Self-position estimation device, self-position estimation method, and moving body control system

Abstract

To provide a self-position estimation device capable of estimating a self-position of a moving body even when there is a deviation between prior information and an environment during movement.SOLUTION: A self-position estimation device for estimating a self-position of a mobile object by using measurement data acquired by a measurement unit mounted on the mobile object moving in a moving environment, the self-position estimation device comprising: a specific object detection unit which detects, as a specific object, an object having a correspondence with an object unique to the moving environment described in the pre-acquired moving environment data from the measurement data measured by the measurement unit; a general object detection unit which detects, as a general object, an object not described in the pre-acquired moving environment data from the measurement data measured by the measurement unit; A position calculation unit configured to obtain self-position coordinates in a moving environment from object detection results detected by the specific object detection unit and the general object detection unit.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a self-position estimation device for a mobile body, a self-position estimation method, and a mobile body control system. [Background technology]

[0002] At nuclear power plants, inspection, maintenance, and repair work on the reactor pressure vessel and reactor internal structures must be carried out in a radioactive environment. To reduce radiation exposure to workers, remote operation devices such as radiation-resistant robots are used to carry out inspection and repair work remotely, without workers having to enter the radioactive environment.

[0003] In many tasks, a robot must move to the target location and perform the specified tasks one by one, so it is necessary for the robot to know its own position as it proceeds with the task.One possible method for estimating the self-position of a mobile object such as a robot is to attach sensors such as cameras and LiDAR to the mobile object and use the information from these sensors.

[0004] Specifically, for example, a database linking positions in the moving environment with the features detected at those positions can be prepared in advance, and the position can be estimated by comparing the database with features extracted from images acquired by sensors or 3D shapes using filtering processes. Furthermore, since creating a database is difficult in unknown environments, a method called Simultaneous Localization and Mapping (SLAM) is used, which generates a map and estimates the vehicle's position simultaneously. In controlled environments such as power plants, prior information such as construction drawings and equipment installation information is available, so it is possible to build a database based on this information and take the former approach.

[0005] However, the above method does not work if there is a discrepancy between the prior information and the environment at the time of movement, because the acquired features will not match the features in the database. These discrepancies arise for a variety of reasons, such as changes in the appearance of the equipment over time, or the installation, relocation, or disposal of equipment that is not reflected in the prior information.

[0006] As described above, a method for estimating the self-position of a moving object needs to be able to estimate its self-position robustly even when there is a discrepancy between prior information such as drawings and the environment during movement. For example, known technologies for estimating the self-position of a moving object using prior information include a camera position information management method, a camera position information management system, a construction site monitoring system, a moving object position information management method, and a moving object position information management system, as described in Patent Document 1.

[0007] Patent Document 1 aims to provide a camera position information management method and system that automatically updates camera position information, and describes it as "a camera position information management method including a position estimation step of estimating the position of the camera based on the result of comparing an image taken by a camera installed at a construction site with a 3D model created in advance for the building under construction, and a position information update step of updating the camera position information stored in a specified memory area in accordance with the estimation result in the position estimation step." [Prior art documents] [Patent documents]

[0008] [Patent Document 1] Japanese Patent Application Publication No. 2023-134924 Summary of the Invention [Problem to be solved by the invention]

[0009] The system described in Patent Document 1 compares the appearance of a camera image on a Building Information Modeling (BIM) model with that of an actual image, and estimates its own position.

[0010] However, the method described in Patent Document 1 has the problem that if there is a discrepancy between the prior information (BIM model) and the environment during movement, the camera images will be different, making it difficult to estimate the user's own position.

[0011] The present invention has been made in consideration of this background, and aims to provide a self-position estimation device, a self-position estimation method, and a mobile body control system that can estimate the self-position of a mobile body even when there is a discrepancy between prior information and the environment during movement. [Means for solving the problem]

[0012] In view of the above, the present invention is a self-position estimation device that estimates the self-position of a moving body using measurement data acquired by a measurement unit mounted on the moving body moving in a moving environment, the self-position estimation device comprising: a specific object detection unit that detects, from the measurement data measured by the measurement unit, an object that corresponds to an object specific to the moving environment described in the moving environment data acquired in advance, as a specific object; a general object detection unit that detects, from the measurement data measured by the measurement unit, an object that is not described in the moving environment data acquired in advance, as a general object; and a position calculation unit that calculates the self-position coordinates in the moving environment from the object detection results detected by the specific object detection unit and the general object detection unit.

[0013] Furthermore, the present invention is described as "a self-location estimation method that is realized using a computer and that estimates the self-location of a moving body using measurement data acquired by a measurement unit, wherein the computer detects, from the measurement data measured by the measurement unit, objects that correspond to objects specific to the moving environment described in the moving environment data acquired in advance, as specific objects, and detects, from the measurement data measured by the measurement unit, objects that are not described in the moving environment data acquired in advance, as general objects, and calculates the self-location coordinates in the moving environment from the specific objects and the general objects."

[0014] Furthermore, the present invention provides a mobile body control system comprising a mobile body, a measurement device, a control device, and a self-position estimation device, wherein the self-position estimation device comprises: a specific object detection unit that detects, as specific objects, objects that correspond to objects specific to the mobile environment in previously acquired mobile environment data from measurement data acquired by a measurement unit mounted on the mobile body moving through the mobile environment; a general object detection unit that detects, as general objects, objects that are not described in the previously acquired mobile environment data from the measurement data acquired by the measurement unit; and a position calculation unit that calculates self-position coordinates in the mobile environment from the object detection results detected by the specific object detection unit and the general object detection unit. [Effects of the Invention]

[0015] According to the present invention, it is possible to provide a self-location estimation device that functions in an environment where there is a deviation from prior information. Problems, configurations, and effects other than those described above will become clear from the following description of the embodiments. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a diagram showing an example of the overall configuration of a mobile object control system according to a first embodiment of the present invention; [Figure 2] 1 is a block diagram showing an example of the functional configuration of a self-location estimation device according to a first embodiment of the present invention; [Figure 3] 3 is a flowchart of a self-position estimation process of the self-position estimation device according to the first embodiment of the present invention. [Figure 4] FIG. 2 is a block diagram showing an example of the functional configuration of a position calculation unit according to the first embodiment of the present invention. [Figure 5] 4 is a flowchart of a position calculation process of a position calculation unit according to the first embodiment of the present invention. [Figure 6] FIG. 3 is a diagram showing an example of a display on an input / output unit according to the first embodiment of the present invention. [Figure 7] 10 is a flowchart of a position calculation process of a position calculation unit according to the second embodiment of the present invention. [Figure 8] 10 is a flowchart of a position calculation process of a position calculation unit according to a third embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0017] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In this specification and drawings, components having substantially the same functions or configurations are designated by the same reference numerals, and redundant description will be omitted. [Example]

[0018] First, an example of the overall configuration of a mobile object control system according to a first embodiment of the present invention will be described with reference to Fig. 1. As shown in Fig. 1, the mobile object control system 1 is configured with a mobile object 7, a measurement device 2, a control device 3, and a self-position estimation device 4, and estimates the self-position in a mobile environment in which objects 5, 6, etc. are located. At least the measurement device 2 is disposed on the mobile object 7.

[0019] The mobile object 7 has at least a moving mechanism such as crawlers or tires, and moves according to commands from the control device 3. The moving form is not limited, and it may be a flying object like a drone, or a swimming object assuming that the moving environment is underwater.

[0020] The measurement device 2 is a device that measures at least data related to the surrounding environment of the moving object 7. The measurement device 2 is, for example, a camera that captures images of the environment, a LiDAR that measures the shape of the environment, etc. Furthermore, there may be more than one measurement device 2. The measurement device 2 may include a device that measures the state of the moving object 7, such as an inertial sensor. The measurement device 2 measures data of the surrounding environment and outputs a detection signal (measurement data) corresponding to the detected content to the self-position estimation device 4. In this embodiment, it is assumed that the measurement device 2 is a camera that captures images of the surrounding environment, etc.

[0021] The control device 3 has an input / output unit that receives operation input from an operator, and outputs a control command to the moving object 7 based on the input operation command to control the movement of the moving object 7. The control command is, for example, a signal that indicates a current value, a voltage value, etc. for an actuator (such as a motor) provided in a movement mechanism or the like provided on the moving object 7. When the moving object 7 receives a control command from the control device 3, a built-in drive circuit supplies a drive signal to the corresponding actuator.

[0022] The self-location estimation device 4 is a terminal device that receives measurement data from the measurement device 2 as input and estimates the self-location of the moving object 7. The self-location estimation device 4 and the control device 3 may be mounted on the moving object 7, or may be installed at a remote location via a wired connection.

[0023] Next, a description will be given of the functional configuration of the self-location estimation device 4. Fig. 2 is a block diagram showing an example of the functional configuration of the self-location estimation device 4 according to the first embodiment of the present invention. The self-location estimation device 4 configured using a computer includes an input / output unit 41, a control unit 42, and a storage unit 43, as shown in Fig. 2.

[0024] User interface devices such as a display, keyboard, and mouse are connected to the input / output unit 41. The input / output unit 41 displays the object detection results of the object detection model (described later) and the self-position estimation results of the position calculation unit. The input / output unit 41 also includes a communication device, and can send and receive data (signals) to and from devices such as the measurement device 2 and the control device 3.

[0025] The storage unit 42 includes storage devices such as a ROM (Read Only Memory), a RAM (Random Access Memory), an SSD (Solid State Drive), etc. The storage unit 42 stores object position data D1, an object detection model M, object detection data D2, and position calculation data D3.

[0026] The control unit 43 is configured, for example, by a CPU (Central Processing Unit) not shown, and when the processing of the control unit 43 is expressed as functions, it can be said to have the functions of an acquisition unit 431, an object detection unit 432, a position calculation unit 433, and a learning unit 434, as shown in Fig. 2. The control unit 43 controls the operation of each component of the self-position estimation device 4. Note that the control unit 43 may be configured by a GPU (Graphics Processing Unit) instead of a CPU, or may be configured by a CPU and a GPU.

[0027] The acquisition unit 431 acquires a control command for the moving object 7 from the control device 3. The acquisition unit 431 also acquires measurement data from the measurement device 2.

[0028] The object detection unit 432 is a device that uses the object detection model M to detect objects from measurement data acquired from the measurement device 2. For example, if the measurement data is image data, the position (pixel position) of the object in the image and the type of object are output. The output does not have to be the pixel position, and may be the size and position of a rectangular bounding box that represents the area where the object exists, or segmentation that indicates the area in the image. Furthermore, multiple objects may be detected simultaneously. In the present invention, object detection in the object detection unit 432 is performed by dividing objects into specific objects and general objects (other than specific objects), as will be described later.

[0029] The position calculation unit 433 calculates the position of the moving object 7 using the detection result of the object detection unit 432, the object position data D1, the object detection data D2, and the position calculation data D3.

[0030] The learning unit 434 learns the object detection model M based on the object information of the object position data D1.

[0031] Here, the object position data D1 is data (prior information) that associates information about objects present in the mobile environment with positions on the drawing based on a drawing of the mobile environment, and the object position data D1 can be called mobile environment data. The object information includes, for example, information about the object's shape, which can uniquely identify an object in the mobile environment. In the case of identical devices, it is desirable to associate the object information with a unique number or the like. To be used for training the object detection model M, the object information in the object position data D1 includes at least information (such as an image or 3D shape information) equivalent to the measurement data acquired from the measurement device 2 so that the object can be detected from the measurement data acquired from the measurement device 2.

[0032] In this way, the object position data D1 is data that includes the moving environment of the observed object and the original positions and original shapes of the objects present therein, and can be said to be environmental map information that has been created and grasped in advance. On the other hand, it would be sufficient if the object detection data D2 actually measured in the moving environment of the observed object were the same as the previously grasped object position data D1, but since differences in position and shape may occur for some reason, the present invention seeks to determine the difference between them to ensure accurate position estimation.

[0033] 2 includes a specific object detection unit 4321 that detects specific objects among objects present in a mobile environment, and a general object detection unit 4322 that detects general objects (other than specific objects), and detects objects using an object detection model M. The configuration of the object detection model M is not important, and an object detection model trained on an open source dataset, such as YOLO, may be used.

[0034] The specific object detection unit 4321 detects a specific object that is an object specific to the moving environment and is included in the object position data D1. Furthermore, if the object position data D1 includes multiple identical or similar devices, the specific object detection unit 4321 uniquely identifies each of them. In this way, the definition of a specific object is "an object specific to the moving environment and included in the object position data D1," or in other words, an object that is known to be in the moving environment.

[0035] The generic object detection unit 4322 detects objects not included in the object position data D1 (all objects that cannot be defined as specific objects) as generic objects. The generic object detection unit 4322 also detects objects in a category that cannot be uniquely identified even if they are included in the object position data D1 (hereinafter referred to as generic objects).

[0036] As a result, for example, if the object position data D1 includes similar pumps A and B, and the measurement data (object detection data D2) obtained from the measuring device 2 includes information about pump A, the specific object detection unit 4321 detects the pump as "pump A" to distinguish between pumps. On the other hand, the general object detection unit 4322 does not identify the object, but detects it as a category called "pump."

[0037] Since the specific object detection unit 4321 needs to uniquely identify an object, if the measurement data obtained from the measurement device 2 deviates from the data included in the object position data D1, the object may not be detected properly. Possible cases of data discrepancy include loss or deterioration of the object, or differences in the quality of the measurement data due to environmental conditions (low resolution, different lighting conditions, etc.). As a result, the specific object detection unit 4321 may not be able to recognize an object as a specific object even though it is actually a specific object.

[0038] On the other hand, the general object detection unit 4322 does not specify an object, and therefore does not require detailed information for detection compared to the specific object detection unit 4321. Therefore, there is a high possibility that a specific object that could not be detected by the specific object detection unit 4321 can be detected as a general object.

[0039] 3 shows a flowchart of the self-position calculation process of the self-position estimation unit 4 using the detection results of the object detection unit 432. By repeating this process, the self-position of the moving object 7 is continuously estimated.

[0040] 3, measurement data is acquired from the measurement device 2. Next, in processing steps S12 and S13, the measurement data is input to the specific object detection unit 4321 and the general object detection unit 4322, respectively, to acquire the object detection results. Furthermore, in processing step S14, the object detection results in processing steps S12 and S13 are input to the position calculation unit 433, which calculates global coordinates on the drawing.

[0041] Then, in processing step S15, the object detection results obtained in processing steps S12 and S13 and the global coordinates obtained in processing step S14 are stored as object detection data D2 and position calculation data D3 in storage unit 2. It is expected that the stored object detection data D2 and position calculation data D3 will be used for display on input / output unit 41. An example of the processing of position calculation unit 433 in processing step S15 will be described later with reference to FIG. 4.

[0042] According to the flow of Figure 3, if there is a discrepancy between prior information such as a drawing and the actual moving environment, there is a high possibility that an object detection result will not be obtained in processing step S12 (the object will not be recognized as a specific object). However, by using general object detection results in which objects are relatively easy to detect, as in processing step S13, it is possible to maintain an estimate of the self-position even if an object is not detected in processing step S12.

[0043] 4 is a block diagram showing an example of the functional configuration of the position calculation unit 433. The position calculation unit 433 is made up of a global coordinate estimation unit 4331, a movement amount estimation unit 4332, and a filter unit 4333, as shown in FIG.

[0044] The global coordinate estimation unit 4331 compares the detection result of the specific object detection unit 4321 with the object information in the object position data D1 to estimate global coordinates on the drawing. The movement amount estimation unit 4332 compares the detection result of the general object detection unit 4322 at time t with the detection result at the previous time t-1 to estimate the movement amount of the moving object 7 at one time.

[0045] It should be noted that objects are assigned to specific objects or general objects based on the measurement data from the measurement device, and depending on the appearance of the measurement data, an object that was detected as a general object [specific object] at time t1 may be detected as a specific object [general object] at time t2. However, if a specific object is detected at any time, the global coordinates can be calculated, and therefore the movement amount estimation unit 4332 can handle this by referring only to general objects.

[0046] The filter unit 4333 filters the global coordinates obtained from the global coordinate estimation unit 4331 and the estimated values ​​of the movement amount obtained from the movement amount estimation unit 4332, and outputs corrected estimated values. By using a probabilistic filter such as a Kalman filter, which has the movement amount and global coordinates as its state, it is expected that the estimation value will be robust against variations. Furthermore, even when multiple objects are detected simultaneously, the use of a probabilistic filter is expected to enable the use of information on multiple objects.

[0047] 5 is a flowchart of the position calculation process of the position calculation unit 433. By this process, the position of the moving body 7 is estimated based on the object detection result.

[0048] First, in processing steps S141 and S142, a specific object detection result and a general object detection result are obtained, respectively. Next, in processing step S143, the global coordinate estimation unit 4331 compares the specific object detection result with the object information in the object position data D1 to obtain estimated values ​​of global coordinates on the drawing. Meanwhile, in processing step S144, the movement amount estimation unit 4332 compares the general object detection result with the result from one time point earlier and estimates the movement amount of the moving object 7 from the amount of change.

[0049] In processing step S145, the global coordinates and estimated values ​​of the amount of movement obtained in processing steps S143 and S144 are input to filter unit 4333, which corrects and outputs the estimated coordinates. Note that this correction can be thought of as correcting global coordinates. Since it is expected that any object detection result will contain noise (including average values ​​and abnormal values), the idea is to make corrections while referring to both values.

[0050] Such a position calculation unit 433 can appropriately refer to the specific / general object detection results, and even if a specific object is not detected, can refer to the general object detection results to estimate the self-position of the moving object 7.

[0051] Fig. 6 shows an example of the display of the input / output unit 41. The display screen 330 in Fig. 6 may be configured to display, for example, object detection results obtained from measurement data such as images and 3D shape data, and the estimated self-position of the mobile object 7, in order to assist the decision-making of an operator who operates and monitors the mobile object 7. In addition, the current timestamp, coordinate values ​​of the self-position, reliability, etc. may also be displayed.

[0052] 5 is a method of simultaneously using the detection results of specific objects and general objects. However, more simply, the position calculation unit 433 may be configured on a rule basis such that when no detection result is obtained from the specific object detection unit 4321, the result from the general object detection unit 4322 is used. [Example]

[0053] In the first embodiment, it is assumed that the actual position of a specific object corresponds to the position in the drawing. In the second embodiment, a method for accurately estimating the self-position of a moving object will be described, even if the actual position of a specific object does not correspond to the position of the object position data (drawing).

[0054] For example, when a specific object whose position differs from that of the drawing is detected at time t, the self-location estimation result of Example 1 is considered to have a large distance between the self-location estimated at time t-1 and the self-location updated at time t. In Example 2, if the distance is large relative to a set threshold L, it is detected that the specific object has moved to a position different from that of the drawing, thereby enabling accurate self-location estimation even when the actual position of the specific object does not correspond to the position of the drawing, as will be described later. The threshold L may be determined based on the allowable error in the estimated position and the maximum amount of movement of the moving object 7 that can occur at one time.

[0055] Fig. 7 is a flowchart of the position calculation process of the position calculation unit 433 according to the second embodiment. The difference from Fig. 5 shown in the first embodiment is that processing steps S146 and S147 are added to detect the movement of a specific object without referring to the position information of the object. The processing steps S141 to S144 and S145 are the same as those in Fig. 5, and therefore detailed description thereof will be omitted.

[0056] In processing step S146, the global coordinate estimation unit 4331 calculates the distance between the global coordinate estimated in processing step S143 and the global coordinate from one time point earlier, and determines whether the distance is smaller than threshold L. If the determination is Yes, it is determined that the specific object has not moved, and the process proceeds to processing step S145 as in Fig. 5. If the determination is No, it is determined that the specific object has moved, and in processing step S147, the global coordinate obtained based on the object that exceeded threshold L is erased, and the process proceeds to processing step S145.

[0057] The processing of the position calculation unit 433 involves evaluating the relationship between the distance between the global coordinate estimated by the global coordinate estimation unit 433 at a specific time and the global coordinate at a time prior to the specific time and a predetermined threshold value, and if the distance exceeds the predetermined threshold value, determining that an object specific to the environment in the previously acquired moving environment data has moved relative to the object position between the time the moving environment data was acquired and the time the position was calculated.

[0058] According to this processing, if a specific object different from the prior information is detected, the moment that specific object is detected, the position of the moving body will move significantly at a certain time, and this will be determined in processing step S146, which will result in the detection of some kind of abnormality.

[0059] A plurality of specific objects may be detected, and a corresponding plurality of global coordinates of the estimated moving object 7 are obtained. Usually, the obtained plurality of global coordinates are appropriately referenced by the filter unit 4334. However, since inputting obviously abnormal values ​​due to the movement of a specific object or the like will cause a decrease in estimation accuracy, the obviously abnormal values ​​are eliminated in this way before inputting them to the filter unit.

[0060] If the position calculation unit 433 determines that the position of a specific object differs from the previous drawing information, the position in the object information of the object position data D1 may be updated. Since the amount of information that can be used to identify the position increases, more robust self-position estimation is expected. [Example]

[0061] In the first embodiment, it is assumed that the object is stationary, but if there is a floating object in water, it is assumed that the object will move due to a water current generated by the movement of the moving body 7. In the third embodiment, a self-localization method will be described in which the dynamic movement of an object is detected, and depending on the detection result, the detection result of the dynamically moving object is not used for self-localization, and the detection result of an object whose dynamic movement is not detected is used when estimating the position of the moving body 7, thereby being able to cope with the case where a moving object is present in a moving environment.

[0062] For example, when a dynamically moving specific object is detected at time t, the distance between the self-position estimated at time t-1 and the self-position updated at time t, or the amount of movement in the case of a general object, becomes large. If the distance is larger than the set threshold L, it is detected whether the referenced object has dynamically moved. Note that the threshold L may be determined based on the allowable error in the estimated position and the maximum amount of movement of the moving object 7 at one time, as in the second embodiment.

[0063] Fig. 8 is a flowchart of the position calculation process of the position calculation unit 433 according to the third embodiment. The difference from Fig. 5 shown in the first embodiment is that processing steps S146 to S149 are added in order to detect dynamic movement of an object without referring to the position information of the object. The processing steps S141 to S145 are the same as the processing in Fig. 5, and therefore detailed description thereof will be omitted.

[0064] 8, in processing step S146, the global coordinate estimation unit 4331 calculates the distance between the global coordinate estimated in processing step S143 and the global coordinate from one time point earlier, and determines whether the distance is smaller than threshold L. If the determination is Yes, it is determined that there is no dynamic movement of the specific object, and the process proceeds to processing step S145 as in Fig. 5. If the determination is No, it is determined that there is dynamic movement of the specific object, and in processing step S147, the global coordinate obtained based on the object that exceeds threshold L is erased, and the process proceeds to processing step S145.

[0065] 8, in processing step S148, the movement amount estimation unit 4332 determines whether the movement amount estimated in processing step S144 is smaller than the threshold value L. If the determination is Yes, it is determined that there is no dynamic movement of the general object, and the process proceeds to processing step S145 as in Fig. 5. If the determination is No, it is determined that there is dynamic movement of the general object, and in processing step S149, the movement amount obtained based on the object that exceeds the threshold value L is deleted, and the process proceeds to processing step S145.

[0066] Although there are exceptions, such as when the moving environment of the moving object 7 is underwater and there is a water current that is not related to the movement of the moving object 7, in most cases the moving object 7 will not move without a control command. Therefore, by referring to the control command and dynamically changing the threshold L, it is possible to accurately detect the movement of an object. If there is no control command and the moving object 7 does not move, there is no movement in the global coordinates, so the threshold L may be reduced.

[0067] The present invention is not limited to the above-described embodiments and includes various modifications. For example, the above-described embodiments have been described in detail to clearly explain the present invention, and the present invention is not necessarily limited to those including all of the described configurations. Furthermore, it is possible to replace part of the configuration of one embodiment with the configuration of another embodiment, or to add the configuration of another embodiment to the configuration of one embodiment. Furthermore, it is also possible to add, delete, or replace part of the configuration of each embodiment with other configurations.

[0068] Furthermore, the above-described configurations, functions, processing units, processing means, etc. may be partially or entirely implemented in hardware, for example, by designing them as integrated circuits. Furthermore, the above-described configurations, functions, etc. may be implemented in software, with a processor interpreting and executing a program that implements each function. Information such as the programs, tables, and files that implement each function can be stored in a memory, a recording device such as a hard disk or SSD (Solid State Drive), or a recording medium such as an IC card, SD card, or DVD.

[0069] In addition, the control lines and information lines shown are those that are considered necessary for the explanation, and do not necessarily show all the control lines and information lines in the product. In reality, it can be assumed that almost all components are interconnected. [Explanation of symbols]

[0070] 1: Mobile control system 7: Mobile 2: Measuring equipment 3: Control device 4: Self-position estimation device 5, 6...object 41: Input / output section 42: Storage part 43: Control unit D1: Object position data M: Object detection model D2: Object detection data D3: Position calculation data 431: Acquisition Department 432: Object detection unit 433: Position calculation section 434: Learning Department 4331: Global coordinate estimation unit 4332: Movement amount estimation part 4333: Filter section 4321: Specific object detection unit 4322: General object detection unit

Claims

1. A self-position estimation device that estimates the self-position of a moving body using measurement data acquired by a measurement unit mounted on the moving body that moves in a mobile environment, A self-position estimation device characterized by comprising: a specific object detection unit that detects, from the measurement data measured by the measurement unit, an object that corresponds to an object specific to the moving environment in the moving environment data acquired in advance, as a specific object; a general object detection unit that detects, from the measurement data measured by the measurement unit, an object that is not described in the moving environment data acquired in advance, as a general object; and a position calculation unit that calculates the self-position coordinates in the moving environment from the object detection results detected by the specific object detection unit and the general object detection unit.

2. The self-location estimation device according to claim 1 , The position calculation unit is a global coordinate estimation unit that estimates the self-position coordinates of the moving body in the moving environment data from the detection result of the specific object detection unit, a movement amount estimation unit that estimates the self-movement amount of the moving body in the moving environment from the detection result of the general object detection unit, and a filter unit that corrects the self-position coordinates from the estimation results of the position coordinates and movement amount estimated by the global coordinate estimation unit and the movement amount estimation unit.

3. The self-location estimation device according to claim 2, The position calculation unit evaluates the relationship between the distance between the global coordinate estimated by the global coordinate estimation unit at a specific time and the global coordinate at a time prior to the specific time and a predetermined threshold, and if the distance exceeds the predetermined threshold, determines that an environment-specific object in the previously acquired mobile environment data has moved relative to the object position between the time the mobile environment data was acquired and the time the position was calculated.

4. The self-location estimation device according to claim 2, the position calculation unit evaluates the relationship between the distance between the global coordinate estimated by the global coordinate estimation unit at a specific time and the global coordinate at a time prior to the specific time and a predetermined threshold, and if the distance exceeds the predetermined threshold, determines that an object specific to the environment described in the previously acquired movement environment data has moved; the position calculation unit evaluates the relationship between the amount of movement estimated by the movement amount estimation unit at the current time and a threshold, and if the amount of movement exceeds the threshold, determines that an object in the previously acquired movement environment data has dynamically moved.

5. 5. The self-location estimation device according to claim 3 or 4, The self-position estimation device is characterized in that the threshold value is calculated from an allowable amount of estimation error and a maximum movement amount of a mobile object at one time.

6. The self-location estimation device according to claim 5, The self-position estimation device is characterized in that the threshold value is set relatively small when there is no control command for the moving body, and set relatively large when there is a control command.

7. The self-location estimation device according to claim 1 , a self-position estimation device comprising an input / output unit that displays the object detection results of the specific object detection unit and the general object detection unit, and the self-position estimation result of the position calculation unit;

8. The self-location estimation device according to claim 1 , A self-position estimation device characterized in that the specific object detection unit detects an object determined from measurement data as a specific object if it is in the same position and has the same shape as an object in previously acquired movement environment data.

9. A self-location estimation method that is realized using a computer and estimates a self-location of a moving object using measurement data acquired by a measurement unit, comprising: The computer detects, from the measurement data measured by the measurement unit, objects that correspond to objects specific to the mobile environment in the mobile environment data acquired in advance as specific objects, and detects, from the measurement data measured by the measurement unit, objects that are not in the mobile environment data acquired in advance as general objects, and calculates the self-position coordinates in the mobile environment from the specific objects and the general objects.

10. A mobile object control system including a mobile object, a measurement device, a control device, and a self-position estimation device, the self-location estimation device, a specific object detection unit that detects, as a specific object, an object that corresponds to an object specific to the mobile environment in the mobile environment data that has been previously acquired from the measurement data acquired by the measurement device mounted on the mobile body that moves through the mobile environment; a general object detection unit that detects, from the measurement data acquired by the measurement device, an object that is not described in the previously acquired movement environment data as a general object; a position calculation unit that calculates the coordinates of the vehicle's own position in the moving environment from the object detection results detected by the specific object detection unit and the general object detection unit.

Images