Detection information correction device, detection information correction program, detection information correction system, and operation control device

Abstract

Regardless of the information detected in various environments and situations, we plan for accurate correction of the detected object information. [Solution] The system includes: an object detection information acquisition unit that acquires object detection information, which includes object state information, which is information about multiple state-related items of an object that is detected by a sensor; a type candidate determination unit that determines a candidate type of the object based on a combination of object state information corresponding to multiple state-related items; a correction target information determination unit that outputs a correction target state-related item and a correction value candidate as correction information based on a combination of the type candidate and object state information; a correction unit that corrects the object state information based on the object detection information and the correction information to generate object information, which is corrected object detection information; and an object information output unit that outputs object information.

Description

Technical Field

[0001] The present disclosure relates to a detection information correction device, a detection information correction program, a detection information correction system, and an operation control device.

Background Art

[0002] In recent years, technologies have been developed to perform various functions such as automatic driving of vehicles by using object detection information obtained by sensors such as cameras or LiDARs to detect objects. When using such sensors, for example, when the installation intervals of the sensors are sparse, or when the distance from the sensor to the object is far, the accuracy of the object detection information obtained by the sensor detecting the object is low, and the reliability of the object that can be recognized from the object detection information may be low.

[0003] In order to solve the above problems, for example, in Patent Document 1, an object included in an input image captured by a surveillance camera is detected, and the type of the object included in the input image is determined based on the probability for each type of the object obtained by inputting the movement history information of the detected object into a learned model, so that when the number of pixels related to the object included in the input image is small, a technique for quickly and accurately determining the type of the object is disclosed.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

[0005] However, the prior art disclosed in Patent Document 1 attempts to compensate for the low reliability of the input image by preparing a pre-trained model so that reliable information (probability for each object type) can be obtained even if the input image has a small number of pixels related to the objects contained in it, in other words, an input image with low reliability. However, this does not make the input image itself a source of information with sufficient reliability. Therefore, the prior art disclosed in Patent Document 1 still cannot solve the above problem. [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] This disclosure is made to solve the above-mentioned problems and aims to provide a detection information correction device that can improve the reliability of detected information by accurately correcting the information of detected objects, even when the information is detected in various environments and situations. [Means for solving the problem]

[0007] The detection information correction device according to this disclosure includes: an object detection information acquisition unit that acquires object detection information including object state information which is information relating to multiple state-related items of an object that is the target of detection by a sensor; a type candidate determination unit that determines a candidate type of the object based on a combination of object state information corresponding to multiple state-related items; a correction target information determination unit that outputs a correction target state-related item and a correction value candidate as correction information based on a combination of the type candidate and object state information; a correction unit that corrects the object state information based on the object detection information and the correction information to generate object information which is corrected object detection information; and an object information output unit that outputs object information. [Effects of the Invention]

[0008] The detection information correction device of this disclosure includes a type candidate determination unit that determines a candidate for the type of object based on the object state information included in the object detection information acquired by the object detection information acquisition unit, a correction target information determination unit that outputs the correction target state-related items to be corrected and their correction value candidates as correction information based on the object type candidate and object state information, and a correction unit that corrects the object state information based on this correction information to generate object information. As a result, the correction information can be flexibly set without fixing the object type or the correction target state-related items, and the detection information correction device can accurately correct the object state information in a variety of situations, thereby improving the accuracy of object detection. [Brief explanation of the drawing]

[0009] [Figure 1] This figure shows an example configuration of the detection information correction device according to Embodiment 1. [Figure 2] This figure shows an example of a determination table for the type candidate determination unit of the detection information correction device according to Embodiment 1. [Figure 3] This figure shows an example of a determination table for the correction target information determination unit of the detection information correction device according to Embodiment 1. [Figure 4] This figure shows an example of the configuration of the correction unit of the detection information correction device according to Embodiment 1. [Figure 5] This diagram shows the flow of information exchanged between each component in the detection information correction device according to Embodiment 1. [Figure 6] This is a flowchart illustrating the operation of the detection information correction device according to Embodiment 1. [Figure 7] This is a flowchart illustrating the details of the process in step ST3 of Figure 4 in the detection information correction device according to Embodiment 1. [Figure 8] This figure shows an example of the hardware configuration of the detection information correction device according to Embodiment 1. [Figure 9] This figure shows an example configuration of the detection information correction device according to Embodiment 2. [Figure 10]It is a diagram showing a configuration example of a detection information correction device according to Embodiment 3. [Figure 11] It is a diagram for explaining an example of the operation of a detection information correction device according to Embodiment 3. [Figure 12] It is a diagram showing a configuration example of a detection information correction device according to Embodiment 4. [Figure 13] It is a diagram for explaining an example of the operation of a detection information correction device according to Embodiment 4.

Embodiments for Carrying Out the Invention

[0010] Hereinafter, preferred embodiments of a detection information correction device according to the present disclosure will be described with reference to the drawings. Note that the same reference numerals are assigned to the same contents and corresponding parts, and detailed descriptions thereof are omitted.

[0011] Embodiment 1. <Definition of Terms> Regarding the main terms used in this embodiment, although the terms will be explained even in the scene where they first appear, the definitions will be described first in summary to facilitate the understanding of this embodiment. Object: A moving body such as a vehicle, a two-wheeled vehicle, a person, etc. Object detection information: Information about an object detected by a sensor (including an identifier capable of identifying the sensor, environmental information at the moment when the sensor detects the object, and object state information). State-related item: An item related to the state of an object (for example, the position of the object, the distance to the object, the speed of the object, the size of the object, the type of the object, the direction of the object, etc.). Object state information: Information about an object related to state-related items included in object detection information. Correction information: State-related items to be corrected (correction target state-related items) according to object type candidates and their correction value candidates. Correction target state-related item: Among the state-related items of the object state information included in the object detection information, the state-related item to be corrected. Object information: Object detection information after correction.

[0012] First reliability: The reliability of the object state information included in the object detection information. Second reliability: The reliability of the correction information. A value used to control the weighting of whether to correct the object state information. Third reliability: The reliability of the object state information included in the history of the object detection information. First reliability information: Information in which the information on the correction target state related items of the object detection information and the first reliability are associated with each other. Second reliability information: Information in which the information corresponding to the correction target state related items included in the correction information and the second reliability are associated with each other. Third reliability information: Information in which the information on the correction target state related items of the history of the object detection information and the third reliability are associated with each other.

[0013] <Configuration of the detection information correction system 1000 and the detection information correction device 300> FIG. 1 is a diagram showing a configuration example of the detection information correction system 1000 and the detection information correction device 300 according to Embodiment 1. This detection information correction device 300 is connected to the sensor 100, and the detection information correction system 1000 is constituted by the detection information correction device 300 and the sensor 100.

[0014] The sensor 100 detects an object existing in a region (hereinafter referred to as "target region") to be detected for an object. The sensor 100 is, for example, a radio wave sensor such as LiDAR (Light detection and ranging), a millimeter wave radar, or a camera. The sensor 100 may be any sensor capable of detecting an object existing in the vicinity including other types of electromagnetic waves. In FIG. 1, only one sensor 100 is illustrated, but this is merely an example. In the detection information correction system 1000, a plurality of sensors 100 may be provided. When a plurality of sensors 100 are provided, the plurality of sensors 100 may be different types of sensors 100.

[0015] Sensor 100 includes a detection unit 200. The detection unit 200 generates information about objects detected by sensor 100 (hereinafter referred to as "object detection information"). Specifically, for example, if sensor 100 is an infrared camera, the detection unit 200 takes the RGB image acquired by sensor 100 as input, performs image recognition processing, obtains the position and type of object captured in the RGB image, and generates object detection information. The information obtained when sensor 100 detects an object has a format specific to the type of sensor 100. The detection unit 200 generates object detection information as abstracted information.

[0016] The object detection information includes an identifier that can identify the sensor 100 that detected the object indicated in the object detection information, environmental information at the moment the sensor 100 detected the object, and information regarding the state of the object (hereinafter referred to as "object state information"). The identifier is, for example, an ID that is uniquely determined for each sensor 100 that detected the object (for example, "001"). The environmental information is, for example, information indicating the installation location of the sensor 100. The object state information includes information on one or more items related to the state of the object (hereinafter referred to as "state-related items"). Examples of state-related items include the position of the object, the distance to the object, the velocity of the object, the size of the object, the type of object, and the direction of the object.

[0017] The detection unit 200 outputs the generated object detection information to the detection information correction device 300. In Figure 1, the sensor 100 is shown to include the detection unit 200, but this is only one example. The detection information correction device 300 may include the detection unit 200, or a device such as a PC (Personal Computer) connected to the sensor 100 and the detection information correction device 300 outside of the sensor 100 may include the detection unit 200. When the sensor 100 and the detection unit 200 are configured as separate units, the signal from the sensor that has detected an object is input to the detection unit 200, and object detection information is generated in the detection unit 200 based on this signal.

[0018] The detection information correction device 300 is installed, for example, on a server and connected to the sensor 100 via a network. The detection information correction device 300 corrects the object detection information regarding objects detected by the sensor 100 in the target area, in other words, the object detection information output from the sensor 100, as needed.

[0019] Specifically, the detection information correction device 300 corrects, as necessary, the information of state-related items that are subject to correction (hereinafter referred to as "state-related items to be corrected") from the state-related items of the object state information included in the object detection information. More specifically, the detection information correction device 300 corrects the information of state-related items that are subject to correction, which are determined to contain values ​​with low reliability, using correction information (details will be described later) or the history of the object detection information. Candidates for which state-related items will be designated as state-related items to be corrected are provided by the correction target information determination unit 304.

[0020] The state-related items that may be subject to correction are those in which the detection unit 200 may output detection results that differ from the actual state of the object due to disturbances and measurement errors input to the sensor. Examples include the size of the object, the type of the object, the speed of movement of the object, and the direction of movement of the object. Specifically, the size of the object may be output as larger than the actual size if noise signals are detected around the object, and thus may be a state-related item subject to correction. Details of the detection information correction device 300 will be described later.

[0021] In this embodiment 1, the sensor 100 is a camera as an example. The sensor 100 is installed, for example, on the shoulder of a road, and captures an image of the area to be detected, detecting moving objects such as vehicles, motorcycles, and people on the sidewalk. In the following description, when "object" is used, it refers to "moving objects such as vehicles, motorcycles, and people." Information about the object detected by the sensor 100 is output to the detection unit 200 as image data. The detection unit 200 performs image recognition processing to determine the position of the object, the type of object, etc., and generates object detection information.

[0022] In this embodiment 1, as an example, the object state information of the object detection information for an object detected by the sensor 100 includes information on five state-related items: the object's position, the object's size, the object's direction, its movement speed, and the object's type. The detection information correction device 300 corrects the information of the state-related items to be corrected, which it has determined to contain values ​​with low reliability among the information of the five state-related items included in the object detection information obtained from the detection unit 200, using correction information or the history of the object detection information, and generates the corrected object detection information as object information. Below, as an example, an example is shown in which the state-related items to be corrected are determined and corrected according to four state-related items (object's position, object's size, object's direction, and movement speed).

[0023] The detection information correction device 300 outputs the generated object information. For example, the detection information correction device 300 outputs the generated object information to the driving control device 2000 (see Figure 1) and the navigation device (not shown) installed in the vehicle. The vehicle is assumed to be a vehicle with an autonomous driving function (hereinafter referred to as "autonomous vehicle"). Even if the driving control device 2000 of the autonomous vehicle is not equipped with surrounding recognition sensors, it can recognize an object (for example, an obstacle that hinders the driving of the autonomous vehicle) based on the object information and map information, and control the driving to overtake the area where the obstacle exists in order to avoid contact with the object related to this object information. For example, the navigation device may superimpose the obstacle based on the object information onto the map and display it on a display device (not shown) provided by the navigation device. The occupants of the autonomous vehicle can recognize the obstacle by looking at the display device.

[0024] Now, let's assume that the detection information correction device 300 outputs object detection information to the autonomous vehicle without correction. In this case, for example, let's say that sensor 100 has determined that the type of object is clearly different from the actual type, or that sensor 100 has detected the size of the object to be clearly smaller than it is actually. In that case, the driving control device 2000 in the autonomous vehicle may, based on the object detection information, attempt to pass a person who is trying to cross the road, or, in a situation where it is overtaking a parked vehicle, it may mistakenly perceive the object as being smaller than it is, leading to a dangerous situation where it could potentially collide with the object.

[0025] In contrast, in Embodiment 1, the detection information correction device 300 outputs object information that has been corrected from the object detection information. The driving control device 2000 recognizes obstacles based on the object information output from the detection information correction device 300 and map information, and can decide to stop or determine a route to pass through the area where the obstacle is located in order to pass through safely. As a result, the autonomous vehicle can continue to operate without the problems described above occurring.

[0026] As shown in Figure 1, the detection information correction device 300 further comprises an object detection information acquisition unit 301, a detection result holding unit 302, a type candidate determination unit 303, a correction target information determination unit 304, a correction unit 305, and an object information output unit 306. The object detection information acquisition unit 301 acquires object detection information from the detection unit 200. The object detection information acquisition unit 301 outputs the acquired object detection information to the correction unit 305, the type candidate determination unit 303, and the correction target information determination unit 304, and also stores it in the detection result holding unit 302. The object detection information acquisition unit 301 stores the object detection information in the detection result holding unit 302, associating it with the date and time of acquisition and identifier of the object detection information.

[0027] The detection result storage unit 302 stores the object detection information acquired by the object detection information acquisition unit 301 as a history. The detection result storage unit 302 stores the object detection information according to pre-set conditions (hereinafter referred to as "history storage conditions"). The history storage conditions include, for example, a predetermined time or a predetermined number of records set by an administrator or the like. The detection result storage unit 302 stores, for example, object detection information for a predetermined time or a predetermined number of object detection records according to the history storage conditions. In Figure 1, the detection result storage unit 302 is provided in the detection information correction device 300, but this is just one example. The detection result storage unit 302 may be provided in a location outside the detection information correction device 300 that is accessible to the detection information correction device 300.

[0028] The type candidate determination unit 303 generates state-related items to be corrected and their candidates based on the object state information included in the object detection information. As an example, the type candidate determination unit 303 maintains a determination table as shown in Figure 2. This determination table is arranged in a predetermined order (first order) for combinations of object state information corresponding to multiple state-related items. This first order could be, for example, arranged in order from the type candidates that are most likely to occur under realistic traffic conditions. The type candidate determination unit 303 uses this determination table to determine the most likely object type based on combinations of object state information corresponding to multiple state-related items.

[0029] For example, if object detection information indicates that the detected object is located on a sidewalk within the detection area, is moving slowly along the sidewalk, and is large enough to be considered a vehicle, it is not typically considered a vehicle traveling on a sidewalk. Therefore, the most likely type is a pedestrian, and the next most likely type is a vehicle. In this way, after considering realistic traffic conditions, a candidate object type with high probability is selected from a combination of multiple state-related items. The type candidate determination unit 303 outputs information about the candidate object type estimated from the state-related items of the object detection information and its probability.

[0030] The Correction Target Information Determination Unit 304 determines the state-related items to be corrected based on the object type candidate obtained from the Type Candidate Determination Unit 303 and the object state information included in the object detection information. As an example, the Correction Target Information Determination Unit 304 maintains a determination table as shown in Figure 3. This determination table is arranged according to a predetermined order (second order) for combinations of type candidates and object state information corresponding to multiple state-related items. This second order could be, for example, arranged in order from the corrected state-related items and correction value candidates that are most likely to be corrected in realistic traffic conditions. The Correction Target Information Determination Unit determines the state-related items that are likely to be erroneous (corrected state-related items) based on the combination of type candidates selected by the Type Candidate Determination Unit 303 and object state information corresponding to multiple state-related items.

[0031] Specifically, if the object detection information indicates that the detected object is located on a sidewalk within the detection area, is moving slowly along the sidewalk, and is large enough to be considered a vehicle, then, according to the output of the object type candidate determination unit 303, if the object is determined to be a pedestrian, the object size in the state-related items is considered to be incorrect. If the object is determined to be a vehicle, then the location and direction of movement in the state-related items are considered to be incorrect. In this way, taking into account realistic traffic conditions, the state-related items considered to be incorrect for the selected object type candidate are selected as targets for correction. As described above, the correction target information determination unit 304 outputs the state-related items to be corrected (correction target state-related items) and their correction value candidates as correction information, according to the object type candidate.

[0032] The correction unit 305 corrects the state-related item information included in the object detection information based on the object detection information acquired by the object detection information acquisition unit 301, the correction information generated by the correction target information determination unit 304, and the history of object detection information acquired by the object detection information acquisition unit 301, in other words, the history of object detection information stored in the detection result holding unit 302, in order to generate object information. In this embodiment, the correction unit 305 corrects the object state information of the state-related items included in the object detection information based on the object detection information, the correction information, and the history of object detection information, in order to generate object information.

[0033] <Confidence Calculation Process> Here, Figure 4 shows a detailed configuration example of the correction unit 305 included in the detection information correction device 300 according to Embodiment 1. The correction unit 305 includes a reliability calculation unit 3051, a reliability comparison unit 3052, and an object information generation unit 3053.

[0034] The reliability calculation unit 3051 calculates the reliability of the object state information (hereinafter referred to as "first reliability") included in the object detection information acquired by the object detection information acquisition unit 301, the reliability of the correction information (hereinafter referred to as "second reliability"), and the reliability of the state-related item information included in the history of the object detection information (hereinafter referred to as "third reliability"). The first reliability, second reliability, and third reliability are expressed as values ​​ranging from "0.0" to "1.0", for example.

[0035] More specifically, the reliability calculation unit 3051 calculates a first reliability of the information on the correction target state-related items included in the object detection information, a second reliability of the information on the correction target state-related items indicated in the correction information, and a third reliability of the information on the correction target state-related items included in the history of the object detection information. In the example of Embodiment 1, the reliability calculation unit 3051 calculates a first reliability of the information on the direction of the object included in the object detection information, in other words, the direction of the object detected by the sensor 100, a second reliability of the information for correcting the direction of the object, and a third reliability of the history of the information on the direction of the object included in the object detection information, in other words, the direction of the object detected by the sensor 100. In Embodiment 1, the process performed by the reliability calculation unit 3051 to calculate the first reliability, second reliability, and third reliability is also called the "reliability calculation process".

[0036] The confidence calculation unit 3051 calculates higher values ​​for the first confidence level, second confidence level, or third confidence level, respectively, the closer the value indicated by the information related to the correction target state of the object detection information, the information corresponding to the information related to the correction target state of the correction information in the correction information, or the information related to the correction target state of the object detection information history is estimated to be to the true value. The confidence calculation unit 3051 only needs to obtain the history of object detection information from the detection result storage unit 302. In doing so, the confidence calculation unit 3051 compares the identifier attached to the object detection information obtained by the object detection information acquisition unit 301 with the identifier attached to the object detection information stored in the detection result storage unit 302, and obtains the matching object detection information from the detection result storage unit 302.

[0037] Here, we will explain how the reliability calculation unit 3051 calculates the first reliability, second reliability, and third reliability. First, we will explain an example of how the reliability calculation unit 3051 calculates the second reliability. In this embodiment 1, the value that the reliability calculation unit 3051 calculates for the second reliability is predetermined. For example, the second reliability is associated with "0.8" in the determination table provided by the correction target information determination unit 304, which determines that the state-related item is a state-related item to be corrected. The above is an example of how the second reliability is calculated, and the reliability value may be set to a different value for each object type and state-related item, depending on the characteristics of the sensor 100 and the detection unit 200. Furthermore, the value of the second reliability may be adjusted according to the reliability of the state-related item of the input object detection information (corresponding to the first reliability).

[0038] As described above, the confidence calculation unit 3051 calculates the second confidence level to a value predetermined by the developer or user of the detection information correction device 300. This second confidence level is determined to be a large value, in other words, a value that is estimated to be close to the true value, depending on the likelihood.

[0039] Next, an example of how the reliability calculation unit 3051 calculates the first reliability will be explained. The reliability calculation unit 3051 also calculates the first reliability so that the more likely the information regarding the corrected state-related items is, the larger the value. For example, it is assumed that the closer the distance to the object being detected, the more accurately the sensor 100 can capture the characteristics of that object. In other words, the closer the distance from the sensor 100 to the object being detected, the closer the characteristics of the object detected by the sensor 100 are assumed to be to the true value. This is because the further the distance from the sensor 100 to the object, the more likely the detected information is to contain errors or noise. Therefore, the reliability calculation unit 3051 calculates the first reliability so that, for example, the closer the distance between the sensor 100 and the object, the larger the value of the first reliability. Furthermore, since the object detection information includes object state information, which includes information about the object's position, and environmental information, which indicates the installation location of the sensor 100, the reliability calculation unit 3051 can calculate the distance between the sensor 100 and the object based on the object detection information.

[0040] The conditions under which the reliability calculation unit 3051 calculates the first reliability value based on the distance between the sensor 100 and the object are determined in advance by the developer or user of the detection information correction device 300 and stored in the buffer of the reliability calculation unit 3051. The developer or user of the detection information correction device 300 determines the conditions for calculating the first reliability value while considering the conditions for determining the second reliability value.

[0041] In detail, the developer or user of the detection information correction device 300 determines the conditions for calculating the first confidence level such that, when it is assumed that the information of the object's direction, which is a state-related item to be corrected included in the object detection information, should be corrected, the information of the object's direction will be corrected using the correction information. Specifically, the developer or user of the detection information correction device 300 determines the conditions for calculating the first confidence level such that, when it is assumed that the information of the state-related item to be corrected included in the object detection information should be corrected, the first confidence level is calculated to be smaller than the second confidence level. When it is assumed that the information of the state-related item to be corrected included in the object detection information should be corrected, for example, when the distance from the sensor 100 to the object is far (for example, farther than a preset distance).

[0042] Therefore, in Embodiment 1, the second confidence level is not a value that indicates whether the information corresponding to the information of the state-related items to be corrected in the correction information is reliable, but rather a value used to control the weighting of whether or not to perform correction on the information of the state-related items to be corrected included in the object detection information.

[0043] Next, an example of how the confidence calculation unit 3051 calculates the third confidence level will be explained. For example, if the confidence level (first confidence level) associated with the information of the correction target state-related item included in the history of object detection information is equal to or greater than a preset threshold (hereinafter referred to as the "first confidence level determination threshold"), the confidence calculation unit 3051 calculates a higher third confidence level for the information of the correction target state-related item in the history of object detection information to which the confidence level equal to or greater than the first confidence level determination threshold is associated.

[0044] Furthermore, when the reliability calculation unit 3051 calculates the first reliability, it stores the calculated first reliability in association with the information of the correction target state-related item corresponding to the first reliability in the object detection information stored in the detection result holding unit 302. The reliability calculation unit 3051 can identify the information of the correction target state-related item corresponding to the first reliability based on the identifier assigned to the object detection information and the date and time the object detection information was acquired.

[0045] For example, the confidence calculation unit 3051 extracts the largest value from the first confidence levels associated with the information of the correction target state-related items contained in the multiple object detection information included in the history. Then, the confidence calculation unit 3051 compares the extracted first confidence level with the first confidence level determination threshold to calculate the third confidence level. For example, the confidence calculation unit 3051 may calculate the third confidence level by comparing the average value of the first confidence levels associated with the information of the correction target state-related items contained in the multiple object detection information included in the history with the first confidence level determination threshold.

[0046] Furthermore, if there are multiple status-related items to be corrected, the reliability calculation unit 3051 calculates a first reliability and a third reliability for each piece of information related to the status-related items to be corrected. In addition, the reliability calculation unit 3051 calculates a second reliability for each piece of information corresponding to the status-related items to be corrected that is included in the correction information.

[0047] The reliability calculation unit 3051 outputs information regarding the first reliability level (hereinafter referred to as "first reliability information"), information regarding the second reliability level (hereinafter referred to as "second reliability information"), and information regarding the third reliability level (hereinafter referred to as "third reliability information") to the reliability comparison unit 3052. The first reliability information is information that associates the information of the correction target state-related items of the object detection information with the first reliability level. If there are multiple correction target state-related items, the first reliability level corresponding to the information of the correction target state-related item is associated with each correction target state-related item.

[0048] The second confidence information is information that associates the second confidence level with the information corresponding to the correction target state-related item in the correction information. If there are multiple pieces of information corresponding to the correction target state-related item, the corresponding second confidence level is associated with each piece of information corresponding to the correction target state-related item. The third confidence information is information that associates the third confidence level with the information of the correction target state-related item in the object detection information history. The confidence calculation unit 3051 outputs the object detection information output from the object detection information acquisition unit 301, the correction information acquired from the correction target information determination unit 304, and the object detection information history acquired from the detection result holding unit 302, along with the first confidence information, the second confidence information, and the third confidence information, to the confidence comparison unit 3052.

[0049] <Confidence comparison process> The reliability comparison unit 3052 compares the first reliability, second reliability, and third reliability calculated by the reliability calculation unit 3051 to determine whether or not the state-related items included in the object detection information acquired by the object detection information acquisition unit 301, more specifically the information of the state-related items to be corrected, should be corrected with the correction information or the state-related items included in the history of the object detection information, more specifically the information of the state-related items to be corrected.

[0050] In Embodiment 1, the process performed by the reliability comparison unit 3052, which compares the first reliability, the second reliability, and the third reliability to determine whether or not the state-related item information included in the object detection information should be corrected with the correction information or the state-related item information included in the history of the object detection information, is also called the "reliability comparison process."

[0051] For example, the reliability comparison unit 3052 compares the first reliability, second reliability, and third reliability, and if the second or third reliability is higher than the first reliability, it determines that the information of the correction target state-related items included in the object detection information should be corrected. The reliability comparison unit 3052 selects the information that was the source of the highest reliability among the first, second, and third reliability. For example, if the second reliability was the highest, the reliability comparison unit 3052 selects the information in the correction information that corresponds to the information of the correction target state-related items that was the source of the calculation of that second reliability.

[0052] In Embodiment 1, if the second confidence level of the object type candidate is the highest, the confidence comparison unit 3052 selects the object type and state-related item values ​​provided from the correction information as the object type candidate and its corresponding state-related item if the second confidence levels of all of the one or more correction target state-related items corresponding to the object type candidate in the correction information are higher than the first and third confidence levels of each item.

[0053] For example, the reliability comparison unit 3052 determines that the information of the state-related items to be corrected included in the object detection information should be corrected if the pre-set conditions (hereinafter referred to as "conditions for determining whether correction is necessary") are met. Here, the reliability comparison unit 3052 determines that, if the conditions for determining whether correction is necessary are met, the object type and state-related items provided in the correction information should be corrected among the state-related items included in the object detection information.

[0054] The conditions for determining whether correction is necessary include, for example, "the first confidence level is less than a preset threshold (hereinafter referred to as the "confidence level determination threshold"), and the second or third confidence level is equal to or greater than the confidence level determination threshold." The conditions for determining whether correction is necessary are set in advance by the developer or user of the detection information correction device 300 and stored in a location accessible to the detection information correction device 300.

[0055] The confidence comparison unit 3052 selects the information that was used to calculate the confidence level (second confidence level or third confidence level) which is equal to or greater than the confidence level determination threshold. For example, in Embodiment 1, if the most likely object type presented by the correction information is a pedestrian, and only the second confidence level of the object size, which is a correction target state related item corresponding to a pedestrian, is equal to or greater than the confidence level determination threshold, the confidence comparison unit 3052 selects the information used to correct the object size, which was used to calculate the second confidence level, from the correction information.

[0056] The reliability comparison unit 3052 associates the selected information with the object type and the correction target state-related items, and outputs it to the object information generation unit 3053 as a reliability comparison result. If the reliability comparison unit 3052 determines that correction of the correction target state-related items is unnecessary, it outputs information indicating this to the object information generation unit 3053 as a reliability comparison result.

[0057] <Object information generation> The object information generation unit 3053 generates object information by correcting the state-related items, more specifically the state-related items to be corrected, included in the object detection information acquired by the object detection information acquisition unit 301, based on the determination result by the reliability comparison unit 3052 of whether or not to correct them with the state-related items, more specifically the state-related items to be corrected, included in the object detection information acquired by the object detection information acquisition unit 301, using the correction information or the history of the object detection information.

[0058] Specifically, the object information generation unit 3053 performs a correction on the information of the correction target state-related items included in the object detection information acquired by the object detection information acquisition unit 301. If the reliability comparison unit 3052 outputs information selected by the reliability comparison unit 3052, the generation unit 3053 overwrites this information with the information output by the reliability comparison unit 3052.

[0059] In this embodiment 1, the object information generation unit 3053 performs a correction by rewriting the object type and size information included in the object detection information acquired by the object detection information acquisition unit 301 with the information provided by the correction information, for example, when the reliability comparison unit 3052 selects the object type and size information presented in the second reliability level. The object information generation unit 3053 performs a correction by rewriting the object size information included in the object detection information acquired by the object detection information acquisition unit 301 with the object size information in the object detection information history, for example, when the reliability comparison unit 3052 determines that the object size information presented in the third reliability level is the most reliable. The object information generation unit 3053 uses the corrected object detection information as object information.

[0060] In the above explanation, three confidence levels—first confidence level, second confidence level, and third confidence level—were calculated, and by comparing these values, it was determined whether or not the object state information contained in the object detection information should be corrected with correction information or the object state information contained in the history of the object detection information. If the result indicated that correction was necessary, object information was generated based on the object state information contained in the correction information or the history of the object detection information. However, it is not essential to use the third confidence level and the object state information contained in the history of the object detection information. Only the first and second confidence levels may be calculated, and by comparing these values, it may be determined whether or not the object state information contained in the object detection information should be corrected with correction information. If the result indicated that correction was necessary, object information may be generated based on the object state information contained in the correction information.

[0061] Furthermore, when the object information generation unit 3053 rewrites the information of the state-related items to be corrected in the object detection information, it may perform some kind of calculation process, such as rewriting the information corresponding to the state-related items to be corrected in the correction information, or the information corresponding to the state-related items to be corrected in the history of object detection information, with an intermediate value between the two, rather than rewriting it with the exact same value. For example, the calculation process of an intermediate value between the information corresponding to the state-related items to be corrected in the object detection information and the information corresponding to the state-related items to be corrected in the correction information may be performed by the correction target information determination unit 304 when generating the correction information. The correction target information determination unit 304 performs the above calculation process to calculate the intermediate value, etc., and generates correction information including the calculated intermediate value, etc. The object information generation unit 3053 can also generate object information by selecting whether to rewrite the object state information with correction information, rewrite it with the object state information included in the history of object detection information, or not rewrite it at all, for each of the multiple types of state-related items.

[0062] The object information output unit 306 outputs the object information generated by the object information generation unit 3053 to an external device. The external device to which the object information output unit 306 outputs the object information is, as described above, for example, a driving control device 2000 or a navigation device installed in an autonomous vehicle. The external device may also be, for example, a display device provided by a management device (not shown) such as a PC that monitors the target area in a control room or the like.

[0063] <Operation of the detection information correction device> The operation of the detection information correction device 300 according to Embodiment 1 will now be described. Figure 5 is a diagram showing the flow of information exchanged between each component in the detection information correction device 300 according to Embodiment 1. For convenience, the illustration of the object detection information, correction information, and history of object detection information output from the reliability calculation unit 3051 to the reliability comparison unit 3052, and the history of object detection information, correction information, and object detection information output from the reliability comparison unit 3052 to the object information generation unit 3053, is omitted in the correction unit 305.

[0064] Figure 6 is a flowchart illustrating the operation of the detection information correction device 300 according to Embodiment 1. The object detection information acquisition unit 301 acquires object detection information from the detection unit 200 (step ST1). The object detection information acquisition unit 301 outputs the object detection information acquired in step ST1 to the correction unit 305 and stores it in the detection result holding unit 302.

[0065] The object type candidate determination unit 303 acquires object detection information and determines an object type candidate. The correction target information determination unit 304 acquires the object detection information and the object type candidate determined by the object type candidate determination unit 303 and generates correction information (step ST2). The correction target information determination unit 304 outputs the generated correction information to the correction unit 305.

[0066] The correction unit 305 corrects the state-related items information included in the object detection information based on the object detection information acquired by the object detection information acquisition unit 301 in step ST1, the correction information acquired by the correction target information determination unit 304 in step ST2, and the history of object detection information acquired by the object detection information acquisition unit 301, in other words, the history of object detection information stored in the detection result holding unit 302, to generate object information (step ST3). The object information output unit 306 outputs the object information generated by the correction unit 305 in step ST3 (step ST4).

[0067] Figure 7 is a flowchart illustrating the details of the process in step ST3 of Figure 6. The confidence calculation unit 3051 performs a confidence calculation process to calculate the first confidence level, the second confidence level, and the third confidence level (step ST11). The confidence calculation unit 3051 outputs the first confidence level information, the second confidence level information, and the third confidence level information to the confidence comparison unit 3052. In addition, the confidence calculation unit 3051 outputs to the confidence comparison unit 3052 the object detection information output from the object detection information acquisition unit 301, the correction information acquired from the correction target information determination unit 304, and the history of object detection information acquired from the detection result holding unit 302, along with the first confidence level information, the second confidence level information, and the third confidence level information.

[0068] The reliability comparison unit 3052 compares the first reliability, second reliability, and third reliability calculated by the reliability calculation unit 3051 in step ST11 and performs a reliability comparison process to determine whether the state-related items included in the object detection information acquired by the object detection information acquisition unit 301, more specifically the information of the state-related items to be corrected, should be corrected with the correction information or the information of the state-related items included in the history of the object detection information (step ST12).

[0069] The reliability comparison unit 3052 associates the information selected in the reliability comparison process with information that can identify which correction target state-related item in the object detection information corresponds to, and outputs it to the object information generation unit 3053 as a reliability comparison result.

[0070] The object information generation unit 3053 generates object information (step ST13) by overwriting the information of the state-related items included in the object detection information with the correction information or the information of the state-related items included in the history of object detection information, based on the determination result by the reliability comparison unit 3052 of whether or not the object state information included in the object detection information acquisition unit 301 should be corrected with correction information or state-related items included in the history of object detection information. The object information generation unit 3053 outputs the generated object information to the object information output unit 306. As described above, the object information generation unit 3053 can also generate object information by selecting whether to rewrite the object state information with correction information, rewrite it with object state information included in the history of object detection information, or not rewrite it for each of the multiple types of state-related items.

[0071] <An example of a specific application scenario for the detection information correction device> The operation of the detection information correction device 300, as described using the flowchart in Figure 6, will now be explained with an example of a specific application scenario. Figure 2 is a diagram illustrating an example of the operation of the detection information correction device 300 when it is applied in the above scenario, as described using the flowchart in Figure 6.

[0072] Figure 2 is a determination table that associates the most likely object types with the state-related items that are likely to be incorrect, i.e., the state-related items that should be corrected, based on the combination of state-related items obtained from object detection information. The type candidate determination unit 303 obtains object detection information from the object detection information acquisition unit 301 and extracts one or more likely type candidates by comparing the state-related items of the acquired object detection information with the above determination table.

[0073] The extracted type candidates are output to the Correction Target Information Determination Unit 304, which, based on the object detection information and type candidates, extracts state-related items to be corrected for each type candidate. The Correction Target Information Determination Unit 304 is a processing unit that generates the Correction Target State-Related Items 1, 2, and 3 in Figure 2. The Correction Target Information Determination Unit 304 determines the Correction Target State-Related Items for the object type candidates obtained from the Type Candidate Determination Unit 303 based on the properties that the object type candidate normally possesses when traffic conditions are taken into consideration.

[0074] For example, according to the state-related items of the object detection information, if a small object detected on a sidewalk is moving at high speed across the sidewalk, and this object is considered a pedestrian, then the speed information of the object detection information is incorrect and should be corrected with other information. Alternatively, if this object is considered a motorcycle, then it is unlikely that a motorcycle would cross a sidewalk, and it is more likely that it is actually moving along the roadway, so the object's position and direction of movement should be corrected. Or, if this object is considered a vehicle, then, similar to a motorcycle, it is actually moving along the roadway, and since a vehicle is not a small object, the object's position, direction of movement, and size should be corrected. The correction target information corresponding to the candidate object type and candidate different types, generated in this way, is transmitted to the correction unit 305 as correction information.

[0075] The correction unit 305 selects one of the object type candidates provided as correction information. The reliability calculation unit 3051 pre-calculates or defines the first reliability, second reliability, and third reliability. In the reliability comparison unit 3052, if the detected object is assumed to be a pedestrian, the correction target state-related item presented in the correction information is speed information. Here, the first reliability, second reliability, and third reliability of the speed information are compared, and if the first reliability or third reliability is greater than the second reliability, it is determined that the object does not require speed information correction, and the possibility that this object is a pedestrian is excluded.

[0076] Next, assuming the detected object is a motorcycle, the first, second, and third confidence levels for position and direction are compared. If the second confidence level is the highest for all information, the object information generation unit 3053 selects a motorcycle as the object type, overwrites the object type, position, and direction information of the object detection information, and uses the corrected object detection information as the object information.

[0077] Regarding position and direction information, standard values ​​for each object type may be defined, for example, a motorcycle being 1.5m from the edge of the road and its direction being along the road, and these may be provided as correction information. Alternatively, information other than object detection information, specifically the history of object detection information, may be selected. If the comparison result of the first confidence level, second confidence level, and third confidence level in the correction unit 305 does not apply to any of the object type candidates provided in the correction information, no correction using the correction information is performed.

[0078] Thus, the detection information correction device 300 according to Embodiment 1 corrects the state-related item information included in the object detection information based on the acquired object detection information, correction information for correcting state-related item information, and the history of the object detection information, and generates object information, which is corrected object detection information. The correction information can be applied without fixing the detected object or the state-related item to be corrected, and it is expected that it can accurately correct information such as the object type and improve detection accuracy even in a variety of situations.

[0079] In the above embodiment 1, the detection information correction device 300 stored a history of object detection information acquired from the detection unit 200 and used this history to correct the object detection information acquired from the detection unit 200. However, this is merely one example, and the detection information correction device 300 does not necessarily need to use the history of object detection information when correcting the object detection information acquired from the detection unit 200. For example, if there is no history of object detection information, the detection information correction device 300 can correct the object detection information acquired from the detection unit 200 without using the history, and the detection information correction device 300 may be configured not to have a function to store the history of object detection information. In this case, the detection information correction device 300 does not necessarily need to include a detection result holding unit 302.

[0080] In other words, the detection information correction device 300 may be configured to acquire object detection information, and based on the acquired object detection information and correction information for correcting state-related item information, correct the state-related item information included in the object detection information to generate object information, which is corrected object detection information. Even with this configuration, the detection information correction device 300 can correct the object detection information obtained by the sensor 100 detecting an object to information that has sufficient reliability to be used in technologies that perform various functions.

[0081] Furthermore, in the above embodiment 1, it was assumed that the detection information correction device 300 was mounted on a server (not shown), but this is merely one example. The detection information correction device 300 may, for example, be mounted on the sensor 100, or the detection unit 200 and the detection information correction device 300 may be mounted on a common device. The detection unit 200 may be provided in the detection information correction device 300.

[0082] Figures 8A and 8B show an example of the hardware configuration of the detection information correction device 300 according to Embodiment 1. In this Embodiment 1, the functions of the object detection information acquisition unit 301, the type candidate determination unit 303, the correction target information determination unit 304, the correction unit 305, and the object information output unit 306 are realized by the processing circuit 1001. That is, the detection information correction device 300 includes a processing circuit 1001 for performing control to correct object detection information related to an object detected by the sensor 100. The processing circuit 1001 may be dedicated hardware as shown in Figure 8A, or it may be a processor 1004 that executes a program stored in memory as shown in Figure 8B.

[0083] If the processing circuit 1001 is dedicated hardware, it may be, for example, a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, an ASIC (Application Specific Integrated Circuit), an FPGA (Field-Programmable Gate Array), or a combination thereof.

[0084] When the processing circuit is a processor 1004, the functions of the object detection information acquisition unit 301, the type candidate determination unit 303, the correction target information determination unit 304, the correction unit 305, and the object information output unit 306 are realized by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in memory 1005.

[0085] The processor 1004 reads and executes the program stored in the memory 1005, thereby executing the functions of the object detection information acquisition unit 301, the type candidate determination unit 303, the correction target information determination unit 304, the correction unit 305, and the object information output unit 306. In other words, the detection information correction device 300 includes a memory 1005 for storing a program that, when executed by the processor 1004, will result in the execution of steps ST1 to ST4 in Figure 6 described above.

[0086] Furthermore, the program stored in memory 1005 can be said to cause the computer to execute the processing procedures or methods of the object detection information acquisition unit 301, the type candidate determination unit 303, the correction target information determination unit 304, the correction unit 305, and the object information output unit 306. Here, memory 1005 refers to non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), flash memory, EPROM (Erasable Programmable Read Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), or magnetic disks, flexible disks, optical disks, compact disks, minidiscs, DVDs (Digital Versatile Discs), etc.

[0087] Furthermore, the functions of the object detection information acquisition unit 301, the type candidate determination unit 303, the correction target information determination unit 304, the correction unit 305, and the object information output unit 306 may be partially implemented by dedicated hardware and partially implemented by software or firmware. For example, the functions of the object detection information acquisition unit 301, the type candidate determination unit 303, and the correction target information determination unit 304 can be implemented by a processing circuit 1001 as dedicated hardware, while the functions of the correction unit 305 and the object information output unit 306 can be implemented by the processor 1004 reading and executing a program stored in memory 1005.

[0088] Furthermore, the detection result holding unit 302 uses, for example, memory 1005. This is just one example, and the detection result holding unit 302 may be configured with an HDD (Hard Disk Drive) or SSD (Solid State Drive), etc. The detection information correction device 300 includes a device such as a sensor 100, and an input interface device 1002 and an output interface device 1003 that perform wired or wireless communication.

[0089] As described above, the detection information correction device 300 according to this embodiment includes a type candidate determination unit 303 that determines a candidate object type based on object state information included in the object detection information acquired by the object detection information acquisition unit 301, a correction target information determination unit 304 that outputs a correction target state-related item to be corrected and a candidate correction value from among the state-related items as correction information based on the candidate object type and object state information, and a correction unit 305 that corrects the object state information based on this correction information to generate object information. For this reason, the correction information can be flexibly set without fixing the object type or the correction target state-related item, and the detection information correction device 300 can accurately correct the object state information even in a variety of situations, thereby improving the accuracy of object detection.

[0090] Embodiment 2. Figure 9 shows an example of the configuration of the detection information correction device according to Embodiment 2. In this embodiment, the type candidate determination unit 303 and the correction target information determination unit 304 may use ambient information obtained from sources other than sensors, in addition to object detection information. For example, signal information can be used as ambient information, and this signal information is transmitted from the ambient information transmitter 101 and input to the type candidate determination unit 303.

[0091] An example of the operation of the detection information correction device 300 in Embodiment 2 will be explained using Figure 3. As ambient information, the type candidate determination unit 303 acquires signal information within the detection target area from the ambient information transmitter 101. If the pedestrian signal is green, the vehicle is considered to be normally stopped and is excluded from the object type candidates. Also, if the vehicle signal is green, it is considered that it is not normally conceivable that there are pedestrians perpendicular to the roadway, i.e., crossing the roadway, and the pedestrian object type is excluded. On the other hand, if the pedestrian signal is green, the object moving in the direction of crossing the roadway is most likely to be a pedestrian, and the object type candidates can be narrowed down. The determination table shown in Figure 3 can be reinterpreted as being arranged from the most likely combinations of multiple state-related items, taking into account realistic traffic conditions including ambient information such as signal information. As described above, since the object type candidates are selected according to a predetermined order (third order), the object detection information can be corrected according to traffic conditions, and the reliability can be improved.

[0092] Embodiment 3. Figure 10 shows an example of the configuration of the detection information correction device according to Embodiment 3. In this embodiment, the correction priority determination unit 307 may assign a priority to a plurality of object type candidates in the correction information provided by the correction target information determination unit 304.

[0093] Figure 11 is a diagram illustrating an example of the operation of the detection information correction device 300 in Embodiment 3. As ambient information, the type candidate determination unit 303 acquires time period information within the detection target area from the ambient information transmitter 101. In the determination by the type candidate determination unit 303, even if the conditions indicate a high probability of it being a pedestrian based on the state-related items of the object detection information, the correction priority determination unit 307 can adjust the priority of pedestrians to be lower if the time period is late at night. In this way, by incorporating ambient information and using it as a basis, the correction priority determination unit 307 makes a decision to prioritize the information to be adopted, thereby enabling object detection information to be prioritized and corrected more appropriately according to the ambient information, and further improving reliability.

[0094] Embodiment 4. Figure 12 shows an example of the configuration of the detection information correction device according to Embodiment 4. In this embodiment, in the correction information provided by the correction target information determination unit 304, the correction priority determination unit 307 may assign a priority to multiple object type candidates, taking into consideration object detection information detected by the detection unit 200 in addition to the main detection target.

[0095] Figure 13 is a diagram illustrating an example of the operation of the detection information correction device 300 according to Embodiment 4. If the detection information correction device 300 also detects a bus, it is assumed that people will get off the bus, and even if the object is detected on the roadway, the correction priority determination unit 307 can increase the priority of pedestrians and determine the candidate object type. By making a decision in the correction priority determination unit 307 based on the object detection information of other detected objects in this way, the object detection information can be prioritized and corrected more appropriately, thereby further improving reliability.

[0096] While this disclosure describes various exemplary embodiments and examples, the various features, aspects, and functions described in one or more embodiments are not limited to the application of a particular embodiment, but can be applied individually or in various combinations to the embodiments. Accordingly, countless variations not illustrated are conceivable within the scope of the art disclosed in this specification. These include, for example, modifying, adding or omitting at least one component, or even extracting at least one component and combining it with components of other embodiments.

[0097] The various aspects of this disclosure are summarized below as an appendix.

[0098] (Note 1) An object detection information acquisition unit acquires object detection information, which includes object state information, which is information about multiple state-related items of an object that is detected by a sensor. A type candidate determination unit that determines a candidate type of object based on a combination of the object state information corresponding to a plurality of state-related items, A correction target information determination unit outputs a correction target state-related item, which is the state-related item to be corrected, and a correction value candidate as correction information based on the combination of the type candidate and the object state information. A correction unit that corrects the object state information based on the object detection information and the correction information to generate object information which is the corrected object detection information, A detection information correction device comprising an object information output unit that outputs the aforementioned object information. (Note 2) The correction unit includes a reliability calculation unit that calculates a first reliability, which is the reliability of the object state information, and a second reliability, which is the reliability of the correction information. A reliability comparison unit compares the first reliability and the second reliability to obtain a determination result of whether or not the object state information should be corrected with the correction information, The detection information correction device according to Appendix 1, further comprising an object information generation unit that, when the determination result indicates that correction is necessary, corrects the object state information based on the correction information to generate the object information. (Note 3) The detection information correction device according to Appendix 1, characterized in that the correction unit corrects the object state information based on the history of the object detection information in addition to the object detection information and the correction information, and generates the object information. (Note 4) The correction unit further calculates a third confidence level, which is the confidence level of the object state information included in the history of the object detection information, A reliability comparison unit compares the first reliability, the second reliability, and the third reliability to obtain a determination result of whether or not the object state information should be corrected with the correction information or the object state information included in the history of the object detection information, The detection information correction device according to Appendix 2, further comprising: an object information generation unit that, when the determination result indicates that correction is necessary, corrects the object state information based on the correction information or the object state information included in the history of the object detection information, and generates the object information. (Note 5) The detection information correction device according to Appendix 4, characterized in that the reliability comparison unit obtains a determination result that the object state information should be corrected with the correction information or the object state information included in the history of the object detection information if the first reliability is less than a predetermined reliability determination threshold and the second reliability or the third reliability is equal to or greater than the reliability determination threshold. (Note 6) The detection information correction device according to Appendix 1, characterized in that the correction unit generates the object information by selecting whether or not to rewrite the object state information with the correction information for each of the multiple state-related items. (Note 7) The detection information correction device according to Appendix 3, characterized in that the correction unit generates object information by selecting, for each of the multiple state-related items, to rewrite the object state information with the correction information, to rewrite it with the object state information included in the history of the object detection information, or not to rewrite it. (Note 8) The detection information correction device according to Appendix 1, wherein the type candidate determination unit has a determination table arranged in a predetermined first order for the combination of object state information, and determines the type candidate based on this determination table. (Note 9) The detection information correction device according to Appendix 1, wherein the correction target information determination unit has a determination table arranged in a predetermined second order for the combination of the type candidate and the object state information, and outputs the correction target state related items and the correction value candidate as correction information based on this determination table. (Note 10) The detection information correction device according to any one of the appendices 1 to 9, characterized in that the type candidate determination unit has a determination table arranged in a predetermined third order for combinations of ambient information input from the ambient information transmitter and the object state information, and outputs the correction target state related items and the correction value candidates as correction information based on this determination table. (Note 11) The detection information correction device according to any one of the appendices 1 to 9, characterized in that the correction target information determination unit determines the state-related items to be corrected for each type candidate based on the combination of ambient information input from the ambient information transmitter, the object state information, and the type candidate determined by the type candidate determination unit. (Note 12) A detection information correction device according to any one of the appendices 1 to 9, comprising a correction priority determination unit that determines the priority order for selecting the type candidate based on a combination of ambient information input from an ambient information transmitter and the object state information. (Note 13) A detection information correction program for causing a computer to function as a detection information correction device as described in any one of the items 1 to 12 of the appendix. (Note 14) A sensor for detecting the aforementioned object, A detection unit that receives a signal from the aforementioned sensor and generates object detection information relating to the aforementioned object, A detection information correction system comprising a detection information correction device described in any one of the appendices 1 to 12. (Note 15) The detection information correction system according to Appendix 14, characterized in that the sensor detects reflected electromagnetic waves that are reflected by the object from electromagnetic waves radiated toward the object. (Note 16) The detection information correction system according to Appendix 14, characterized in that the sensor is a camera that photographs the object. (Note 17) The object information is input from the detection information correction system described in any one of the appendices 14 to 16. A driving control device that uses the object information and map information to control a vehicle in order to avoid contact with the object related to the object information. [Explanation of Symbols]

[0099] 100: Sensor, 301: Object detection information acquisition unit, 303: Type candidate determination unit, 304: Correction target information determination unit, 305: Correction unit, 306: Object information output unit, 300: Detection information correction device, 3051: Reliability calculation unit, 3052: Reliability comparison unit, 3053: Object information generation unit, 101: Surrounding information transmitter, 307: Correction priority determination unit, 200: Detection unit, 1000: Detection information correction system, 2000: Driving control device.

Claims

1. An object detection information acquisition unit acquires object detection information, which includes object state information, which is information about multiple state-related items of an object that is detected by a sensor. A type candidate determination unit that determines a candidate type of object based on a combination of the object state information corresponding to a plurality of state-related items, A correction target information determination unit outputs a correction target state-related item, which is the state-related item to be corrected, and a correction value candidate as correction information based on the combination of the type candidate and the object state information. A correction unit that corrects the object state information based on the object detection information and the correction information to generate object information which is the corrected object detection information, A detection information correction device comprising an object information output unit that outputs the aforementioned object information.

2. The correction unit includes a reliability calculation unit that calculates a first reliability, which is the reliability of the object state information, and a second reliability, which is the reliability of the correction information. A reliability comparison unit compares the first reliability and the second reliability to obtain a determination result of whether or not the object state information should be corrected with the correction information, The detection information correction device according to claim 1, further comprising: an object information generation unit that, if the determination result indicates that correction is necessary, corrects the object state information based on the correction information and generates the object information.

3. The detection information correction device according to claim 1, characterized in that the correction unit corrects the object state information based on the history of the object detection information in addition to the object detection information and the correction information, and generates the object information.

4. The correction unit further calculates a third confidence level, which is the confidence level of the object state information included in the history of the object detection information, A reliability comparison unit compares the first reliability, the second reliability, and the third reliability to obtain a determination result of whether or not the object state information should be corrected with the correction information or the object state information included in the history of the object detection information, The detection information correction device according to claim 2, further comprising: an object information generation unit that, when the determination result indicates that correction is necessary, corrects the object state information based on the correction information or the object state information included in the history of the object detection information and generates the object information.

5. The detection information correction device according to claim 4, characterized in that the reliability comparison unit obtains a determination result that the object state information should be corrected with the correction information or the object state information included in the history of the object detection information if the first reliability is less than a predetermined reliability determination threshold and the second reliability or the third reliability is equal to or greater than the reliability determination threshold.

6. The detection information correction device according to claim 1, characterized in that the correction unit generates the object information by selecting whether or not to rewrite the object state information with the correction information for each of the multiple state-related items.

7. The detection information correction device according to claim 3, characterized in that the correction unit generates object information by selecting, for each of the multiple state-related items, to rewrite the object state information with the correction information, to rewrite it with the object state information included in the history of the object detection information, or not to rewrite it.

8. The detection information correction device according to claim 1, wherein the type candidate determination unit has a determination table arranged in a predetermined first order for the combination of object state information, and determines the type candidate based on this determination table.

9. The detection information correction device according to claim 1, wherein the correction target information determination unit has a determination table arranged in a predetermined second order for the combination of the type candidate and the object state information, and outputs the correction target state related items and the correction value candidate as correction information based on this determination table.

10. The detection information correction device according to claim 1, characterized in that the type candidate determination unit has a determination table arranged in a predetermined third order for combinations of ambient information input from the ambient information transmitter and the object state information, and outputs the correction target state related items and the correction value candidates as correction information based on this determination table.

11. The detection information correction device according to claim 1, characterized in that the correction target information determination unit determines the state-related items to be corrected for each type candidate based on the combination of ambient information input from the ambient information transmitter, the object state information, and the type candidate determined by the type candidate determination unit.

12. The detection information correction device according to claim 1, further comprising a correction priority determination unit that determines the priority order for selecting the type candidate based on a combination of ambient information input from an ambient information transmitter and the object state information.

13. A detection information correction program for causing a computer to function as a detection information correction device according to any one of claims 1 to 12.

14. A sensor for detecting the aforementioned object, A detection unit that receives a signal from the aforementioned sensor and generates object detection information relating to the aforementioned object, A detection information correction system comprising a detection information correction device according to any one of claims 1 to 12.

15. The detection information correction system according to claim 14, characterized in that the sensor detects reflected electromagnetic waves that are reflected by the object from electromagnetic waves radiated toward the object.

16. The detection information correction system according to claim 14, characterized in that the sensor is a camera that photographs the object.

17. The object information is input from the detection information correction system described in claim 14, A driving control device that uses the object information and map information to control a vehicle in order to avoid contact with the object related to the object information.

Images