Map generation - self-location inference apparatus

Abstract

The present invention provides a map generation-self position estimation device capable of highly accurately estimating a self vehicle position and posture. The map generation-self position estimation device of the present invention includes a map generation unit (110) that generates a map composed of a point group and a travel position of a self vehicle based on an output of a sensor; a required accuracy determination unit (120) that determines a position and posture estimation accuracy (required accuracy) required for self position estimation at a next travel based on the output of the sensor and the map generated by the map generation unit (110); a data selection unit (130) that selects a point group included in the map based on the map generated by the map generation unit (110) and the required accuracy determined by the required accuracy determination unit (120); a map recording unit (140) that records a selected map including the point group selected by the data selection unit (130), the travel position, and the required accuracy as a recorded map; and a self position estimation unit (150) that associates a current map composed of the point group selected by the data selection unit (130) and the travel position of the self vehicle with the recorded map recorded in the map recording unit (140), thereby estimating a current position and posture of the vehicle on the recorded map.

Description

TECHNICAL FIELD

[0001] The present application relates to a map generation-self position estimation device. BACKGROUND

[0002] To expand the range of use of an automated driving, a driving assistance system, information acquisition from a map based on self position estimation is important. However, for example, although a map for a driving assistance system has been perfected for an expressway, a map for a residential area such as an ordinary road or a home neighborhood has not been perfected. In this regard, Patent Literature 1 describes the following: "The method of the present application relates to a method for assisting a driver of a vehicle when parking in a parking space, in particular a parking space in which the vehicle is parked every day, such as a garage, using a driver assistance device for the vehicle. Furthermore, the driver assistance device has a learning mode and an action mode. In the learning mode, during parking of the vehicle in the parking space by the driver, the driver assistance device detects and stores reference data related to the surroundings of the parking space using a sensor device. In the learning mode, the driver assistance system also records a reference target position reached by the vehicle in the learning mode. Data having information related to the reference target position, in particular information related to the surroundings, is stored. In the subsequent action mode, sensor data is detected by the sensor device and compared with the reference data. Based on the result of the comparison, the surroundings of the parking space are recognized or identified using the detected sensor data, and the current position of the vehicle relative to the reference target position is determined. The driver assistance device determines a parking path based on the current position of the vehicle relative to the reference target position, along which the vehicle is parked from the current position to the parking space."

[0003] PRIOR ART DOCUMENTS

[0004] PATENT LITERATURE

[0005] Patent Literature 1: Japanese Patent No. 6022447 SUMMARY

[0006] PROBLEMS TO BE SOLVED BY THE INVENTION

[0007] In the related art described in Patent Literature 1, a map is generated from data on objects around the vehicle and a travel route at the time of initial travel, and the own vehicle position is estimated on the generated map at the time of travel after the initial travel, whereby automatic parking using the travel route included in the map is realized. However, if data on objects around the vehicle acquired by a low-precision sensor or method is used for map generation, there is a problem that the precision of the own vehicle position estimation decreases. On the other hand, if data on objects around the vehicle acquired by a low-precision sensor or method is not used for map generation, the information available decreases, so there is a problem that the range in which the own vehicle position can be estimated increases. Although automatic parking is assumed in Patent Literature 1, in automated driving or driving assistance that requires operation in a wider range and under various conditions, both the improvement of the precision of the own vehicle position estimation and the expansion of the range in which the own vehicle position can be estimated must be taken into account.

[0008] The present application has been achieved in view of the above-described circumstances, and has an object to provide a map generation-own vehicle position estimation device capable of estimating the own vehicle position and posture with high precision.

[0009] Technical means for solving the problem

[0010] One of representative map generation-own vehicle position estimation devices of the present application is characterized by comprising: a map generation section that generates a map composed of a travel position of an own vehicle and a point group of measurement results of objects around the own vehicle, from information acquired by a plurality of sensors different in characteristics; a required precision determination section that determines a required precision of own vehicle position estimation on a prescribed place of the travel position of the own vehicle, based on the information acquired by the sensors and the map; a data selection section that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required precision; a map recording section that records the map as a recorded map; and an own vehicle position estimation section that estimates a current own vehicle position in the recorded map by comparison between the selection map generated in travel of the own vehicle and the recorded map.

[0011] Effects of the Invention

[0012] According to the present application, a map generation-own vehicle position estimation device capable of estimating the own vehicle position and posture with high precision can be realized.

[0013] The above-described problems, configurations, and effects will be made clear by the following embodiments.

[0014] Figure 1 A diagram showing the block configuration of the map generation-own vehicle position estimation device 100 of Embodiment 1.

[0015] Figure 2 FIG. 1 is a diagram for illustrating an example of an environment through which a vehicle travels.

[0016] Figure 3 FIG. 2 is a diagram for illustrating an example of a map 310 generated by the map generation section 110.

[0017] Figure 4 FIG. 3 is a diagram for illustrating an example of a required accuracy determination using the point group 311, the point group 312.

[0018] Figure 5 FIG. 4 is a diagram for illustrating an example of a required accuracy determination using a detection result of an object.

[0019] Figure 6 FIG. 5 is a diagram for illustrating an example of a required accuracy determination using a steering angular velocity as vehicle information.

[0020] Figure 7 FIG. 6 is a diagram for illustrating an example of a required accuracy determination using an action state of a driving assist system as vehicle information.

[0021] Figure 8 FIG. 7 is a diagram for illustrating an example of a selection map 380 selected by the data selection section 130.

[0022] Figure 9 FIG. 8 is a diagram for illustrating an example of a grid sampling.

[0023] Figure 10 FIG. 9 is a diagram for illustrating an example of a position and posture of a host vehicle inferred by the host position inference section 150.

[0024] Figure 11 FIG. 10 is a diagram for illustrating a block configuration of the map generation-host position inference apparatus 500 of the second embodiment. DETAILED DESCRIPTION

[0025] Hereinafter, a mode of embodying the present application will be described with reference to the drawings. Also, in all the drawings used for describing the mode of embodying the present application, the same symbols are attached to the portions having the same function, and the repeated description will be omitted at times.

[0026] [First Embodiment]

[0027] Hereinafter, the first embodiment will be described with reference to Figures 1-10The first embodiment of the map generation-self position estimation device will be described. Further, the map generation-self position estimation device is configured in the form of a computer having a processor such as a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), a HDD (Hard Disk Drive), a storage device, and the like. Each function of the map generation-self position estimation device is realized by executing a program stored in the ROM by the processor. The RAM stores data such as intermediate data of the operation under the program executed by the processor.

[0028] (Block configuration)

[0029] Figure 1 A diagram showing the block configuration of the map generation-self position estimation device 100 of the first embodiment will be described. The map generation-self position estimation device 100 takes measurement data of a sensor group 200 (hereinafter sometimes referred to as a sensor 200) as input. The map generation-self position estimation device 100 has a map generation section 110, a required accuracy determination section 120, a data selection section 130, a map recording section 140, and a self position estimation section 150. The sensor 200 has a first external environment sensor 210, a second external environment sensor 220, a relative position sensor 230, and a vehicle information sensor 240.

[0030] The operation of the map generation-self position estimation device 100 is composed of a map generation mode and a position estimation mode. The map generation mode is to generate a map composed of a point group (a set of points obtained in the form of measurement results of objects around the own vehicle) and a travel position of the own vehicle from the output of the sensor 200, and to save the map as a recorded map. The position estimation mode is to estimate the position and posture of the own vehicle in the recorded map from the output of the sensor 200 and the recorded map.

[0031] In the map generation mode, the map generation section 110, the required accuracy determination section 120, the data selection section 130, and the map recording section 140 operate. The map generation section 110 generates a map composed of a point group and a travel position of the own vehicle from the output of the sensor 200. The required accuracy determination section 120 determines the required position and posture estimation accuracy (hereinafter sometimes referred to as required accuracy) of the self position estimation at the next travel from the output of the sensor 200 and the map generated by the map generation section 110. The data selection section 130 selects a point group included in the map from the map generated by the map generation section 110 and the required accuracy determined by the required accuracy determination section 120. The map recording section 140 records the selected map including the point group selected by the data selection section 130, the travel position, and the required accuracy as a recorded map.

[0032] In the position inference mode, the map generation section 110, the data selection section 130, and the own position inference section 150 act. The map generation section 110 and the data selection section 130 act as in the map generation mode. Here, the data selection section 130 performs selection of the point group included in the map using the required accuracy recorded in the map recording section 140 instead of the required accuracy determined by the required accuracy determination section 120. The own position inference section 150 associates the current map constituted by the point group selected by the data selection section 130 and the travel position of the own vehicle with the recorded map recorded in the map recording section 140, thereby inferring the position and posture of the current vehicle on the recorded map.

[0033] Further, in the map generation mode, the map generation section 110 acts on line. That is, the map generation section 110 performs processing for each input from the sensor 200. The required accuracy determination section 120, the data selection section 130, and the map recording section 140 can act on line or off line. For example, the required accuracy determination section 120, the data selection section 130, and the map recording section 140 can be caused to act only once at the end of the map generation mode. In the position inference mode, the map generation section 110, the data selection section 130, and the own position inference section 150 act on line.

[0034] Further, in the present embodiment, the map generation section 110 and the data selection section 130 that act in the map generation mode and the position inference mode are the same (common), but different (another) map generation section 110 and data selection section 130 can be prepared to act in the map generation mode and the position inference mode.

[0035] (Configuration of Sensor)

[0036] The configuration of the sensor 200 that inputs measurement data to the map generation-own position inference apparatus 100 will be described. The sensor 200 is configured to include a plurality of sensors (210, 220, 230, 240) that differ in characteristics.

[0037] The first external environment sensor 210 and the second external environment sensor 220 are external environment sensors that are mounted on a vehicle (own vehicle) to measure the environment around the vehicle. The external environment sensor is, for example, a monocular camera, a stereo camera, a LiDAR, a millimeter wave radar, a sonar, or the like, and measures the three-dimensional position of an object present around the vehicle. Further, in the case of using a monocular camera, the data acquired is an image, and although it is not possible to directly acquire the three-dimensional position, it is possible to measure the three-dimensional position by using a plurality of images with the use of a publicly known motion stereoscopy or the like. In addition, it is also possible to infer the three-dimensional position of a white line, a stop line, a pedestrian crossing, or the like detected in the image by assuming the shape of the road surface.

[0038] The measurement results of the plurality of outside sensors are input to the map generation-self position estimation device 100. In the present embodiment, an example in which the measurement results of two outside sensors different in characteristics, the first outside sensor 210, the second outside sensor 220, and the like are input is described. However, the number of outside sensors is not limited to two, and the measurement results of any number of outside sensors equal to or more than two can be input. Further, the first outside sensor 210 and the second outside sensor 220 include sensors different in processing method of the measurement results of the sensors in addition to sensors different in physical kind. Thus, for example, the first outside sensor 210 can use three-dimensional position measurement under motion stereo method based on an image of a monocular camera, and the second outside sensor 220 can use three-dimensional position measurement under object recognition and road surface shape assumption based on an image of a monocular camera. In this case, even if sensors physically of one kind such as a monocular camera are used, since the processing methods for measuring three-dimensional positions are different, the present embodiment regards the outside sensors as different in characteristics.

[0039] In the present embodiment, a sonar is used as the first outside sensor 210, and motion stereo for an image of a surround camera (monocular camera) is used as the second outside sensor 220. In a driving assistance system in a low-speed region and an automatic driving system, a sonar and a surround camera are mounted on a large number of vehicle models. In the sonar and the motion stereo for an image of a surround camera, there are differences in characteristics such as that the measurement accuracy of the three-dimensional position of the sonar is high but the measurable range is narrow, and the measurable range of the motion stereo for an image of a surround camera is wide but the measurement accuracy of the three-dimensional position is low. However, the first outside sensor 210 is not limited to a sonar, and the second outside sensor 220 is not limited to motion stereo for an image of a surround camera, and can be other outside sensors.

[0040] The relative position sensor 230 is a sensor that outputs the relative position of the own vehicle. Here, like the outside sensor, processing of estimating the relative position of the own vehicle from the measurement result of the sensor is included. Here, the relative position means a position and an attitude with a certain time of the vehicle and the attitude as a reference. For example, the relative position sensor 230 can be a well-known wheel odometry method of estimating the relative movement of the own vehicle from the steering angle of the vehicle and the rotation amount of the tire. Further, it can be a well-known image odometry method or a LiDAR odometry method of estimating the relative movement of the own vehicle from the measurement result of a camera or a LiDAR.

[0041] The vehicle information sensor 240 is a sensor that acquires information of the vehicle such as the speed of the vehicle, the steering angle (corresponding to the driving operation amount of the driver), the gear position, the operation state of a driving assistance system (described later) that assists the driving operation of the driver, and the like.

[0042] (Map generation - Actions of the self-position inference device)

[0043] use Figures 2-10 The processing of each part of the map generation-self-position inference device 100, which takes the measurement data of sensor 200 as input, will be explained.

[0044] (Actions of the map generation department)

[0045] First, use Figure 2 , Figure 3 The processing in the map generation unit 110 will be explained. The map generation unit 110 generates a map consisting of a group of points and the vehicle's own driving position based on the output of the sensor 200 (that is, information obtained by multiple sensors with different characteristics).

[0046] Figure 2 This diagram illustrates an example of the environment a vehicle travels through. The vehicle 300 is traveling at a driving position (driving path) 301, and there are objects such as lane markings 302, buildings 303, trees 304, and utility poles 305 around the driving position 301.

[0047] Figure 3 This is a diagram illustrating an example of a map 310 generated by the map generation unit 110. Figure 3 The map generation unit 110 was shown. Figure 2 Map 310 is generated in the environment shown. Map 310 consists of point group 311 obtained by the first external sensor 210 (sonar in this example), point group 312 obtained by the second external sensor 220 (motion stereo view of the image from the panoramic camera in this example), and driving position 313 obtained by the relative position sensor 230.

[0048] Point groups 311 and 312 are sets of points obtained by external sensors detecting (measuring) objects around the vehicle 300 during its operation.

[0049] The driving position 313 is the position and attitude of the vehicle 300 at various times, obtained by the relative position sensor 230, with the position and attitude of the vehicle 300 at a certain moment as the origin of the coordinate system.

[0050] The point group 311 acquired by the first external sensor 210 and the point group 312 acquired by the second external sensor 220 are acquired by converting the three-dimensional positions of the objects acquired by each external sensor in the sensor frame to the same coordinate system as the travel position 313 using the mounting position and posture of the sensor on the vehicle and the travel position 313 at the time point at which the point group was measured by the sensor. That is, the point group 311, the point group 312, and the travel position 313 are positions and postures in the same coordinate system.

[0051] Further, the point group 311, the point group 312, and the travel position 313 can be positions and postures in a three-dimensional space or a two-dimensional space. When a three-dimensional space is used, the height and the distance that cannot be expressed in a two-dimensional space can be inferred. On the other hand, when a two-dimensional space is used, the data capacity of the map 310 can be reduced.

[0052] (Action of the required accuracy determination section)

[0053] Next, the contents of the processing in the required accuracy determination section 120 will be described. Figures 4-7 The required accuracy determination section 120 determines the position and posture estimation accuracy (required accuracy) required for the self-position estimation at the next travel, based on the output of the sensor 200 and the map 310 generated by the map generation section 110.

[0054] For example, the required accuracy determination section 120 determines the required accuracy at each time point of the travel position 313. Further, for example, the required accuracy determination section 120 determines the required accuracy with two values of high and low, which are the position and posture estimation accuracy required for the self-position estimation at the next travel. Furthermore, the timing of the determination of the required accuracy and the determination reference are not limited to this, and the required accuracy can be determined with more than two values, as long as the high and low of the position and posture estimation accuracy can be discriminated.

[0055] Figure 4 A diagram for illustrating an example of the required accuracy determination using the point group 311 and the point group 312. Figure 4 In the example, the travel position 313 is determined by the required accuracy determination section 120 as a travel position 321 (solid line portion of the travel position 313) in which the required accuracy is high and a travel position 322 (dotted line portion of the travel position 313) in which the required accuracy is low.

[0056] In the required accuracy determination using the point group 311, the point group 312, the distance 331 from the traveling position 314 at each time to the point group on the left side of the vehicle 300 and the distance 332 to the point group on the right side of the vehicle 300 are used to determine the required accuracy. The distance 331 from the traveling position 314 at each time to the point group on the left side of the vehicle 300 and the distance 332 to the point group on the right side of the vehicle 300 are distances from the traveling position 314 at each time to the point closest to the vehicle 300 among the points included in the exploration range 333, 334 set in advance for the vehicle 300. In a case where the sum of the distances 331, 332 is smaller than a threshold value set in advance, it is determined that the required accuracy is high, and in other cases, it is determined that the required accuracy is low. That is, in a case where the distance to the obstacle is short, it is determined that high-accuracy self-position estimation and control are required to implement the driving assistance, the automatic driving using the map generation-self-position estimation device 100 at the next traveling.

[0057] Figure 5 A figure is shown to represent an example of the required accuracy determination using the detection result of the object. In the required accuracy determination using the detection result of the object, the position of the object detected by the first external environment sensor 210, the second external environment sensor 220, or another sensor is used to determine the required accuracy. Figure 5 A case where the temporary stop line 340 is used as the object is shown in the figure. In a case where the distance 341 from the traveling position 314 at each time to the temporary stop line 340 is smaller than a threshold value set in advance, it is determined that the traveling position 321 (the solid line portion of the traveling position 313) is a high-accuracy traveling position, and in other cases, it is determined that the traveling position 322 (the broken line portion of the traveling position 313) is a low-accuracy traveling position. Here, the object is not limited to the temporary stop line 340, and an object such as a traffic signal, a sign, a road marking, an intersection, or the like can be used. That is, in a case where the distance to an object important in a control set in advance is short, it is determined that high-accuracy self-position estimation and control are required to implement the driving assistance, the automatic driving using the map generation-self-position estimation device 100 at the next traveling.

[0058] Figure 6 A figure is shown to represent an example of the required accuracy determination using the steering angular velocity as the vehicle information. In the required accuracy determination using the vehicle information, the vehicle information acquired from the vehicle information sensor 240 is used to determine the required accuracy. In a case where the steering angular velocity is large, it is determined that the required accuracy is high, and in other cases, it is determined that the required accuracy is low. That is, in a case where the steering angular velocity is large, it is determined that high-accuracy self-position estimation and control are required to implement the driving assistance, the automatic driving using the map generation-self-position estimation device 100 at the next traveling. Figure 6In the example shown using the steering angular velocity, first, the position 350 where the steering angular velocity becomes greater than a preset threshold value is detected. Next, in the case where the distance 351 from the travel position 314 at each time to the position 350 is smaller than a preset threshold value, the travel position 321 requiring high accuracy (solid line portion of the travel position 313) is determined, and in the case other than this, the travel position 322 requiring low accuracy (dotted line portion of the travel position 313) is determined. Further, in the case where there are a plurality of positions 350 (two in this example) where the steering angular velocity becomes greater than the preset threshold value, the distance 351 is calculated from the travel position 314 at each time and the closest position 350. Here, the vehicle information is not limited to the steering angular velocity, and the braking amount, the speed, the acceleration, the angular velocity, the angular acceleration, the gear change, the turning light illumination, the hazard warning light illumination of the vehicle can also be used as the vehicle information. That is, at a place where the driving action of the driver in the travel or the behavior change of the vehicle following the driving action is large, it is determined that high-accuracy self-position estimation and control are required for implementing the driving assistance using the map generation-self-position estimation device 100 at the next travel. Figure 6

[0059] Figure 7 An example of the request accuracy determination using the driving assistance system action as the vehicle information is shown in the figure. In the case where the lane keeping system, which is one of the driving assistance systems, is functioning, Figure 7 The case where the lane keeping system is turned off (also called canceled) at the position 371 because the lane marking 370 is not recognized due to a whiteout or the like, and the case where the lane keeping system is turned on (also called restored) at the position 372 because the lane marking 302 is recognized are shown. In the case where the travel position 314 at each time is between the position 371 and the position 372, the travel position 321 requiring high accuracy (solid line portion of the travel position 313) is determined. Further, in the case where the distance 373 from the travel position 314 at each time to the closer one of the position 371 and the position 372 is smaller than a preset threshold value, the travel position 321 requiring high accuracy (solid line portion of the travel position 313) is determined. In the case other than this, the travel position 322 requiring low accuracy (dotted line portion of the travel position 313) is determined. That is, in the section where the driving assistance system is turned off in the travel, it is determined that high-accuracy self-position estimation and control are required for implementing the driving assistance using the map generation-self-position estimation device 100 at the next travel.

[0060] ​Here, regarding the thresholds for each distance, different thresholds can be used for each object or each vehicle information. Furthermore, the thresholds can be changed based on whether the driving position 314 at each time point is closer or farther than the positions detected by each object or vehicle information. Additionally, time can be used instead of position, or time can be used in addition to position.

[0061] The accuracy determination unit 120 can also combine multiple determination indicators as described above to determine the final required accuracy. For example, if there are N or more indicators that determine the driving position 321 with high required accuracy, then the driving position 321 with high required accuracy can be determined as the driving position 321 with high required accuracy. Here, N is an integer greater than or equal to 1.

[0062] (Actions of the data selection department)

[0063] Next, use Figure 8 , Figure 9 The processing in the data selection unit 130 will be explained. The data selection unit 130 selects point groups included in the map based on the map 310 generated by the map generation unit 110 and the required accuracy. Here, in map generation mode, the required accuracy determined by the required accuracy determination unit 120 is used; in location inference mode, the required accuracy included in the recorded map recorded by the map recording unit 140 is used. Furthermore, the data selection unit 130 generates a selection map that includes the selected point groups (selected point groups), the driving position, and the required accuracy.

[0064] In this embodiment, the data selection unit 130 generates a selection map by selecting map data (point groups) from the map 310 generated by the map generation unit 110 that corresponds to the type of sensor or processing method used to measure the point groups, based on the required accuracy described above. In other words, the data selection unit 130 selects data (point groups) from the map 310 generated by the map generation unit 110 that corresponds to the type of sensor or processing method used to measure the point groups (along with the required accuracy at a specified location of the vehicle's driving position) to generate the selection map.

[0065] Figure 8 This is a diagram illustrating an example of the selection map 380 chosen by the data selection unit 130. Figure 8 The text demonstrates the use of... Figure 3 Map 310 and shown Figure 4The result obtained by selecting data using the indicated determination result of required accuracy (traveling position 321 where required accuracy is high, traveling position 322 where required accuracy is low) (selected map 380). In the case where the traveling position 314 at each time is the traveling position 321 where required accuracy is high, the data selection section 130 selects only the point group 311 acquired by the first external sensor 210 (a sensor with high measurement accuracy of three-dimensional position) at this position, and does not select the point group 312 acquired by the second external sensor 220 (a sensor with low measurement accuracy of three-dimensional position). In the case where the traveling position 314 at each time is the traveling position 322 where required accuracy is low, the data selection section 130 selects both the point group 311 acquired by the first external sensor 210 and the point group 312 acquired by the second external sensor 220 at this position. That is, the data selection section 130 uses only the point group (information) acquired from the external sensor with high measurement accuracy of three-dimensional position (compared to the traveling position where required accuracy is low on the map) at the traveling position where required accuracy is high on the map.

[0066] Further, in the case where required accuracy is included in the recorded map recorded in the map recording section 140 in the position estimation mode, the traveling position 314 of the map 310 generated in the travel of the own vehicle at each time is converted into the position attitude on the recorded map using the position attitude estimated by the own position estimation section 150, and the determination result of required accuracy for the traveling position on the nearest recorded map is used.

[0067] Further, the data selection section 130 can also use a known sampling method such as grid sampling. Figure 9Fig. 6 is a diagram for illustrating an example of grid sampling. In grid sampling, a grid of a predetermined size is set, and only one point in each grid is selected, thereby reducing the number of points. Further, the mean or median of the points in each grid can be selected as a point. Here, the data selection section 130 performs sampling of the point group corresponding to the required precision. That is, the data selection section 130 sets a grid 385 for the point group measured on the travel location 321 for which the required precision is high and a grid 386 for the point group measured on the travel location 322 for which the required precision is low, and makes the size of each grid of the grid 386 larger than the size of each grid of the grid 385. Further, the data selection section 130 dynamically determines the size of the grid in accordance with the capacity (that is, the maximum capacity of the recorded map that is set in advance) available for recording the map, and performs sampling of the point group. For example, first, the size of the grid 385 and the grid 386 is set to the same value. Then, the size of the grid 386 is gradually increased, and the size at which the capacity of the map becomes equal to or less than the maximum capacity of the map set in advance is adopted. Thus, it is possible to give priority to retaining the points measured on the travel location 321 for which the required precision is high (in other words, to reduce the points measured on the travel location 322 for which the required precision is low) while controlling the capacity of the map to be equal to or less than the maximum capacity determined by the constraints of the hardware and the system.

[0068] (Action of the map recording section)

[0069] The map recording section 140 records a selection map including the point group selected by the data selection section 130 (selected point group), travel location, and required precision as a recorded map. Further, the map generated by the map generation section 110, that is, the map constituted by the travel location (travel path) of the own vehicle and the measurement result of the objects around the own vehicle, that is, the point group (for example, bound to the required precision) can be recorded as a recorded map in place of the selection map or in addition to the selection map.

[0070] (Action of the own position estimation section)

[0071] Next, the contents of the processing in the own position estimation section 150 will be described. Figure 10 The own position estimation section 150 correlates (compares) the selection map 380 (recorded map) recorded in the map recording section 140 with the selection map 380 (current map) generated on the basis of the output of the current (traveling) sensor 200, and thereby estimates the position and posture of the current (traveling) vehicle on the recorded map.

[0072] Figure 10 Fig. 8 is a diagram for illustrating an example of the position and posture of the own vehicle estimated by the own position estimation section 150. Figure 10In this case, the point group 311, the point group 312, the travel position 313, and the determination result of the required precision (the travel position 321 in which the required precision is high, the travel position 322 in which the required precision is low) indicate data of the recorded map. The self-position estimation unit 150 estimates the position and posture 390 of the current self-vehicle 300 and the travel position (travel path) 391 on these recorded maps.

[0073] The self-position estimation unit 150 estimates the position and posture between the recorded map and the current map by associating the point group 311 and the point group 312 included in each map. The estimation of the position and posture based on the association of the point groups uses a known ICP (Iterative Closest Point) method or an NDT (Normal Distributions Transform) method. When the position and posture between the recorded map and the current map is used, the position and posture on one map can be converted into the position and posture on another map. Thus, the position and posture 390 of the current self-vehicle 300 and the travel position (travel path) 391 on the recorded map can be estimated by converting the travel position 313 included in the current map into the position and posture on the recorded map using the position and posture between the recorded map and the current map.

[0074] In the ICP method and the NDT method, in a case where the number of input points is small or points are concentrated in a part of the space, the estimation of the position and posture can fail (error is extremely large). On the other hand, if points of low precision are input, the precision of the position estimation decreases. The self-position estimation unit 150 uses both the point group 311 acquired by the first external sensor 210 (a sensor of which the measurement precision of the three-dimensional position is high) and the point group 312 acquired by the second external sensor 220 (a sensor of which the measurement precision of the three-dimensional position is low) on the travel position 322 in which the required precision of the position estimation is low, and thus the number of points increases and the distribution of the points in the space is wide, so the position estimation can be performed without failing. Further, only the point group 311 acquired by the high-precision first external sensor 210 is used on the travel position 321 in which the required precision of the position estimation is high, so high-precision position estimation can be performed.

[0075] Further, in the ICP method and the NDT method, high-precision estimation can be achieved when the characteristics of the two input point groups are similar. The current map used in the self-position estimation unit 150 includes the point group selected by the data selection unit 130 according to the required precision included in the recorded map. That is, the criterion of the data selection of the map performed by the data selection unit 130 in the position estimation mode is consistent between the current map and the recorded map, so the self-position estimation unit 150 can estimate the position and posture with high precision.

[0076] (EFFECTS)

[0077] According to the above-described first embodiment, the following advantageous effects are obtained.

[0078] (1) Map generation - The map generation device 100 is provided with a map generation section 110, a required accuracy determination section 120, a data selection section 130, a map recording section 140, and a self-position estimation section 150. The map generation section 110 generates a map composed of point groups and travel positions of the own vehicle based on the output of the sensor 200. The required accuracy determination section 120 determines the position and posture estimation accuracy required for self-position estimation at the next travel (required accuracy) based on the output of the sensor 200 and the map generated by the map generation section 110. The data selection section 130 selects point groups included in the map based on the map generated by the map generation section 110 and the required accuracy determined by the required accuracy determination section 120. The map recording section 140 records a selection map including the point groups selected by the data selection section 130, the travel positions, and the required accuracy as a recorded map. The self-position estimation section 150 correlates the current map composed of the point groups selected by the data selection section 130 and the travel positions of the own vehicle with the recorded map recorded in the map recording section 140, thereby estimating the position and posture of the current vehicle on the recorded map. Figure 1 The data selection section 130 selects only point groups acquired from the external sensors having high measurement accuracy of three-dimensional positions at travel positions where the required accuracy is high, and selects both point groups acquired from the external sensors having high measurement accuracy of three-dimensional positions and point groups acquired from the external sensors having low measurement accuracy of three-dimensional positions at travel positions where the required accuracy is low, thereby selecting point groups acquired from more external sensors than at travel positions where the required accuracy is high. Figure 8 Thus, only (preferentially) high-accuracy point groups are used at travel positions where the required accuracy of position estimation is high, thereby enabling high-accuracy position estimation. Further, a large number of point groups are used at travel positions where the required accuracy of position estimation is low, thereby enabling position estimation over a wide range.

[0079] (2) The map recording section 140 records the point groups selected by the data selection section 130 (selection map 380). Thus, the capacity (amount of data) of the recorded map can be reduced.

[0080] (3) In the map generation mode, the data selection section 130 selects data (point groups) from the map using the required accuracy determined by the required accuracy determination section 120. In the position estimation mode, the data selection section 130 selects data (point groups) from the map generated from the travel of the own vehicle using the required accuracy included in the recorded map recorded in the map recording section 140. Figure 1 Thus, the recorded map used by the self-position estimation section 150 in the position estimation mode coincides with the criterion for data selection of the current map, so the position and posture can be estimated with high accuracy.

[0081] (4) The data selection unit 130 reduces the number of points by sampling a group of points corresponding to the required accuracy. Figure 9 Therefore, it is possible to reduce the size of the recorded map while prioritizing the retention of the point group measured at the driving position 321 where high accuracy is required.

[0082] (5) The data selection unit 130 reduces the number of points by sampling point groups corresponding to the preset maximum capacity of the recording map. Therefore, the map capacity can be controlled below the maximum capacity determined by hardware and system constraints.

[0083] (6) Map Generation - Self-Position Inference Device 100: In map generation mode, the accuracy requirement determination unit 120, data selection unit 130, and map recording unit 140 operate offline only once when the map generation mode ends. The map recording unit 140 records the selected map as the recorded map. Therefore, the computational cost of online processing that requires periodic processing can be reduced. Furthermore, since the accuracy requirement determination unit 120 uses the entire map, the accuracy requirement at each time point can be determined using data from the sensor 200 at times prior to each time point. Therefore, the accuracy requirement can be determined with high precision.

[0084] (7) When the distance from the vehicle's own driving position to the point group is close, the accuracy determination unit 120 determines that high-precision self-position inference and control are required. Figure 4 Therefore, it can accurately infer position and attitude in environments with narrow roads and equidistant obstacles.

[0085] (8) When the accuracy determination unit 120 is close to a preset target (the target detected by sensor 200) at its own driving position, it determines that high-precision self-position inference and control are required. Figure 5 Therefore, it is possible to accurately infer the position and attitude of objects such as those stopped at temporary stop lines or traffic lights.

[0086] (9) The accuracy determination unit 120 requires the use of vehicle information such as steering angle velocity (corresponding to the driver's driving operation amount), gear position, and the operating status of the driving assistance system that assists the driver's driving operation to determine the required accuracy. Figure 6 , Figure 7 Therefore, it can accurately infer position and attitude in situations such as turning, reversing, and the actions of driver assistance systems.

[0087] As described above, the map generation-self position estimation device 100 of the first embodiment described above includes: a map generation section 110 that generates a map composed of a travel position (travel path) of a host vehicle and a point cloud of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; a required accuracy determination section 120 that determines a required accuracy of estimation of a host vehicle position (host position) on a prescribed place of the travel position of the host vehicle (at the next travel) based on the information acquired by the sensors and the map; a data selection section 130 that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point cloud from the map according to the required accuracy; a map recording section 140 that records the map as a recorded map; and a host position estimation section 150 that estimates a current host vehicle position in the recorded map by comparison between the selection map generated in travel of the host vehicle and the recorded map.

[0088] The data selection section 130 preferentially selects information acquired by a sensor having a high position measurement accuracy among the plurality of sensors to make a selection map on the travel position where the required accuracy is high, compared to the travel position where the required accuracy is low.

[0089] The data selection section 130 selects information acquired by a sensor having a high position measurement accuracy among the plurality of sensors to make a selection map on the travel position where the required accuracy is high, and selects both information acquired by a sensor having a high position measurement accuracy and information acquired by a sensor having a low position measurement accuracy among the plurality of sensors to make a selection map on the travel position where the required accuracy is low.

[0090] Further, the map recording section 140 also records the required accuracy with respect to the recorded map (selection map), and the data selection section 130 uses the required accuracy included in the recorded map to perform data selection with respect to the map generated in travel of the host vehicle.

[0091] More specifically, the map generation-self position estimation device 100 of the above-described first embodiment includes, in the map generation mode, the map generation section 110 that generates a map constituted by a travel position (travel path) of the own vehicle and a point group that is a measurement result of an object in the vicinity of the own vehicle, based on information acquired by a plurality of sensors different in characteristics, the required accuracy determination section 120 that determines a required accuracy of self vehicle position (self position) estimation on a prescribed place on the travel position of the own vehicle (at the next travel) based on the information acquired by the sensors and the map, the data selection section 130 that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy determined by the required accuracy determination section 120, and the map recording section 140 that records the selection map as a recorded map, and includes, in the position estimation mode, the map generation section 110 that generates a map constituted by a travel position (travel path) of the own vehicle and a point group that is a measurement result of an object in the vicinity of the own vehicle, based on information acquired by a plurality of sensors different in characteristics, the data selection section 130 that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy included in the selection map (recorded map) recorded in the map recording section 140 (in the travel of the own vehicle), and the self position estimation section 150 that estimates a current own vehicle position in the recorded map by comparison between the selection map (current map) generated in the travel of the own vehicle and the selection map (recorded map) recorded in the map recording section 140.

[0092] For example, in automatic driving or driving assistance that requires actions in a wider range and various conditions compared to automatic parking, it is necessary to take into account both improvement of self position estimation accuracy and expansion of a self position estimable range, and according to the above-described first embodiment, it is possible to realize the map generation-self position estimation device 100 capable of estimating a self vehicle position and posture with high accuracy.

[0093] [Second Embodiment]

[0094] Hereinafter, the above-described first embodiment will be described with reference to Figure 11A second embodiment of the map generation-self-position inference device will be described. In the following description, the same symbols are used for components identical to those in the first embodiment, and the differences are mainly explained. Items not specifically described are the same as in the first embodiment. This embodiment focuses on a map generation-self-position inference device that, in addition to the map generation-self-position inference device 100 of the first embodiment, also includes a data temporary storage unit and an offline map generation unit. The data temporary storage unit temporarily stores sensor data, and the offline map generation unit generates a map through offline processing based on the data stored in the data temporary storage unit.

[0095] (Block Composition)

[0096] Figure 11 This is a diagram illustrating the block structure of the map generation-self-position inference device 500 according to the second embodiment. The map generation-self-position inference device 500 includes a map generation unit 110, a required accuracy determination unit 120, a data selection unit 130, a map recording unit 140, a self-position inference unit 150, a data temporary storage unit 560, and an offline map generation unit 570. The data temporary storage unit 560 temporarily stores data from the sensor 200 (online processing). The offline map generation unit 570 generates a map through offline processing based on the data stored in the data temporary storage unit 560.

[0097] (Actions of the temporary data storage department)

[0098] The processing in the temporary data storage unit 560 will be explained. The temporary data storage unit 560 temporarily stores the data from the sensor 200.

[0099] The data temporary storage unit 560 operates in map generation mode, temporarily storing data from the sensor 200 acquired at the driving position 313 of the driving position 321, which is determined by the accuracy requirement determination unit 120 to require high accuracy. Here, the data from the first external sensor 210 and the second external sensor 220 are not limited to point groups, but can also be the outputs of each sensor before processing. For example, when using motion stereoscopic views of the image from a panoramic camera as the second external sensor 220, the data temporary storage unit 560 can store the panoramic camera image in addition to the point groups obtained through the motion stereoscopic views. The data stored in the data temporary storage unit 560 is maintained until the operation of the offline map generation unit 570 is completed.

[0100] (Actions of the offline map generation department)

[0101] Next, the processing in the offline map generation unit 570 will be explained. The offline map generation unit 570 generates a map through offline processing based on the data stored in the data temporary storage unit 560.

[0102] The offline map generation section 570 acts after determining that the travel in the map generation mode has ended, and generates a map by offline processing based on the data of the sensor 200 at the travel position 321 for which high precision is required, which is held in the data temporary holding section 560. Since the offline map generation section 570 generates a map by offline processing, a map with higher precision than that of the map generation section 110 can be generated. For example, in a case where an image is held in the data temporary holding section 560, a known SfM (Structure from Motion) method of inferring the position and posture of a camera and the three-dimensional positions of feature points from the image can be used. By using the SfM method, a travel position and a point group with higher precision than those inferred by wheel odometry and those inferred by motion stereo can be obtained.

[0103] Further, the offline map generation section 570 is algorithmically premised on online processing, but can generate a map by an algorithm that cannot be run in the map generation section 110 due to insufficient processing time. For example, a travel position can be inferred by wheel odometry that can be processed at high speed in the map generation section 110, and an image odometry of inferring a travel position from an image or a Visual SLAM (Simultaneous Localization and Mapping) method of inferring a travel position and a point group from an image can be used in the offline map generation section 570.

[0104] Here, the map generated by the map generation section 110 and the map generated by the offline map generation section 570 are integrated into one map and recorded to the map recording section 140. Specifically, a travel position after a start point of a travel position for which high precision is required is calculated using the relative position and posture from the start point inferred by the offline map generation section 570. Similarly, a travel position after an end point of a travel position for which high precision is required is calculated using the relative position and posture from the end point inferred by the relative position sensor 230. Thus, the results of the inference of two different relative position and postures can be linked to obtain a travel position in one coordinate system.

[0105] Further, the action of the offline map generation section 570 can be after the ignition is turned off, or after the driver gets out of the vehicle. Further, the offline map generation section 570 does not need to act continuously, and can implement processing gradually during a period when the load of other processing is low.

[0106] (EFFECTS)

[0107] According to the above-described second embodiment, the following effects are obtained.

[0108] (1) The offline map generation section 570 acts when it is determined that the travel in the map generation mode has ended, and generates a map (a map composed of the travel position of the own vehicle and the measurement results of the objects in the vicinity of the own vehicle, i.e., a point group) by offline processing based on the data of the sensor 200 held in the data temporary holding section 560. Thus, a high-precision map can be obtained by generating a map by offline processing. Further, by using a high-precision map, the position and posture can be accurately estimated in the position estimation mode.

[0109] (2) The data temporary holding section 560 acts in the map generation mode, and temporarily holds the data of the sensor 200 acquired at the travel position of the travel position 321 for which it is determined in the required accuracy determination section 120 that the required accuracy is high. Thus, only the data of the sensor 200 acquired at the travel position 321 for which the required accuracy is high is held, and the data of the sensor 200 acquired at the travel position 322 for which the required accuracy is low is not held, so the capacity required for temporary holding of data can be reduced.

[0110] For example, in automatic driving or driving assistance that requires action in a wider range and under various conditions compared to automatic parking, both improvement of the estimation accuracy of the own position and expansion of the range in which the own position can be estimated are required, and as described above, according to the above-described second embodiment, as with the first embodiment, the map generation-own position estimation device 500 that can accurately estimate the position and posture of the own vehicle can be implemented.

[0111] Further, the present application includes various modifications, and is not limited to the above-described embodiments. For example, the above-described embodiments are detailed descriptions made in order to make the present application easy to understand, and are not necessarily limited to all the configurations described. Other modes thought within the scope of the technical idea of the present application are also included in the scope of the present application. Further, a part of the configuration of an embodiment can be replaced with the configuration of another embodiment, and further, the configuration of an embodiment can be added with the configuration of another embodiment. Further, a part of the configuration of each embodiment can be added, deleted, or replaced with another configuration. Further, a part or all of each configuration, function, processing section, processing unit, and the like described above can be implemented in hardware, for example, by designing using an integrated circuit. Further, each configuration, function, and the like described above can be implemented in software by a processor interpreting and executing a program that implements each function. The program, table, file, and the like that implement each function can be placed in a storage device such as a memory, a hard disk, an SSD (Solid State Drive), or an IC card, an SD card, a DVD, or the like.

[0112] Explanation of Symbols

[0113] 100 … map generation-self position estimation device (1st embodiment), 110 … map generation unit, 120 … required accuracy determination unit, 130 … data selection unit, 140 … map recording unit, 150 … self position estimation unit, 200 … sensor group, 210 … 1st external environment sensor, 220 … 2nd external environment sensor, 230 … relative position sensor, 240 … vehicle information sensor, 310 … map, 380 … selected map, 500 … map generation-self position estimation device (2nd embodiment), 560 … data temporary storage unit, 570 … offline map generation unit.

Claims

1. A map generation-self position inference apparatus characterized by comprising: Possessing: a map generation section that generates a map composed of a travel position of a host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; a required accuracy determination section that determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; a data selection section that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy; a map recording section that records the map as a recorded map; and a host position estimation section that estimates a current host vehicle position in the recorded map by comparing the selection map generated in travel of the host vehicle with the recorded map, the data selection section preferentially selects information acquired by a sensor having a high position measurement accuracy among the plurality of sensors to make a selection map on a travel position where the required accuracy is high, compared to a travel position where the required accuracy is low, the data selection section selects information acquired by a sensor having a high position measurement accuracy among the plurality of sensors to make a selection map on a travel position where the required accuracy is high, and selects both information acquired by a sensor having a high position measurement accuracy and information acquired by a sensor having a low position measurement accuracy among the plurality of sensors to make a selection map on a travel position where the required accuracy is low.

2. The map generation-host position estimation device according to claim 1, characterized in that, in a map generation mode, the map generation section generates a map composed of a travel position of a host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; the required accuracy determination section determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy determined by the required accuracy determination section; and the map recording section records the selection map as a recorded map, in a position estimation mode, the map generation section generates a map composed of a travel position of a host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy included in the selection map recorded in the map recording section; and the host position estimation section estimates a current host vehicle position in the recorded map by comparing the selection map generated in travel of the host vehicle with the selection map recorded in the map recording section. Possessing: ​ ​ 3. A map generation-self position inference apparatus characterized by comprising: ​ a map generation unit that generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; a required accuracy determination unit that determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; a data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy; a map recording unit that records the map as a recorded map; and a host position estimation unit that estimates a current host vehicle position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the recorded map, the map recording unit records the selection map instead of the map as the recorded map, the map recording unit further records the required accuracy for the recorded map, the data selection unit uses the required accuracy included in the recorded map to perform data selection for the map generated in the travel of the host vehicle.

4. The map generation-host position estimation device according to claim 3, wherein in a map generation mode, the map generation unit generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; the required accuracy determination unit determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; the data selection unit generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy determined by the required accuracy determination unit; and the map recording unit records the selection map as a recorded map; in a position estimation mode, the map generation unit generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; the data selection unit generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy included in the selection map recorded in the map recording unit; and the host position estimation unit estimates a current host vehicle position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the selection map recorded in the map recording unit. provided with: a map generation unit that generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; ​ 5. A map generation-self position inference apparatus characterized by comprising: ​ ​ a required accuracy determination unit that determines a required accuracy of self-vehicle position estimation on a prescribed location on a travel position of the self-vehicle, based on information acquired by the sensors and the map; a data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group, from the map, according to the required accuracy; a map recording unit that records the map as a recording map; and a self-position estimation unit that estimates a current self-vehicle position in the recording map, by comparing the selection map generated in the travel of the self-vehicle with the recording map, the data selection unit sets a first grid for the point group measured on the travel position for which the required accuracy is high and a second grid for the point group measured on the travel position for which the required accuracy is low, respectively, and makes the size of the second grid larger than the size of the first grid, and performs sampling of the point group corresponding to the required accuracy, the data selection unit first sets the sizes of the first grid and the second grid to the same value, then gradually increases the size of the second grid, adopts the size at which the capacity of the recording map becomes below a maximum capacity of the recording map set in advance, and dynamically determines the sizes of the first grid and the second grid according to the maximum capacity of the recording map set in advance, and performs sampling of the point group corresponding to the maximum capacity of the recording map set in advance.

6. The map generation-self-position estimation device according to claim 5, characterized in that, in a map generation mode, it includes: the map generation unit that generates a map constituted by a travel position of the self-vehicle and a measurement result of an object around the self-vehicle, i.e., a point group, from information acquired by a plurality of sensors different in characteristics; the required accuracy determination unit that determines a required accuracy of self-vehicle position estimation on a prescribed location on a travel position of the self-vehicle, based on information acquired by the sensors and the map; the data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group, from the map, according to the required accuracy determined by the required accuracy determination unit; and the map recording unit that records the selection map as a recording map; in a position estimation mode, it includes: the map generation unit that generates a map constituted by a travel position of the self-vehicle and a measurement result of an object around the self-vehicle, i.e., a point group, from information acquired by a plurality of sensors different in characteristics; the data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group, from the map, according to the required accuracy included in the selection map recorded in the map recording unit; and the self-position estimation unit that estimates a current self-vehicle position in the recording map, by comparing the selection map generated in the travel of the self-vehicle with the selection map recorded in the map recording unit. it has:

7. A map generation-self position inference apparatus characterized by comprising: ​ a map generation unit that generates a map composed of a travel position of the host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; a required accuracy determination unit that determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; a data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy; a map recording unit that records the map as a recorded map; and a host position estimation unit that estimates a current host vehicle position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the recorded map, in a map generation mode, causing the required accuracy determination unit, the data selection unit, and the map recording unit to act offline only once at the end of the map generation mode, and recording the selection map as the recorded map in the map recording unit.

8. The map generation-host position estimation device according to claim 7, characterized in that, in a map generation mode, includes: the map generation unit that generates a map composed of a travel position of the host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; the required accuracy determination unit that determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; the data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy determined by the required accuracy determination unit; and the map recording unit that records the selection map as a recorded map; in a position estimation mode, includes: the map generation unit that generates a map composed of a travel position of the host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; the data selection unit that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensors that measured the point group from the map according to the required accuracy included in the selection map recorded in the map recording unit; and the host position estimation unit that estimates a current host vehicle position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the selection map recorded in the map recording unit. includes:

9. A map generation-self position inference apparatus characterized by comprising: a map generation unit that generates a map composed of a travel position of the host vehicle and a point group that is a measurement result of an object around the host vehicle, based on information acquired by a plurality of sensors different in characteristics; a required accuracy determination unit that determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensors and the map; ​ a data selection section that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy; a map recording section that records the map as a recorded map; and a self-position estimation section that estimates a current self-vehicle position in the recorded map by comparing the selection map generated in the travel of the self-vehicle with the recorded map, the required accuracy determination section determines the required accuracy according to a distance from a travel position of the self-vehicle to a point group.

10. The map generation-self-position estimation device according to claim 9, wherein in a map generation mode, the map generation section generates a map constituted by a travel position of a self-vehicle and a measurement result of an object around the self-vehicle, i.e., a point group, from information acquired by a plurality of sensors different in characteristics; the required accuracy determination section determines a required accuracy of self-vehicle position estimation on a prescribed place of the travel position of the self-vehicle based on the information acquired by the sensors and the map; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy determined by the required accuracy determination section; and the map recording section records the selection map as a recorded map; in a position estimation mode, the map generation section generates a map constituted by a travel position of a self-vehicle and a measurement result of an object around the self-vehicle, i.e., a point group, from information acquired by a plurality of sensors different in characteristics; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy included in the selection map recorded in the map recording section; and the self-position estimation section estimates a current self-vehicle position in the recorded map by comparing the selection map generated in the travel of the self-vehicle with the selection map recorded in the map recording section. provided with a map generation section that generates a map constituted by a travel position of a self-vehicle and a measurement result of an object around the self-vehicle, i.e., a point group, from information acquired by a plurality of sensors different in characteristics; a required accuracy determination section that determines a required accuracy of self-vehicle position estimation on a prescribed place of the travel position of the self-vehicle based on the information acquired by the sensors and the map; 11. A map generation-self position inference apparatus characterized by comprising: a data selection section that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy; a map recording section that records the map as a recorded map; and a self-position estimation section that estimates a current self-vehicle position in the recorded map by comparing the selection map generated in the travel of the self-vehicle with the recorded map, ​ ​ ​ ​ The required accuracy determination section determines the required accuracy based on a distance from a travel position of the host vehicle to a target detected by the sensor.

12. The map generation-host position estimation device according to claim 11, characterized in that, in the map generation mode includes: the map generation section generates a map composed of a travel position of the host vehicle and a point cloud of measurement results of objects around the host vehicle based on information acquired by a plurality of sensors different in characteristics; the required accuracy determination section determines a required accuracy of host position estimation at a prescribed position of the travel position of the host vehicle based on the information acquired by the sensors and the map; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point cloud from the map according to the required accuracy determined by the required accuracy determination section; and the map recording section records the selection map as a recorded map; in the position estimation mode includes: the map generation section generates a map composed of a travel position of the host vehicle and a point cloud of measurement results of objects around the host vehicle based on information acquired by a plurality of sensors different in characteristics; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point cloud from the map according to the required accuracy included in the selection map recorded in the map recording section; and the host position estimation section estimates a current host position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the selection map recorded in the map recording section.

13. A map generation-self position inference apparatus characterized by comprising: provided with: a map generation section that generates a map composed of a travel position of the host vehicle and a point cloud of measurement results of objects around the host vehicle based on information acquired by a plurality of sensors different in characteristics; a required accuracy determination section that determines a required accuracy of host position estimation at a prescribed position of the travel position of the host vehicle based on the information acquired by the sensors and the map; a data selection section that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point cloud from the map according to the required accuracy; a map recording section that records the map as a recorded map; and a host position estimation section that estimates a current host position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the recorded map, the required accuracy determination section determines the required accuracy based on vehicle information including at least one of an amount of a driving operation of a driver, a gear position, and an operation state of a driving assist system that assists the driving operation of the driver.

14. The map generation-host position estimation device according to claim 13, characterized in that, in the map generation mode includes: ​ the map generation section generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; the required accuracy determination section determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensor and the map; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy determined by the required accuracy determination section; and the map recording section records the selection map as a recorded map; in a position estimation mode, the map generation section generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; the data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy included in the selection map recorded in the map recording section; and the host position estimation section estimates a current host vehicle position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the selection map recorded in the map recording section. provided with:

15. A map generation-self position inference apparatus characterized by comprising: a map generation section that generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; a required accuracy determination section that determines a required accuracy of host vehicle position estimation on a prescribed place of the travel position of the host vehicle, based on the information acquired by the sensor and the map; a data selection section that generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map according to the required accuracy; a map recording section that records the map as a recorded map; a host position estimation section that estimates a current host vehicle position in the recorded map by comparing the selection map generated in the travel of the host vehicle with the recorded map; a data temporary storage section that temporarily stores data acquired by the sensor; and an offline map generation section that generates a map by offline processing according to data stored in the data temporary storage section.

16. The map generation-host position estimation device according to claim 15, wherein the data temporary storage section stores only data acquired on a travel position of the travel position of the host vehicle on which the required accuracy is high.

17. The map generation-host position estimation device according to claim 15 or 16, wherein in a map generation mode, the map generation section generates a map composed of a travel position of the host vehicle and a point group of measurement results of objects around the host vehicle, from information acquired by a plurality of sensors different in characteristics; ​ ​ ​ The required accuracy determination section determines a required accuracy of self-vehicle position estimation on a prescribed location of a travel position of the self-vehicle, based on information acquired by the sensor and the map; The data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map, according to the required accuracy determined by the required accuracy determination section; and The map recording section records the selection map as a recorded map; In a position estimation mode, the map generation section generates a map composed of a travel position of a self-vehicle and a measurement result of an object in the vicinity of the self-vehicle, i.e., a point group, from information acquired by a plurality of sensors having different characteristics; The data selection section generates a selection map by selecting data of the map corresponding to a kind or a processing method of the sensor that measured the point group from the map, according to the required accuracy included in the selection map recorded in the map recording section; and The self-position estimation section estimates a current self-vehicle position in the recorded map by comparing the selection map generated in the travel of the self-vehicle with the selection map recorded in the map recording section. ​

Images