor. In one embodiment, the translating includes determining if each vehicle in the plurality is in a slotted parking area or in an unslotted parking area. If a vehicle is in the slotted parking area, the sensor information is translated to parking space counts. If a vehicle is in the unslotted parking area, the sensor information is translated to a parking space map. In one embodiment, the receiving of sensor information includes receiving video from a camera (e.g., a webcam) associated with each sensor in the plurality. The video can include a plurality of images that are time stamped.
In one embodiment, the receiving further includes determining that the location coordinates fall within a range of location coordinates associated with the start of the receiving. Stopping of the receiving step can occur when the location coordinates fall outside of the range of location coordinates associated with the start of the receiving. In one embodiment, the translating further includes determining the width of a dip in sensor information. The determining of the width can include determining a number of vehicles to which the dip corresponds, comparing the width to a threshold width, and/or determining the depth of a dip.
In one embodiment, the translation is based on an aggregate of sensor information corresponding to the plurality of vehicles. In one embodiment, a location estimate of the vehicle is corrected by matching time-series sensor information (sensor information collected over a period of time) to sensor information from prior trips along a road. In one embodiment, traces are aligned based on distinct signatures generated by fixed road-side objects. In one embodiment, the translating is based on continually comparing time-series sensor information observed by each sensor to a set of signatures known to correspond to typical vehicles. In one embodiment, the translating further includes calculating the parking statistics information over a predetermined period of time (e.g., average parking availability in a city block on a Saturday afternoon over the past month). In one embodiment, the parking statistics information calculated over the predetermined period of time is used to predict availability of parking (e.g., future availability or current availability).
These and other aspects and embodiments will be apparent to those of ordinary skill in the art by reference to the following detailed description and the accompanying drawings.
In the drawing figures, which are not to scale, and where like reference numerals indicate like elements throughout the several views:
Embodiments are now discussed in more detail referring to the drawings that accompany the present application. In the accompanying drawings, like and/or corresponding elements are referred to by like reference numbers.
Various embodiments are disclosed herein; however, it is to be understood that the disclosed embodiments are merely illustrative of the disclosure that can be embodied in various forms. In addition, each of the examples given in connection with the various embodiments is intended to be illustrative, and not restrictive. Further, the figures are not necessarily to scale, and some features may be exaggerated to show details of particular components (and any size, material and similar details shown in the figures are intended to be illustrative and not restrictive). Therefore, specific structural and functional details disclosed herein are not to be interpreted as limiting, but merely as a representative basis for teaching one skilled in the art to variously employ the disclosed embodiments.
In one embodiment, the sensor 110 is an ultrasonic rangefinder and the vehicle 105 includes a Global Positioning System (GPS) (e.g., a Garmin 18-5 Hz GPS with 12 channel receiver). These are used to monitor road-side parking availability.
Referring to
In one embodiment, the vehicle 105 also includes a computing device 230 in communication with the sensor 110. The communication between the sensor 110 and the computing device 230 is represented by arrow 235. The computing device 230 may be internal to or external from the vehicle 105. For purposes of this disclosure (and as described in more detail below with respect to
In one embodiment, the server 205 determines from the sensor information 210 (e.g., GPS coordinates) if the vehicle 105 is in a slotted parking area or an unslotted parking area (step 325). If the vehicle 105 is in a slotted parking area, the server 105 translates the sensor information 210 to parking space counts (step 330). If the vehicle 105 is in an unslotted parking area, the server 105 translates the sensor information 210 to a parking space map (step 335). The server 205 then communicates the corresponding parking statistics information 225 to the vehicle 105 (step 340).
Referring to
In one embodiment, each sensor vehicle 105 carries a passenger-side facing ultrasonic rangefinder to detect the presence or absence of parked vehicles. In one embodiment, its range is equal to at least half the width of urban roads and the sampling rate is high enough to provide several samples over the length of a car at maximum city speeds. In one embodiment, the ultrasonic rangefinder is magnetically mounted to the side of the vehicle 105.
In one embodiment, the ultrasonic rangefinder is a Maxbotix WR1 waterproof rangefinder. This sensor emits sound waves every 50 ms at a frequency of 42 KHz. The sensor provides a single range reading from 12 inches to 255 inches every cycle, which corresponds to the distance to the nearest obstacle or the maximum range of 255 inches if no obstacle is detected. In one embodiment, the sensor measurements at each vehicle 105 are time-stamped and location-stamped with inputs from a 5 Hz GPS receiver, producing the following sensor information 210:
In one embodiment, vehicles transmit a collection of these measurements (a trace 408) to the server 205. The server 205 continuously aggregates and processes this sensor information using probabilistic detection algorithms. Other sensor options also exist.
In one embodiment, to obtain ground truth information (to determine what is located on the ground at a parking spot) for system evaluation purposes and to be able to analyze erroneous readings, a webcam (e.g., a Sony Playstation 3 Eye webcam) is integrated into the passenger-side sensor mount. In one embodiment, to avoid angular and shift errors with respect to the sensor, the camera is mounted just above the sensor and its orientation is aligned to the sensor. In one embodiment, the associated program captures about 20 frames per second (fps) and tags each image with a kernel time stamp. This time stamp links images to the sensor records obtained at approximately the same time. Each image then is inspected (e.g., manually) and the ground truth sensor data is entered. This can facilitate the determining of the estimated aiming of the ultrasound sensor for error checking purposes.
In one embodiment, since GPS coordinates can oscillate due to positioning errors, the tripbox implementation can include a guard distance and a guard time to avoid repeatedly triggering the same tripbox functions. The guard distance is a minimum distance that must be traveled between two tripbox boundary crossings. Similarly, the guard time is the minimum time that must be spent before the next tripbox function can be triggered. This avoids triggering the start and the stop functions repeatedly due to GPS errors.
In one embodiment, the server 205 (or, in another embodiment, the vehicle 105) then executes a detection algorithm to detect available parking spots. The detection algorithm translates the ultrasound distance-reading trace into a count of available parking spaces. The distance-reading trace provides a one-dimensional view of the distance to the nearest obstacle as the sensing vehicle 105 moves forward.
An ultrasonic sensor does not have a perfectly narrow beam-width, but instead the beam width of the sound waves emitted widens with distance. This implies that the sensor receives echos, not just from objects that are directly in front, but also from objects that are at an angle. This can affect how the sensor 110 perceives vehicles that are parked very close to each another. Instead of clearly sensing the gap between these vehicles, the “dips” in the sensor reading can become merged, as depicted in dips 620, 625 of
The inaccuracy of latitude and longitude values obtained from the GPS unit adds another challenge to the detection problem. The location estimate provided by a commercial grade GPS receiver suffers from well known errors. Without a priori knowledge of how the GPS error varies in space and time, it is possible that GPS errors can make a parked car appear to be shorter or longer than its true length. Since the detection of parked vehicles depends upon distinguishing objects that are about the length of a car from other, smaller obstacles in the sensor's path (such as trees, recycle bins, people, etc.), the detection sometimes leads to false alarms (i.e., dips caused by objects other than cars to be classified as parked cars), and missed detections (i.e., parked vehicles to be classified as something other than a parked car).
With respect to the detection algorithms, in one embodiment a slotted model exists and an unslotted model exists.
Slotted model: Each dip in the sensor trace has a depth and a width that correspond to the distance from the sensor 110 to the object causing the dip, and the size of the object in the direction of motion of the sensing vehicle 105. The sensor trace first is pre-processed to remove all dips that have too few readings (less than 6 sensor readings, assuming a maximum speed of 37 mph and a car length of 5 meters) and could not possible have arisen from a parked car. To detect a parked car, in one embodiment the width and depth of each dip in the sensor reading is compared against thresholds. These thresholds can be determined using training data.
In one embodiment, training data refers to a recording of spatial width and depths of dips produced by vehicles. In one embodiment, training data is collected using the webcam, which allows the visual determination of whether the sensor is pointed at a vehicle at any given point in time and thereby enables association of a given dip with an actual parked vehicle.
In one embodiment, all remaining dips are checked for spatial width, and compared against a threshold representing the typical length of a car. For this, the interpolated GPS coordinates belonging to the starting and ending sample of the dip are converted to UTM (meters) and the distance in meters between the starting and ending sample is computed. Since some dips correspond to multiple cars parked very close together, in one embodiment, dips of a width greater than twice the threshold for one car are classified to belong to two cars, and so on. This allows the counting of the number of cars on a stretch of road. Subtracting this from the total number of slots on the road, as given by the map, provides an estimate of the number of vacant spaces.
Unslotted model: For the unslotted parking model, the number of cars that can be accommodated on a given stretch of road depends upon the manner in which cars are parked on it at any given instant of time. Since each successive pair of parked cars in this model can have a variable amount of space between them, in one embodiment, the space between successive parked cars is estimated to determine whether the space is large enough to accommodate one or more cars. To accomplish this, in one embodiment, the sensor trace is used to estimate the spatial distance between dips that have been classified as parked cars. The estimated length of the vacant stretch then is compared against the length of a standard parking space (which, in one embodiment, is assigned a value of 6 meters).
In one embodiment, slotted and unslotted street-parking models are handled separately. Further, in one embodiment, it is assumed that it is easy to obtain information about which streets have which type of parking as prior knowledge. For the slotted model, detecting how many of the parking spaces on a road segment are vacant is of interest.
For example, it is assumed that a street segment with the slotted parking model is known to have N parking slots and that at a given instant of time, n of these slots are vacant. A sensing vehicle that drives through this street determines that n{circle around ( )} of the slots are vacant. The value of n{circle around ( )} can differ from n due to missed detections as well as false positives. In one embodiment, the missed detection rate (pm) is of interest, i.e. the probability that a parked car is not detected. Further, the false positive rate (pf) is also of interest, i.e. the probability that there is no parked car in a given slot but the detection algorithm detects one. The ratio n{circle around ( )}/n captures the performance of the detection algorithm in estimating the number of vacant spaces. This ratio can be smaller or larger than 1, for a given run, depending on whether there are a greater number of missed detections or false positives. In one embodiment, since the thresholds for dip classification are chosen from the training data to minimize the overall error rate, and this is known to occur when the probability of false alarm equals the missed detection probability, it is expected that the ratio n{circle around ( )}/n has a mean close to 1.
For the unslotted model, in one embodiment, the appropriate metric of interest is: ‘How many more cars can be accommodated on a given road segment, given the cars that are presently parked on it?’. As described above, estimating this number uses estimation of the space between parked cars. As in the slotted parking model, it is assumed that the locations of stretches where unslotted parking spaces are available is known and the detection algorithm is executed over such stretches. Whenever the detection algorithm ascertains that a space between two parked cars is large enough to accommodate another car, it records the estimated space {circumflex over (d)}. Suppose the actual space between the cars is d, then {circumflex over (d)} can be larger or smaller than d and, as before, the measure of accuracy is taken to be {circumflex over (d)}/d. Further, the miss detection rate pm, is of interest, i.e. the probability that the algorithm decides that there isn't enough space for a single car, when there actually is, and the false positive rate p f, is also of interest, i.e. the probability that the detection algorithm declares that one or more cars can be accommodated in a space between two parked cars, whereas in reality there is not enough space for a single car. In one embodiment, it is assumed that a vehicle of length 5 meters and at least half meter on either side for parking, for a minimum of 6 meters, qualifies for a parking space.
In one embodiment, to evaluate the detection algorithm, the images recorded by the webcam are utilized. Since the webcam records images at a rate of 21 frames per second, it matches the rate at which sensor readings are recorded fairly well. Each image is labeled manually based on whether the center of the image has a car in front or not. The time stamp associated with each image allows the interpolation of a location stamp for each image. This provides the ground truth for the training data set and the evaluation data set.
For the unslotted model, the estimate of space between two successive cars is compared with the true value as computed using the ground truth generated by the tagged video images.
While the counting of available parking spaces does not require high absolute position accuracy, creating an occupancy map of parking increases accuracy requirements since a detected car has to be matched to a spot on a reference map. In one embodiment, the location coordinates provided by a GPS receiver are typically accurate to 3 m (standard deviation) when the Wide Area Augmentation System (WAAS) service is available. Given a parking spot length of about 7 m, one can expect a significant rate of errors—any error greater than 3.5 m could lead to matching a vehicle to an incorrect adjacent spot.
To address the occupancy map challenge, an occupancy map creation algorithm is used that exploits both patterns in the sequence of parking spots, as well as an Environmental GPS position correction method, to improve location accuracy with respect to the parking spot map. In one embodiment, the error in GPS coordinates is studied based on how it behaves as a function of distance. The positioning accuracy of a GPS receiver is affected by several factors, including ionospheric effects, satellite orbit shifts, clock errors, and multipath. Ionospheric effects typically dominate the other error sources, except for errors that experience satellite occlusion (e.g., in urban canyons). Ionospheric effects remain similar over distances of several 10s of kilometers and they contain significant components whose rate of change is on the order of tens of minutes or longer. GPS errors therefore can be expected to be correlated in time and space. However, the Wide Area Augmentation System was designed to reduce these ionospheric and some other errors, raising the question whether the resulting GPS errors with WAAS still exhibit strong spatio-temporal correlation.
In one embodiment, the GPS error is highly correlated at short distances, and the correlation tapers off with distance. Motivated by this observation, the server (or vehicle) executes a method to improve absolute location precision by an environmental fingerprinting approach. In particular, the sensor reading is used to detect certain fixed objects that persistently appear in the ultrasound sensor traces, and utilize these to correct the error in the GPS trace. To validate the approach, it is tested on the slot-matching problem described above. It is expected that the environmental fingerprinting approach will benefit any mobile sensing application that requires precise estimates of location or distance between two points, as is the case in some of the scenarios in the sensing application.
In one embodiment, certain fixed objects (such as trees, recycle bins, the edges of street signs, etc., which also would be picked up by the sensor) are location-tagged in the video traces on a given street over multiple different runs from different days. The data is tagged with the same video tagging application developed for evaluating the detection algorithm. It was determined that the tagged coordinates for a given object from multiple runs varied significantly. Using 29 different runs and 8 objects on a street, the standard deviation of error was found to be 4.6 m in the X-direction and 5.2 meters in the Y-direction. The error due to variation in the lateral position of the sensing vehicle was not corrected, because the street chosen for this was narrow enough to allow the lateral variation to be within ±½ meter. Also this street was almost parallel to the X axis and so a larger error in the Y direction to slight variations in the sensing vehicle's lateral position was expected to be observed.
In one embodiment, the error between GPS coordinates is correlated from one object to the next.
The above investigation suggests that if the GPS error is corrected at a given point, then it is likely to remain corrected for an appreciable distance. In plot 1400 of
Fingerprinting the environment by relying on features in the sensor trace that are produced by fixed objects in the environment provides a possible means to improve location accuracy beyond that provided by GPS alone. However, fingerprinting a street requires multiple traces from that street, from which the locations of objects that are likely fixed can be determined.
Estimating the GPS error using the sensor trace involves a task comparing the reported location of the pattern (dips) produced by a series of fixed objects to the a priori known location of this pattern (as determined from multiple previous traces from the same road segment). The offset between the two gives an estimate of the error in the reported location.
For example, to detect the dips corresponding to two successive fixed objects from an experimental trace, a set of candidate dips is identified for each object from the dips that are not classified as vehicles. Each candidate set consists of dips within a radius of 20 meters of the known mean location of the fixed object (mean computed from past traces). One dip then is selected from each candidate set so that the distance between the successive selected dips best matches the known distance between the mean locations of the objects to which they correspond. The vector offset between the known locations and the reported locations of the objects is the GPS error estimate. The correction procedure is repeated with another set of objects once the vehicle travel distance has exceeded the correlation distance.
Form such objects, i=1, . . . m, the location stamps 1i(x, y) of the dips corresponding to each object is recorded. These then are subtracted from the known true location of the object ti(x, y) (assuming the centroid of the 29 locations as above), giving an estimate of the error vector ei(x, y)=ti(x, y)−1i(x, y). Next, this error vector from a given object is added to the location estimates of detected cars that are detected to be within 100 meters of this object.
Motivated by the observation of correlation between GPS error in space, the specific application of matching detected parked cars to their respective slots on a street with slotted parking has been observed. To accomplish this, the output of the algorithm for detecting cars in the slotted model (see
In one embodiment, the power source for the in-car nodes is a power inverter used to convert the 12 volt DC vehicle power supply to AC power suitable for a standard PC power supply. In another embodiment, DC to DC power supplies are installed in each car node and they are connected directly to the fuse box.
In one embodiment, moving vehicles (e.g., in a different lane than the sensing vehicle) can be distinguished from parked vehicles by the length of sensor dips. A car moving at similar speeds as the sensing vehicle, for example, generates a very long dip. In another embodiment, a sensor with a much larger range can greatly help lane detection.
In one embodiment, the server 205 displays (or causes to be displayed) a list of a few best possible (e.g., closest, most likely to remain unoccupied, price per hour) parking spaces to the vehicle 105 on the road looking for a parking space. In another embodiment, the server 205 transmits to the vehicle 105 (or causes to be displayed) a gross indication of the availability of parking spaces on the streets in an urban area. For example, the gross indication may be that 10-20% of spaces are available, 20%-50% of spaces are available, 50%-75% of spaces are available, etc. In one embodiment, the server 205 provides real time information about the level of parking space availability on nearby streets to parking garages to allow them to dynamically tune their prices for parking in time.
In one embodiment, the vehicle detection algorithm is based on sensing changes in the variance of the perceived range as a vehicle drives by parked vehicles.
In one embodiment, the vehicle detection algorithm is based on windowed variance and threshold detection.
Apart from the above windowed variance algorithm, other methods for detecting parked vehicles include, for example, windowed mean, or windowed mean in combination with windowed variance. In an algorithm that uses windowed mean in combination with windowed variance, a parked vehicle will be declared to be detected if both the windowed mean and the windowed variance are seen to fall below pre-defined thresholds set for the mean and variance respectively. Additionally, information about the width of the excursions in the windowed mean and/or windowed variance can be used in conjunction with the speed of the sensor vehicle (reported by the GPS receiver) to infer the number of vehicles parked in a contiguous block one after the other. In the example of
Memory 1704 interfaces with computer bus 1702 so as to provide information stored in memory 1704 to CPU 1712 during execution of software programs such as an operating system, application programs, device drivers, and software modules that comprise program code, and/or computer-executable process steps, incorporating functionality described herein, e.g., one or more of process flows described herein. CPU 1712 first loads computer-executable process steps from storage, e.g., memory 1704, storage medium/media 1706, removable media drive, and/or other storage device. CPU 1712 then can execute the stored process steps in order to execute the loaded computer-executable process steps. Stored data, e.g., data stored by a storage device, can be accessed by CPU 1712 during the execution of computer-executable process steps.
Persistent storage medium/media 1706 is a computer readable storage medium(s) that can be used to store software and data, e.g., an operating system and one or more application programs. Persistent storage medium/media 1706 also can be used to store device drivers, such as one or more of a digital camera driver, monitor driver, printer driver, scanner driver, or other device drivers, web pages, content files, playlists and other files. Persistent storage medium/media 1706 can further include program modules and data files used to implement one or more embodiments of the present disclosure. Persistent storage medium/media 1706 can be either remote storage or local storage in communication with the computing device.
For the purposes of this disclosure, a computer readable storage medium tangibly stores computer data, which data can include computer program code executable by a computer, in machine readable form. Computer storage media includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EPROM, EEPROM, flash memory or other solid state memory technology, CD-ROM, DVD, or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by the computer.
Those skilled in the art will recognize that the methods and systems of the present disclosure may be implemented in many manners and as such are not to be limited by the foregoing exemplary embodiments and examples. In other words, functional elements being performed by single or multiple components, in various combinations of hardware and software or firmware, and individual functions, may be distributed among software applications at either the client or server or both. In this regard, any number of the features of the different embodiments described herein may be combined into single or multiple embodiments, and alternate embodiments having fewer than, or more than, all of the features described herein are possible. Functionality may also be, in whole or in part, distributed among multiple components, in manners now known or to become known. Thus, myriad software/hardware/firmware combinations are possible in achieving the functions, features, interfaces and preferences described herein. Moreover, the scope of the present disclosure covers conventionally known manners for carrying out the described features and functions and interfaces, as well as those variations and modifications that may be made to the hardware or software or firmware components described herein as would be understood by those skilled in the art now and hereafter.
While the system and method have been described in terms of one or more embodiments, it is to be understood that the disclosure need not be limited to the disclosed embodiments. It is intended to cover various modifications and similar arrangements included within the spirit and scope of the claims, the scope of which should be accorded the broadest interpretation so as to encompass all such modifications and similar structures. The present disclosure includes any and all embodiments of the following claims.
This application is a continuation of U.S. Application Ser. No. 16/809,969 filed Mar. 5, 2020, which is a continuation of U.S. application Ser. No. 16/030,149, filed Jul. 9, 2018, now U.S. Pat. No. 10,657,815, issued May 19, 2020, which is a continuation of U.S. application Ser. No. 15/424,442, filed Feb. 3, 2017, now U.S Pat. No. 10,043,389, issued Aug. 7, 2018, which is a continuation of U.S. application Ser. No. 14/841,085, filed Aug. 31, 2015, now U.S. Pat. No. 9,564,052, issued Feb. 7, 2017, which is a continuation of U.S. application Ser. No. 13/942,274, filed Jul. 15, 2013, now U.S. Pat. No. 9,123,245, issued Sep. 1, 2015, which is a continuation of U.S. application Ser. No. 13/639,755, filed May 13, 2010, now abandoned, which, is a national stage entry of PCT/US2010/034729, filed May 13, 2010, which claims the benefit of U.S. Provisional Application Ser. No. 61/177,710 filed May 13, 2009, the disclosures of which are hereby incorporated in their entirety by reference herein.
| Number | Date | Country | |
|---|---|---|---|
| 61177710 | May 2009 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 16809969 | Mar 2020 | US |
| Child | 18099076 | US | |
| Parent | 16030149 | Jul 2018 | US |
| Child | 16809969 | US | |
| Parent | 15424442 | Feb 2017 | US |
| Child | 16030149 | US | |
| Parent | 14841085 | Aug 2015 | US |
| Child | 15424442 | US | |
| Parent | 13942274 | Jul 2013 | US |
| Child | 14841085 | US | |
| Parent | 13639755 | Jan 0001 | US |
| Child | 13942274 | US |
