Ground object data collection device, ground object data collection method, and ground object data collection computer program

Abstract

The present application relates to a terrestrial object data collection device, a terrestrial object data collection method, and a computer program for terrestrial object data collection. The terrestrial object data collection device has: a storage section that stores map information including a probability distribution indicating, for each position, a degree of reliability of the presence of a terrestrial object at the position; a reception processing section that, each time terrestrial object data indicating a terrestrial object position is received from any of one or more vehicles via a communication section capable of communicating with the vehicle, stores the received terrestrial object data in the storage section; an update section that updates the probability distribution indicating, for each position, the degree of reliability of the presence of a terrestrial object, based on the positions of terrestrial objects indicated by one or more of the received terrestrial object data; and a collection stop instruction section that, with respect to terrestrial objects whose extent of the updated probability distribution is below a predetermined threshold, transmits an instruction to stop collecting terrestrial object data to the one or more vehicles via the communication section.

Description

TECHNICAL FIELD

[0001] The present application relates to a terrestrial object data collection device that collects data of a terrestrial object presented by a map, a terrestrial object data collection method, and a computer program for terrestrial object data collection. BACKGROUND

[0002] In a high-precision map that an automatic driving system of a vehicle refers to in order to perform automatic driving control of the vehicle, it is required to accurately present information about a terrestrial object that is set on a road or around the road and is associated with travel of the vehicle. Thus, data indicating a terrestrial object is collected from a vehicle that actually travels on a road. A technique of suppressing a communication load between the vehicle and a device that collects data indicating a terrestrial object at the time of the collection is proposed (refer to International Publication No. 2018 / 180097).

[0003] In the technique disclosed in International Publication No. 2018 / 180097, a server device receives information indicating a situation of automatic driving from a first mobile body that is capable of automatic driving based on a situation around the first mobile body and a map. Then, the server device transmits a request for state information indicating a situation of a place where the first mobile body moves to a second mobile body that is capable of transmitting the state information, and suppresses transmission of the request in a case where the received situation information indicates that automatic driving is possible. SUMMARY

[0004] In order to update a map so as to be able to ensure accuracy of positions of each terrestrial object presented by the map, it is desirable to collect data indicating each terrestrial object in an amount of more than a certain amount for each terrestrial object. However, due to influences of a frequency of passage of a vehicle at each place or a collection environment, or the like, a collection speed of data indicating each terrestrial object is sometimes different for each terrestrial object. In addition, accuracy of a position of a terrestrial object included in data indicating the terrestrial object is not constant, and thus an amount of data indicating the terrestrial object required until sufficient accuracy of the position is obtained is sometimes different for each terrestrial object. Thus, it is difficult to appropriately set a timing of ending collection of data indicating each terrestrial object for each terrestrial object.

[0005] Thus, an object of the present application is to provide a terrestrial object data collection device capable of appropriately setting a timing of stopping collection of data indicating each terrestrial object for each terrestrial object.

[0006] According to one embodiment, a feature data collection device is provided. The feature data collection device includes: a storage unit storing map information regarding features associated with vehicle travel, the map information including, for each location, a probability distribution indicating the reliability of the feature's presence at the location; a reception processing unit storing the received feature data in the storage unit whenever feature data indicating the location of a feature is received from any of one or more vehicles via a communication unit capable of communicating with the vehicle; an updating unit updating the probability distribution indicating the reliability of the feature's presence at each location based on the location of the feature indicated by each of the one or more received feature data; and a collection stop instruction unit transmitting an instruction to stop collecting feature data to the one or more vehicles via the communication unit when the extent of the updated probability distribution falls below a predetermined threshold.

[0007] In the feature data collection device, the feature data preferably further includes information indicating the distance between the vehicle generating the feature data and the position of the feature represented by the feature data, and the updating unit preferably increases the contribution of the feature data to the update of the probability distribution as the distance decreases.

[0008] In addition, in the ground feature data collection device, it is preferred that the probability distribution is represented by a normal distribution, and the collection stop indication unit determines that the degree of expansion of the updated probability distribution is below a predetermined threshold when the variance value in any direction in the updated probability distribution is also below a predetermined variance threshold.

[0009] Alternatively, in the ground feature data collection device, it is preferred that the probability distribution is represented by a normal distribution, and when the reliability of the average value of the position of the collection stop indication unit in the updated probability distribution is above a predetermined reliability threshold, it is determined that the degree of expansion of the updated probability distribution is below a predetermined threshold.

[0010] According to another embodiment, a method for collecting feature data is provided. The method includes: each time feature data indicating the position of a feature associated with the travel of one or more vehicles is received from the vehicle via a communication unit capable of communicating with the vehicle, the method stores the received feature data in a storage unit; updating, for each position included in map information, a probability distribution indicating the reliability of the presence of a feature at the position based on the position of the feature indicated by each of the one or more received feature data; and transmitting, via the communication unit, an instruction to stop collecting feature data to the one or more vehicles when a spread of the updated probability distribution falls below a predetermined threshold.

[0011] Furthermore, according to another embodiment, a computer program for collecting feature data is provided. The computer program for collecting feature data includes instructions for causing a computer to execute the following steps: whenever feature data indicating the position of a feature associated with the travel of one or more vehicles is received from the vehicle via a communication unit capable of communicating with the vehicle, the computer program stores the received feature data in a storage unit; updates, for each position included in map information, a probability distribution indicating the reliability of the presence of a feature at the position based on the position of the feature indicated by each of the one or more received feature data; and, when the spread of the updated probability distribution falls below a predetermined threshold, transmits an instruction to the one or more vehicles via the communication unit to stop collecting feature data.

[0012] The feature data collection device of the present invention has an effect of being able to appropriately set the timing for stopping the collection of data representing each feature. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a schematic diagram of the structure of a feature data collection system equipped with a feature data collection device.

[0014] Figure 2 This is a schematic diagram of the vehicle structure.

[0015] Figure 3 It is a hardware structure diagram of the data acquisition device.

[0016] Figure 4 This is a hardware configuration diagram of a server as an example of a feature data collection device.

[0017] Figure 5 This is a functional block diagram of a server processor related to feature data collection and processing.

[0018] Figure 6 Schematic diagram showing the reliability distribution of the positions of features.

[0019] Figure 7 This is a flowchart of the operation of collecting and processing ground feature data. DETAILED DESCRIPTION

[0020] A feature data collection device, a feature data collection method executed by the feature data collection device, and a computer program for collecting feature data will be described below with reference to the drawings. The feature data collection device collects data representing features associated with vehicle travel (hereinafter referred to as feature data) from one or more communicable vehicles in a predetermined area.

[0021] The feature data collection device represents the location of each feature using a probability distribution (hereinafter referred to as a reliability distribution) indicating the degree of confidence, or reliability, that the feature exists at that location. For each feature, the feature data collection device updates the reliability distribution for that feature using the location of the feature represented by the received feature data. The feature data collection device then stops collecting feature data for features whose spread in the reliability distribution falls below a predetermined threshold.

[0022] Examples of the land features to be detected include various road signs, various road markings, traffic lights, and land features related to the travel of other vehicles.

[0023] Figure 1 This is a schematic diagram of a feature data collection system equipped with a feature data collection device. In this embodiment, the feature data collection system 1 includes at least one vehicle 2 and a server 3 as an example of a feature data collection device. Each vehicle 2 is connected to the server 3 via the wireless base station 5 and the communication network 4 by accessing a wireless base station 5, for example. The wireless base station 5 is connected to the communication network 4 connected to the server 3 via a gateway (not shown). In addition, Figure 1 In the figure, only one vehicle 2 is shown for simplicity, but the feature data collection system 1 may also include a plurality of vehicles 2. Figure 1 In the figure, only one wireless base station 5 is shown, but a plurality of wireless base stations 5 may be connected to the communication network 4. In addition, the server 3 may be communicably connected to a traffic information server (not shown) that manages traffic information via the communication network.

[0024] Figure 2 This is a schematic diagram of the structure of vehicle 2. Vehicle 2 includes a camera 11, a GPS receiver 12, a wireless communication terminal 13, and a data acquisition device 14. Camera 11, GPS receiver 12, wireless communication terminal 13, and data acquisition device 14 are communicatively connected via an in-vehicle network compliant with a standard such as a controller area network. Vehicle 2 may also include a navigation device (not shown) that searches for a planned travel route for vehicle 2 and guides vehicle 2 along that planned travel route.

[0025] The camera 11 is an example of a camera unit for photographing the surroundings of the vehicle 2, and has a two-dimensional detector composed of an array of photoelectric conversion elements such as CCD or C-MOS that are sensitive to visible light, and an imaging optical system that forms an image of the area to be photographed on the two-dimensional detector. Moreover, the camera 11 is installed in the interior of the vehicle 2, for example, in a manner facing the front of the vehicle 2. Moreover, the camera 11 photographs the area in front of the vehicle 2 at predetermined shooting cycles (for example, 1 / 30 second to 1 / 10 second) to generate an image of the area in front. The image obtained by the camera 11 can be either a color image or a grayscale image. In addition, a plurality of cameras 11 with different shooting directions or focal distances can also be provided in the vehicle 2.

[0026] Every time the camera 11 generates an image, it outputs the generated image to the data acquisition device 14 via the in-vehicle network.

[0027] The GPS receiver 12 receives GPS signals from GPS satellites at predetermined intervals and determines the position of the vehicle 2 based on the received GPS signals. The GPS receiver 12 then outputs positioning information indicating the positioning result of the vehicle 2 based on the GPS signals to the data acquisition device 14 at predetermined intervals via the in-vehicle network. Alternatively, the vehicle 2 may include a receiver that complies with a satellite positioning system other than the GPS receiver 12. In this case, the receiver alone may determine the position of the vehicle 2.

[0028] The wireless communication terminal 13 is an example of a communication unit and is a device that performs wireless communication processing in accordance with a predetermined wireless communication standard. For example, it accesses the wireless base station 5 and thereby connects to the server 3 via the wireless base station 5 and the communication network 4. The wireless communication terminal 13 then generates an uplink wireless signal containing feature data, etc., received from the data acquisition device 14. The wireless communication terminal 13 then transmits this uplink wireless signal to the wireless base station 5, thereby transmitting the feature data and driving information, etc., to the server 3. Furthermore, the wireless communication terminal 13 receives a downlink wireless signal from the wireless base station 5 and transmits the collection instruction or collection stop instruction, etc., from the server 3, contained in the wireless signal, to the data acquisition device 14 or the electronic control unit (ECU, not shown) that controls the driving of the vehicle 2.

[0029] Figure 3 1 is a hardware configuration diagram of the data acquisition device. The data acquisition device 14 generates feature data based on the images generated by the camera 11. Furthermore, the data acquisition device 14 generates driving information of the vehicle 2. To this end, the data acquisition device 14 includes a communication interface 21, a memory 22, and a processor 23.

[0030] The communication interface 21 is an example of an in-vehicle communication unit and includes an interface circuit for connecting the data acquisition device 14 to the in-vehicle network. Specifically, the communication interface 21 connects to the camera 11, the GPS receiver 12, and the wireless communication terminal 13 via the in-vehicle network. Whenever the communication interface 21 receives an image from the camera 11, it transmits the received image to the processor 23. Furthermore, whenever the communication interface 21 receives positioning information from the GPS receiver 12, it transmits the received positioning information to the processor 23. Furthermore, the communication interface 21 transmits to the processor 23 instructions for collecting feature data and instructions for stopping collection received from the server 3 via the wireless communication terminal 13. Furthermore, the communication interface 21 outputs the feature data received from the processor 23 to the wireless communication terminal 13 via the in-vehicle network.

[0031] The memory 22 includes, for example, volatile semiconductor memory and nonvolatile semiconductor memory. The memory 22 may also include other storage devices such as a hard disk drive. The memory 22 stores various data used in processing related to feature data generation, which is executed by the processor 23 of the data acquisition device 14. This data includes, for example, a road map, vehicle 2 identification information, camera 11 parameters such as the camera 11's installation height, shooting direction, and field of view, and a parameter set for determining an identifier for detecting features from images. The road map, for example, can be a map used in a navigation device and contains information such as the location and length of each road section within the area represented by the road map, and the connection relationship between road sections at each intersection. The memory 22 may also store images received from the camera 11 and positioning information received from the GPS receiver 12 for a certain period of time. Furthermore, the memory 22 stores information indicating an area designated by a feature data collection instruction for feature data generation and collection (hereinafter sometimes referred to as a collection target area), and information indicating an area designated by a collection stop instruction for feature data collection (hereinafter sometimes referred to as a collection stop area). Furthermore, the memory 22 may also store computer programs and the like for realizing each process executed by the processor 23 .

[0032] The processor 23 includes one or more CPUs (Central Processing Units) and their peripheral circuits. The processor 23 may also include other arithmetic circuits such as a logic operation unit, a numerical operation unit, or a graphics processing unit. The processor 23 then stores images received from the camera 11 and positioning information received from the GPS receiver 12 in the memory 22. Furthermore, the processor 23 executes processing related to generating feature data at predetermined intervals (e.g., 0.1 to 10 seconds) while the vehicle 2 is traveling.

[0033] As part of processing related to generating feature data, the processor 23 determines, for example, whether the vehicle 2's position, indicated by positioning information received from the GPS receiver 12, is within the target collection area. If the vehicle's position is within the target collection area, the processor 23 generates feature data based on the image received from the camera 11.

[0034] For example, the processor 23 detects features in the input image (hereinafter sometimes referred to as the input image) by inputting the image received from the camera 11 to a classifier pre-trained to detect features as detection targets. The processor 23 then generates information indicating the type of the detected features as feature data. For example, the processor 23 can use a deep neural network (DNN) pre-trained to detect features in the input image as this classifier. Examples of such DNNs include those with a convolutional neural network (CNN) architecture, such as the Single Shot MultiBox Detector (SSD) or Faster R-CNN. In this case, the classifier calculates a confidence level, indicating the likelihood that a feature is present in various regions of the input image for each type of feature (e.g., lane markings, crosswalks, stop signs, etc.). The classifier determines that a feature of that type is present in a region where the confidence level for any feature type exceeds a predetermined detection threshold. The recognizer then outputs information indicating the region in the input image that includes the detected feature (e.g., the bounding rectangle of the detected feature, hereinafter referred to as the object region) and information indicating the type of feature present in the object region. Therefore, the processor 23 can generate feature data that includes information indicating the type of feature present in the detected object region.

[0035] Further, the processor 23 determines the position of the aboveground object in the real space indicated by the aboveground object data, and includes information indicating the position in the aboveground object data. Here, the position of each pixel on the image corresponds one-to-one to the orientation from the camera 11 to the object indicated by the pixel. Thus, the processor 23 estimates the position of the aboveground object represented by the object region detected from the image based on the orientation from the camera 11 corresponding to the center of gravity of the object region, the own vehicle position of the vehicle 2 at the time of generation of the image used to generate the aboveground object data, the traveling direction of the vehicle 2, and the imaging direction, the field of view, and the installation height of the camera 11. At this time, the processor 23 can use the position indicated by the positioning information received from the GPS receiver 12 at the closest timing from the generation of the image used to generate the aboveground object data as the own vehicle position of the vehicle 2. Alternatively, in the case where the ECU (not shown) estimates the own vehicle position of the vehicle 2, the processor 23 can acquire information indicating the estimated own vehicle position of the vehicle 2 from the ECU via the communication interface 21. Further, the processor 23 can acquire information indicating the traveling direction of the vehicle 2 from the ECU (not shown). Alternatively, the processor 23 can estimate the position of the aboveground object indicated by the aboveground object data based on so-called Structure from Motion (SfM). In this case, the processor 23 uses optical flow between two images obtained at different timings to correspond the object regions in which the same aboveground object is represented to each other. Then, the processor 23 can estimate the position of the aboveground object by triangulation based on the positions of the vehicle 2 at the times of obtaining the two images, the traveling direction of the vehicle 2, the parameters of the camera 11, and the positions of the object regions in the respective images.

[0036] The processor 23 includes the latitude and the longitude indicating the position of the aboveground object indicated by the aboveground object data as information indicating the position of the aboveground object indicated by the aboveground object data in the aboveground object data. Further, the processor 23 refers to the road map, and determines a route section that is a road section including the position of the aboveground object indicated by the aboveground object data or the closest road section to the position. Then, the processor 23 includes the identification number of the determined route section in the aboveground object data.

[0037] The processor 23 may further include vehicle 2 identification information in the feature data. Furthermore, the processor 23 may include information used to estimate the location of features, such as camera 11 parameters and the location of features in the image. Furthermore, the processor 23 may include the location of vehicle 2 at the time the feature data was generated, its travel direction, and the GPS signal reception strength used to determine the location of vehicle 2, which are used to estimate the location of the features. Each time the processor 23 generates feature data, it outputs the generated feature data to the wireless communication terminal 13 via the communication interface 21. The feature data is thereby transmitted to the server 3. Furthermore, the processor 23 may also transmit the information used to estimate the location of features, along with the vehicle 2 identification information, to the server 3 via the wireless communication terminal 13, separately from the feature data.

[0038] According to a modified example, the processor 23 may also use the image generated by the camera 11 itself (hereinafter referred to as the entire image) or a partial image obtained by cutting out an area showing the road surface from the entire image as the feature data. In this case, the processor 23 may also include the vehicle 2's position and travel direction, as well as the camera 11 parameters, at the time the feature data was generated, in the feature data, allowing the server 3 to detect the location of the feature from the entire image or the partial image.

[0039] Furthermore, when the position of a detected feature is included in the collection stop area specified by the collection stop instruction received from the server 3 , the processor 23 does not generate feature data about the feature, regardless of whether the position of the feature is included in the collection target area.

[0040] Next, the server 3 as an example of the feature data collection device will be described.

[0041] Figure 4 This is a hardware configuration diagram of a server 3, an example of a feature data collection device. The server 3 includes a communication interface 31, a storage device 32, a memory 33, and a processor 34. The communication interface 31, storage device 32, and memory 33 are connected to the processor 34 via signal lines. The server 3 may also include input devices such as a keyboard and mouse, and a display device such as a liquid crystal display.

[0042] The communication interface 31 is an example of a communication unit and includes an interface circuit for connecting the server 3 to the communication network 4. Furthermore, the communication interface 31 is configured to communicate with the vehicle 2 via the communication network 4 and the wireless base station 5. Specifically, the communication interface 31 transmits feature data received from the vehicle 2 via the wireless base station 5 and the communication network 4 to the processor 34. Furthermore, the communication interface 31 transmits collection instructions and collection stop instructions received from the processor 34 to the vehicle 2 via the communication network 4 and the wireless base station 5.

[0043] The storage device 32 is an example of a storage unit, and includes, for example, a hard disk drive or optical recording medium and its access device. The storage device 32 stores various data and information used in the map data collection process. For example, the storage device 32 stores a map to be generated or updated, a reliability distribution of the positions of various features represented on the map, and identification information for each vehicle 2. Furthermore, the map to be generated or updated is an example of map information that includes a reliability distribution of the positions of various features. Furthermore, the storage device 32 stores feature data received from each vehicle 2. Furthermore, the storage device 32 may also store a computer program for executing the feature data collection process executed by the processor 34.

[0044] The memory 33 is another example of a storage unit, and includes, for example, a nonvolatile semiconductor memory and a volatile semiconductor memory, and temporarily stores various data generated during the execution of the feature data collection process.

[0045] The processor 34 is an example of a control unit and includes one or more CPUs (Central Processing Units) and their peripheral circuits. The processor 34 may also include other arithmetic circuits such as a logical operation unit or a numerical operation unit. Furthermore, the processor 34 performs feature data collection processing.

[0046] Figure 5 This is a functional block diagram of the processor 34 associated with feature data collection and processing. The processor 34 includes a collection instruction unit 41, a reception processing unit 42, an update unit 43, and a collection stop instruction unit 44. These components of the processor 34 are, for example, functional modules implemented by a computer program running on the processor 34. Alternatively, these components of the processor 34 may be dedicated arithmetic circuits provided in the processor 34.

[0047] The collection instruction section 41 generates a collection instruction indicating collection of overhead object data regarding overhead objects present in a collection target region, with respect to the vehicle 2. Each collection target region can be set, for example, as any of a plurality of regions set by dividing the entire region represented by a map that is a generation or update target into a lattice shape at predetermined lengths (e.g., several tens of m to several hundreds of m) apart. However, the example is not limiting, and each collection target region can be set in a manner that the area of the collection target region is narrower in a region in which the density of roads is higher. Or, each collection target region can be set in a manner that the area of the region is narrower in a region in which the density of a specific configuration (e.g., an intersection, a merging point, or a branching point) of roads is higher. Or, in addition, each collection target region can be set in a manner that each collection target region includes one road section or one intersection. For example, in the case of newly generating a map, the collection instruction section 41 sets a plurality of regions represented by the map as collection target regions respectively. Or, the collection instruction section 41 can set a region of a plurality of regions represented by a map that is an update target, which has elapsed for a predetermined period from the previous update, as a collection target region. Or, in addition, the collection instruction section 41 can set a collection target region in accordance with information designating the collection target region input via an input device. Further, in addition, the collection instruction section 41 can set a region including a site at which a work is performed as a collection target region when receiving work information indicating the site at which the work is performed from a traffic information server.

[0048] The collection instruction section 41 generates a collection instruction containing information that determines a collection target region. Then, the collection instruction section 41 transmits the generated collection instruction to the vehicle 2 via the communication interface 31.

[0049] The reception processing section 42 saves the received overhead object data in the storage 33 or the storage device 32 each time the overhead object data is received from the vehicle 2 via the wireless base station 5, the communication network 4, and the communication interface 31. In addition, the reception processing section 42 delivers the received overhead object data to the update section 43.

[0050] The update section 43 updates the reliability distribution regarding the position of the corresponding overhead object, with respect to each overhead object in the collection target region, in accordance with the position of the overhead object indicated by each of the one or more received overhead object data. Further, the update section 43 can execute the update processing described below each time the overhead object data is received, or the update section 43 can execute the update processing each time two or more overhead object data of a predetermined number are received.

[0051] In this embodiment, the reliability probability distribution for the location of a feature can be a two-dimensional normal distribution along the road surface. Alternatively, the reliability distribution can be a three-dimensional normal distribution. As described above, this reliability distribution indicates, for each location, the reliability of the feature at that location.

[0052] The updating unit 43 associates the features represented by the received feature data with any features among the features represented on the map. To do this, the updating unit 43 calculates the distance from the location with the highest reliability for the feature, i.e., the average value in the reliability distribution, to the location of the feature represented by the feature data, for each feature represented by the map within the collection target area that includes the location of the feature represented by the feature data. This distance can also be expressed as the Mahalanobis distance. The updating unit 43 then associates the feature represented by the feature data with the feature of the same type as the feature represented by the feature data, for which the calculated distance is the smallest among the features represented by the map, and which is less than a predetermined distance threshold.

[0053] The updating unit 43 performs maximum likelihood estimation on the positions of features represented by the received feature data and previously collected feature data for the features shown on the map, thereby updating the reliability distribution for the positions of the features shown on the map. The reliability distribution is thus updated so that the reliability of the positions represented by the received feature data increases.

[0054] Alternatively, the updating unit 43 may update the reliability distribution of the locations of features represented on the map and corresponding to the features indicated by the received feature data using Bayesian updating. In this case, the range of features represented on the map and corresponding to the features indicated by the received feature data, where the features are likely to be present, is pre-divided into a plurality of grid-like partitions. Then, for each partition, the reliability of the feature is set according to the reliability distribution before the update. The initial reliability value may be the same for each partition, or a higher reliability may be set for each partition where the likelihood of the feature is higher. Upon receiving feature data, the updating unit 43 updates the reliability of each partition so that the reliability of the partition including the location of the feature indicated by the received feature data is increased. Alternatively, the updating unit 43 may update the reliability of each partition so that the reliability of each partition within a predetermined range from the location of the feature indicated by the received feature data is increased. At this time, the updating unit 43 may increase the reliability increase rate for partitions closer to the location of the feature. Alternatively, the updating unit 43 may set a probability distribution for each partition, centered on the partition and having a variance-covariance matrix with a probability distribution before the update. The updating unit 43 may then calculate, for each partition, the posterior probability of the feature existing in the partition at the location of the feature represented by the received feature data based on the partition's probability distribution, and use this posterior probability as the updated reliability for the partition (i.e., the prior probability for the partition at the next update). The updating unit 43 then calculates the updated reliability distribution for the location of the feature by approximating the reliability of each partition to a normal distribution. In this case, the feature data used to update the reliability distribution can be discarded, thereby simplifying the management of the feature data and reducing the memory capacity used to store the feature data.

[0055] Alternatively, the update unit 43 may set multiple candidates for the reliability distribution of each feature's location. In this case, each candidate can be a normal distribution represented by the mean value and variance-covariance matrix of the location. Then, upon receiving feature data, the update unit 43 calculates the posterior probability of each candidate for the feature's location represented by the received feature data and uses this posterior probability as the prior probability for each candidate in the next update. The update unit 43 uses the normal distribution corresponding to the candidate with the highest prior probability as the reliability distribution for the feature's location. In this case, the update unit 43 can also simplify feature data management and reduce the memory capacity required to store feature data.

[0056] Alternatively, the update section 43 can also update the reliability distribution of each of a plurality of aboveground objects of the same kind included in the collection target region, in units of the collection target region. In this case, a mixed normal distribution including the reliability distribution of each of a plurality of aboveground objects of the same kind is defined. Then, the update section 43 applies the expectation-maximization algorithm to the position of the aboveground object indicated by the aboveground object data received with respect to the collection target region and the aboveground object indicated by the aboveground object data collected before, thereby updating the defined mixed normal distribution. Thus, the reliability distribution of each of the aboveground objects included in the updated mixed normal distribution is updated. According to this modification, even in a case where the aboveground object indicated by the received aboveground object data is difficult to correspond to, such as in a case where a plurality of aboveground objects of the same kind are in close contact within the collection target region, the update section 43 can appropriately update the reliability distribution with respect to the position of each of the aboveground objects.

[0057] The update section 43 notifies the collection stop instruction section 44 of the updated reliability distribution with respect to the aboveground object whose reliability distribution with respect to the position is updated, and stores the same in the storage device 32.

[0058] The collection stop instruction section 44 determines whether to stop collecting the aboveground object data with respect to the aboveground object whose reliability distribution with respect to the position is updated, based on the spread of the reliability distribution.

[0059] As described above, in a case where the reliability distribution with respect to the position is represented by a normal distribution, the reliability distribution is defined by the mean value of the position and the variance-covariance matrix. Then, the higher the reliability at the mean value of the position or the smaller the value of each element of the variance-covariance matrix, the smaller the spread of the reliability distribution. Then, the smaller the spread of the reliability distribution, the more accurate the position of the aboveground object is required to be. Thus, the collection stop instruction section 44 stops collecting the aboveground object data with respect to the aboveground object whose reliability distribution with respect to the position is updated, in a case where the spread of the updated reliability distribution is below a threshold value related to the spread. Specifically, the collection stop instruction section 44 determines that the spread of the reliability distribution is below the threshold value related to the spread, in a case where the maximum variance value among the variance values in each direction in the variance-covariance matrix representing the updated reliability distribution is below a predetermined variance threshold value. Then, the collection stop instruction section 44 stops collecting the aboveground object data with respect to the aboveground object. Alternatively, the collection stop instruction section 44 determines that the spread of the reliability distribution is below the threshold value related to the spread, in a case where the reliability at the mean value of the position of the updated reliability distribution is above a predetermined reliability threshold value. Then, the collection stop instruction section 44 stops collecting the aboveground object data with respect to the aboveground object. In this way, by determining the stop of the collection of the aboveground object data, the collection stop instruction section 44 can stop collecting the aboveground object data at a timing at which the position of the aboveground object is accurately found.

[0060] The collection stop instructing unit 44 designates a predetermined range centered around the average value of the position in the reliability distribution of the feature's position as a collection stop area for features determined to be for which collection of feature data should be stopped. The collection stop instructing unit 44 then creates a collection stop instruction including information indicating the collection stop area and transmits the created collection stop instruction to the vehicle 2 via the communication interface 31. This stops the collection of feature data related to features located within the collection stop area.

[0061] Figure 6 is a schematic diagram showing the reliability distribution of the location of ground objects. Figure 6 As shown, for feature 601, a reliability distribution 611 centered around the average value μ1 at that location is obtained. Similarly, for feature 602 and feature 603, a reliability distribution 612 centered around the average value μ2 at that location and a reliability distribution 613 centered around the average value μ3 at that location are obtained. Figure 6 On the upper side, for each reliability distribution, the range where the Mahalanobis distance is 1 is shown. Figure 6 On the lower side, regarding each reliability distribution, the reliability of each position along the extending direction of the road 600 is shown.

[0062] Regarding feature 601, the variance σ1 of the reliability distribution 611 in the direction of the road 600 is greater than the variance threshold Thσ. Furthermore, the reliability c1 at the average value μ1 is less than the reliability threshold Thc. Therefore, feature data collection for feature 601 continues.

[0063] On the other hand, for feature 602, the variance value σ2 of reliability distribution 612 is always below the variance threshold Thσ in all directions. Therefore, it is estimated that the position of feature 602 represented by reliability distribution 612 is sufficiently reliable. As a result, feature data collection for feature 602 is stopped.

[0064] Furthermore, for feature 603, the reliability c3 at the average value μ3 of the reliability distribution 613 is greater than the reliability threshold Thc. Therefore, it is estimated that the position of feature 603 represented by reliability distribution 613 is sufficiently reliable. As a result, feature data collection for feature 603 is stopped.

[0065] Figure 7 This is an operational flowchart of the feature data collection process in the server 3. The processor 34 of the server 3 may execute the feature data collection process in accordance with the operational flowchart shown below at predetermined intervals.

[0066] The reception processing unit 42 of the processor 34 stores the feature data received from the vehicle 2 in the memory 33 or the storage device 32 (step S101). Furthermore, the updating unit 43 of the processor 34 associates the features represented by the received feature data with any features on the map based on the locations and types of the features represented by the feature data (step S102). The updating unit 43 then updates the reliability distribution for the associated features based on the locations of the features represented by the feature data (step S103).

[0067] The collection stop instruction unit 44 of the processor 34 determines whether the degree of spread of the updated reliability distribution is below a predetermined threshold (step S104). If the degree of spread is below the predetermined threshold (step S104 - Yes), the collection stop instruction unit 44 generates a collection stop instruction to stop collecting feature data for features corresponding to the reliability distribution. The collection stop instruction unit 44 then transmits the collection stop instruction to the vehicle 2 via the communication interface 31 (step S105).

[0068] After step S105 or if the spread of the reliability distribution is greater than a predetermined threshold in step S104 (step S104 —No), the processor 34 determines to continue collecting feature data for the feature corresponding to the reliability distribution and then terminates the feature data collection process.

[0069] As described above, the feature data collection device represents the location of each feature within the collection target area using a reliability distribution. For each feature, the feature data collection device updates the reliability distribution for that feature's location using the location represented by the received feature data. The feature data collection device then stops collecting feature data for features whose spread in the reliability distribution falls below a predetermined threshold. This allows the feature data collection device to stop collecting feature data for each feature when the reliability of the feature's location becomes sufficiently high, allowing for appropriate timing to be set for stopping the collection of data representing that feature.

[0070] According to a modified example, the data acquisition device 14 may include the distance from the vehicle 2 to the position of the detected feature in the feature data. It is assumed that the shorter the distance from the vehicle 2 to the position of the detected feature, the more accurate the feature's position. Therefore, the shorter the distance from the vehicle 2 to the position of the detected feature included in the feature data, the higher the contribution of the feature's position represented by that feature data to the update of the reliability distribution for the feature's position. For example, when updating the reliability distribution using maximum likelihood estimation, the updating unit 43 uses the weighted position obtained by multiplying the position of each feature data item by a weighting coefficient for maximum likelihood estimation. In this case, the shorter the distance from the vehicle 2 to the position of the detected feature, the higher the weighting coefficient. Furthermore, when updating the reliability distribution using Bayesian updating, the updating unit 43 increases the rate of increase in the reliability of the partition containing that location as the distance from the vehicle 2 to the position of the detected feature decreases.

[0071] According to this modification, the feature data collection device can more appropriately update the reliability distribution regarding the positions of features.

[0072] Furthermore, a computer program that causes a computer to implement the functions of each component of the processor of the feature data collection device according to each of the above-described embodiments or variations may also be provided in the form of a computer-readable recording medium. The computer-readable recording medium may be, for example, a magnetic recording medium, an optical recording medium, or a semiconductor memory.

[0073] As described above, those skilled in the art can make various modifications in accordance with the embodiments within the scope of the present invention.

Claims

1. A ground feature data collection device comprising: a storage unit storing map information regarding features associated with vehicle travel, the map information including, for each position, a probability distribution indicating a reliability of the presence of the feature at the position; a reception processing unit that stores the received feature data in the storage unit whenever feature data indicating the position of the feature is received from any of the one or more vehicles via a communication unit capable of communicating with the vehicle; an updating unit that updates the probability distribution based on the position of the feature represented by each of the received one or more feature data; as well as a collection stop instruction unit that transmits an instruction to stop collecting the feature data to the one or more vehicles via the communication unit when the spread of the updated probability distribution is below a predetermined threshold value; The probability distribution is represented by a normal distribution, and the collection stop instruction unit determines that the spread of the updated probability distribution is less than a predetermined threshold when the variance value in any direction of the updated probability distribution is also less than a predetermined variance threshold.

2. A ground feature data collection device comprising: a storage unit storing map information regarding features associated with vehicle travel, the map information including, for each position, a probability distribution indicating a reliability of the presence of the feature at the position; a reception processing unit that stores the received feature data in the storage unit whenever feature data indicating the position of the feature is received from any of the one or more vehicles via a communication unit capable of communicating with the vehicle; an updating unit that updates the probability distribution based on the position of the feature represented by each of the received one or more feature data; as well as a collection stop instruction unit that transmits an instruction to stop collecting the feature data to the one or more vehicles via the communication unit when the spread of the updated probability distribution is below a predetermined threshold value; The probability distribution is represented by a normal distribution, and the collection stop instruction unit determines that the spread of the updated probability distribution is below a predetermined threshold when the reliability of the average value of the positions in the updated probability distribution is above a predetermined reliability threshold.

3. The ground feature data collection device according to claim 1 or 2, wherein: The feature data further includes information indicating a distance between a vehicle generating the feature data and a position of the feature indicated by the feature data. The updating unit increases the contribution of the feature data to the update of the probability distribution as the distance is smaller.

4. A method for collecting ground feature data, comprising: Whenever feature data indicating a position of a feature associated with the travel of the vehicle is received from any of the one or more vehicles via a communication unit capable of communicating with the vehicle, the received feature data is stored in a storage unit. updating, based on the positions of the features indicated by the received one or more feature data, a probability distribution indicating the reliability of the presence of the feature at each position included in the map information; When the spread of the updated probability distribution is equal to or less than a predetermined threshold, an instruction to stop collecting the feature data is transmitted to the one or more vehicles via the communication unit. The probability distribution is represented by a normal distribution. If the variance value in any direction of the updated probability distribution is also less than a predetermined variance threshold, it is determined that the spread of the updated probability distribution is less than the predetermined threshold.

5. A method for collecting ground feature data, comprising: Whenever feature data indicating a position of a feature associated with the travel of the vehicle is received from any of the one or more vehicles via a communication unit capable of communicating with the vehicle, the received feature data is stored in a storage unit. updating, based on the positions of the features indicated by the received one or more feature data, a probability distribution indicating the reliability of the presence of the feature at each position included in the map information; When the spread of the updated probability distribution is equal to or less than a predetermined threshold, an instruction to stop collecting the feature data is transmitted to the one or more vehicles via the communication unit. The probability distribution is represented by a normal distribution, and when the reliability at the average value of the positions in the updated probability distribution is greater than or equal to a predetermined reliability threshold, it is determined that the spread of the updated probability distribution is less than or equal to the predetermined threshold.

6. A computer program product for collecting ground feature data, configured to cause a computer to execute: Whenever feature data indicating a position of a feature associated with the travel of the vehicle is received from any of the one or more vehicles via a communication unit capable of communicating with the vehicle, the received feature data is stored in a storage unit. updating, based on the positions of the features indicated by the received one or more feature data, a probability distribution indicating the reliability of the presence of the feature at each position included in the map information; When the spread of the updated probability distribution is equal to or less than a predetermined threshold, an instruction to stop collecting the feature data is transmitted to the one or more vehicles via the communication unit. The probability distribution is represented by a normal distribution. If the variance value in any direction of the updated probability distribution is also less than a predetermined variance threshold, it is determined that the spread of the updated probability distribution is less than the predetermined threshold.

7. A computer program product for collecting ground feature data, configured to cause a computer to execute: Whenever feature data indicating a position of a feature associated with the travel of the vehicle is received from any of the one or more vehicles via a communication unit capable of communicating with the vehicle, the received feature data is stored in a storage unit. updating, based on the positions of the features indicated by the received one or more feature data, a probability distribution indicating the reliability of the presence of the feature at each position included in the map information; When the spread of the updated probability distribution is equal to or less than a predetermined threshold, an instruction to stop collecting the feature data is transmitted to the one or more vehicles via the communication unit. The probability distribution is represented by a normal distribution, and when the reliability at the average value of the positions in the updated probability distribution is greater than or equal to a predetermined reliability threshold, it is determined that the spread of the updated probability distribution is less than or equal to the predetermined threshold.

Images