JP2018084960A - Self-position estimation method and self-position estimation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve estimation accuracy of a self-position even if it is not possible to check all targets with map information without inconsistency.SOLUTION: A self-position estimation method of the present invention includes: moving a relative position between a target present around a vehicle and the vehicle by an amount of movement of the vehicle to store it as target position data; extracting a plurality of travel lane boundaries present right and left sides of the vehicle from the stored target position data; setting priority to each of the extracted travel lane boundaries to select a travel lane boundary with high priority; and checking target position data on the selected travel lane boundary with target position information included in map information to estimate a self-position of the vehicle.SELECTED DRAWING: Figure 3

Description

  The present invention relates to a self-position estimation method and a self-position estimation apparatus that estimate a self-position of a host vehicle.

  Conventionally, it is limited to environmental information existing in a predetermined area with a moving body as a reference, and the self-location is estimated by comparing this limited environmental information with a previously stored environmental map ( Patent Document 1).

JP 2008-250906 A

  However, in the conventional self-position estimation method described above, if the relative error between the paired targets such as white lines becomes large, not all targets can be matched with map information without contradiction. There was a problem that the estimation accuracy was lowered.

  The present invention has been made in view of the above problems, and its purpose is self-position estimation that can improve self-position estimation accuracy even when all targets cannot be compared with map information without contradiction. It is to provide a method and apparatus thereof.

  In order to solve the above-described problem, a self-position estimation method and apparatus according to an aspect of the present invention provides a relative position between a target existing around the host vehicle and the host vehicle by an amount of movement of the host vehicle. It is moved and stored as target position data. Then, a plurality of road boundaries existing on the left and right of the host vehicle are extracted from the accumulated target position data, a priority is set for each of the extracted plurality of road boundaries, and a road boundary with a high priority is selected. . From this result, the target position of the host vehicle is estimated by comparing the target position data of the selected road boundary with the target position information included in the map information.

  According to the present invention, the self-position estimation accuracy can be improved even when all targets cannot be compared with map information without contradiction.

FIG. 1 is a block diagram illustrating a configuration of a self-position estimation system including a self-position estimation apparatus according to an embodiment of the present invention. FIG. 2 is a diagram illustrating an example of a camera mounted on a vehicle. FIG. 3 is a flowchart showing a processing procedure of self-position estimation processing by the self-position estimation apparatus according to the embodiment of the present invention. FIG. 4 is a diagram for explaining a priority setting method by the self-position estimation apparatus according to the embodiment of the present invention. FIG. 5 is a diagram for explaining a priority setting method when there is an oncoming vehicle by the self-position estimating apparatus according to the embodiment of the present invention. FIG. 6 is a diagram for explaining a priority setting method when a parallel vehicle and an oncoming vehicle are present by the self-position estimating apparatus according to the embodiment of the present invention. FIG. 7 is a diagram for explaining a priority setting method when a pedestrian exists by the self-position estimation apparatus according to the embodiment of the present invention.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the description of the drawings, the same or similar parts are denoted by the same or similar reference numerals, and redundant description is omitted.

[Self-position estimation system]
FIG. 1 is a block diagram illustrating a configuration of a self-position estimation system including a self-position estimation apparatus according to the present embodiment. As shown in FIG. 1, a self-position estimation system 100 according to the present embodiment includes a surrounding sensor group 1, a self-position estimation device 3, a storage device 4, and a vehicle sensor group 5. The self-position estimation system 100 according to the present embodiment is mounted on the vehicle V (see FIG. 2) and estimates the self-position of the vehicle V.

  In this embodiment, as the self-position of the vehicle V to be estimated, a position in the east-west direction (X-axis direction) (X coordinate [m]), a position in the north-south direction (Y-axis direction) (Y coordinate [m]), A vehicle azimuth angle θ (yaw angle [rad]) is estimated as posture angle information. That is, the position and posture angle with a total of three degrees of freedom on the two-dimensional plane are estimated.

  The ambient sensor group 1 includes a plurality of cameras 11 and 13 that capture digital images that can be processed. However, a laser range finder may be further provided, and the ambient sensor group 1 only needs to be composed of a plurality of sensors for detecting a target existing around the vehicle V.

  FIG. 2 is a diagram illustrating an example of a state in which the surrounding sensor group 1 is mounted on the vehicle V. The cameras 11 and 13 can be mounted on door mirrors on both the left and right sides of the vehicle V, for example. The cameras 11 and 13 take images with a solid-state imaging device such as a CCD or CMOS, for example. The cameras 11 and 13, for example, photograph a road surface on the side of the vehicle V, and sequentially output the captured images to the self-position estimation device 3. The cameras 11 and 13 can also be applied in place of a camera that images the road surface in the front direction of the vehicle V or a camera that images the road surface in the rear direction of the vehicle V. In addition, a camera that captures a road surface on the side of the vehicle V, a camera that captures a road surface in the front direction of the vehicle V, and a camera that captures a road surface in the rear direction of the vehicle V can be applied simultaneously.

  The storage device 4 stores map information 41 including target position information of targets existing on the map. The storage device 4 can be composed of a semiconductor memory, a magnetic disk, or the like. The target recorded in the map information 41 is, for example, a road marking indicating a stop line, a pedestrian crossing, a pedestrian crossing notice, a lane marking, a structure such as a curb, and various other sensors that can be detected by the surrounding sensor group 1. Includes facilities. In the map information 41, position information such as curbstones and white lines is defined by a collection of straight line information having two-dimensional position information of both end points. The map information 41 is described as straight line information on a two-dimensional plane approximated by a broken line when the shape of the real environment is a curve.

  The vehicle sensor group 5 includes a GPS receiver 51, an accelerator sensor 52, a steering sensor 53, a brake sensor 54, a vehicle speed sensor 55, an acceleration sensor 56, a wheel speed sensor 57, and other sensors 58 such as a yaw rate sensor. Each of the sensors 51 to 58 is connected to the self-position estimation device 3 and sequentially outputs various detection results to the self-position estimation device 3.

  The self-position estimation device 3 is a controller that estimates target position that is the current position of the host vehicle by comparing target position data of targets existing around the host vehicle with target position information included in the map information. is there. The self-position estimation device 3 calculates the approximate position of the vehicle V in the map information 41 using each detection result of the vehicle sensor group 5, or calculates an odometry indicating the amount of movement of the vehicle V per unit time. The self-position estimation device 3 includes a target position detection unit 31, a movement amount estimation unit 32, a target position accumulation unit 33, a map information acquisition unit 34, a target selection unit 35, and a self-position estimation unit 36. Have The self-position estimation device 3 includes a general-purpose electronic circuit including a microcomputer, a microprocessor, and a CPU, and peripheral devices such as a memory. Then, by executing a specific program, the target position detection unit 31, the movement amount estimation unit 32, the target position accumulation unit 33, the map information acquisition unit 34, the target selection unit 35, and the self-position estimation unit 36 described above. Works as. Each function of such a self-position estimation apparatus 3 can be implemented by one or a plurality of processing circuits. The processing circuit includes a programmed processing device such as, for example, a processing device including an electrical circuit, and an application specific integrated circuit (ASIC) or conventional circuit arranged to perform the functions described in the embodiments. It also includes devices such as parts. The processing circuit may also be used as an electronic control unit (ECU) used for other controls of the vehicle V such as automatic driving control.

  The target position detection unit 31 is a target detection sensor that detects target information of a target existing in the vicinity of the host vehicle from images taken by the cameras 11 and 13. In particular, the target position detection unit 31 detects the relative position between the target and the vehicle V present around the vehicle V based on the imaging results of the cameras 11 and 13. Here, the target to be detected is a lane line such as a white line existing around the runway, or a runway boundary such as a curb or a median. For example, a coordinate position having a luminance change between pixels is detected as an edge point from images that are the imaging results of the cameras 11 and 13, and a road boundary such as a lane marking, a curbstone, or a median strip is detected as an edge point group. Further, instead of the cameras 11 and 13, a laser range finder, a laser radar, or the like can be used. It can detect similarly by detecting the coordinate position with a reflection intensity change as an edge point.

  The movement amount estimation unit 32 estimates the movement amount of the vehicle V per unit time based on the detection result of one of the sensors in the vehicle sensor group 5. The amount of movement of the host vehicle may be measured using odometry measurement based on the number of rotations of the tire, inertia measurement using a gyro and an acceleration sensor, or the like. Further, various methods such as a method of receiving radio waves from a satellite such as GNSS (Global Navigation Satellite System) or SLAM that estimates the amount of movement from a change in the measurement value of an external sensor may be used. However, in this embodiment, the movement amount estimation method is not particularly limited.

  The target position accumulation unit 33 moves the relative position detected by the target position detection unit 31 by the movement amount estimated by the movement amount estimation unit 32 and accumulates it as target position data. At this time, the time To when the relative position is detected is stored together with the relative position. Further, the target position accumulating unit 33 projects the target position data onto the vehicle coordinate system, and converts the detected relative position of the target into the position of the vehicle coordinate system. In the vehicle coordinate system, for example, the center of the rear wheel axle of the vehicle V is the origin, the forward direction is the positive direction of the x axis, the left direction is the positive direction of the y axis, and the upward direction is the positive direction of the z axis. A conversion formula from the coordinate system of the cameras 11 and 13 to the vehicle coordinate system is set in the target position storage unit 33 in advance. The same applies to road surface parameters in the vehicle coordinate system.

  The map information acquisition unit 34 acquires map information 41 around the host vehicle from the storage device 4 in order to estimate the own position of the host vehicle. In particular, the map information acquisition unit 34 acquires the position information of the host vehicle from the GPS receiver 51 and the like, and based on the acquired position information, the target position information of the target existing around the host vehicle from the map information 41. To get.

  The target selection unit 35 extracts a plurality of road boundaries existing on the left and right sides of the host vehicle from the target position data accumulated in the target position accumulation unit 33, and assigns priorities to each of the extracted plurality of road boundaries. Set and select a high priority track boundary. This priority is set according to the influence given to the own vehicle, and a higher value is set as the influence given to the own vehicle is larger. For example, the target selection unit 35 sets the higher priority as the relative position between the vehicle boundary and the object associated with the road boundary or the vehicle becomes closer. Further, the priority may be set higher as the height of the object associated with the road boundary or the road boundary becomes higher. Furthermore, the priority may be set higher as the relative speed between the object associated with the road boundary and the host vehicle becomes faster, or the priority may be set according to the attribute of the object associated with the road boundary. Also good. The target selection unit 35 sets the priority for each of the extracted plurality of lane boundaries using at least one of these priority setting methods. Since the plurality of runway boundaries are substantially parallel, if at least one runway boundary can be selected, the target position information of the target recorded in the map information 41 can be collated.

  The self-position estimating unit 36 collates the target position data selected by the target selecting unit 35 with the target position information of the target recorded in the map information 41 to estimate the self-position of the vehicle V. It is also possible to select only the road boundary with the highest priority and collate it with the target position information of the target recorded in the map information 41. In that case, the relative position of the road boundary having the highest priority with respect to the vehicle V's own position is substantially the same as the detected relative position, and the vehicle V's self relative to the road boundary having the highest priority. The position is most accurate.

[Self-position estimation method]
Next, with reference to the flowchart of FIG. 3, the procedure of the self-position estimation process performed by the self-position estimation apparatus according to the present embodiment will be described.

  First, in step S <b> 10, the outside world around the vehicle V is sensed by the cameras 11 and 13 of the ambient sensor group 1.

  In step S <b> 20, the target position detection unit 31 detects target information of a target that exists in the vicinity of the host vehicle from images captured by the cameras 11 and 13. In particular, the target position detection unit 31 detects a relative position between a target existing around the host vehicle and the host vehicle from images captured by the cameras 11 and 13. Specifically, the target position detection unit 31 acquires images captured by the cameras 11 and 13, detects lane boundaries such as white lines, curbstones, and runway boundaries such as median strips as line segments to determine relative positions. Ask. For example, white lines existing on the left and right of the vehicle V are detected from the luminance information of the images captured by the cameras 11 and 13. Note that a representative candidate point may be obtained for each predetermined distance section such as a white line, and then a line segment may be obtained from a group of representative candidate points. The group of representative candidate points may be detected at the same time or at different times. In this step, the detection range is limited to the vicinity of the host vehicle, and the reason is to detect the target position as accurate as possible. The vicinity of the own vehicle is a range in which white lines adjacent to the left and right of the own vehicle can be detected.

  In step S <b> 30, the movement amount estimation unit 32 first estimates the movement amount of the vehicle V based on the detection result of the vehicle sensor group 5. Then, the target position accumulation unit 33 moves the relative position detected in step S20 by the movement amount estimated by the movement amount estimation unit 32 and accumulates it as target position data. At this time, the time To when the relative position is detected is stored together with the relative position.

  In step S40, the target position storage unit 33 projects the target position data stored in step S30 onto a vehicle coordinate system centered on the vehicle position. The target position accumulation unit 33 converts the target position data accumulated in step S30 into the current vehicle coordinate system based on the position information Po of the vehicle V at the time To when the relative position of the target is detected. To do.

  In step S50, the target selection unit 35 extracts a plurality of runway boundaries existing on the left and right of the host vehicle from the target position data converted into the vehicle coordinate system in step S40, and each of the extracted plurality of runway boundaries. Set the priority to. The priority setting method will be described below.

  FIG. 4 shows a situation where the host vehicle is traveling on a road on which the curb and the white line are installed substantially in parallel. In this case, the curbstones A1 and A2 have a height while the white line B has no height, and therefore, when the host vehicle V contacts the curbstones A1 and A2, there is a greater influence. Therefore, a higher priority than the white line B is set for the curbs A1 and A2. For example, the priority weight of the curbs A1 and A2 is set to 3, and the priority weight of the white line B is set to 1. Further, when the road sign D1 exists in the vicinity of the curb A1, the road sign D1 is associated with the curb A1, and the height of the road sign D1 is high. Therefore, the priority weight of the curb A1 is a higher value. Set to 5. Thus, the target selection unit 35 sets the priority higher as the height of the road boundary or the object associated with the road boundary becomes higher.

  In FIG. 4, the curb A1 and the white line B are close to the host vehicle V and close to each other, so that a higher priority is set than the curb A2 that is far from the relative position. For example, the priority weight of the white line B and the curb A1 is set to 2, and the priority weight of the curb A2 is set to 1. Since the road sign D1 is associated with the curb A1, the priority of the curb A1 may be set according to the relative position of the road sign D1. As described above, the target selection unit 35 sets a higher priority as the relative position between the traveling road boundary or the object associated with the traveling road boundary and the host vehicle becomes closer.

  In the case of FIG. 4, the target selection unit 35 adds the priority weight set according to the height of the target and the priority weight set according to the relative position, and the curb A <b> 1. The priority weight of 7 is 7, the priority weight of curb A2 is 4, and the priority weight of white line B is 3. Thereby, a priority is set in order of the curb A1, the curb A2, and the white line B.

  Next, FIG. 5 shows a situation where an oncoming vehicle is running in addition to the situation of FIG. In this case, the oncoming vehicle C1 is associated with the white line B existing between the own vehicle V and the oncoming vehicle C1, and the oncoming vehicle C1 has a higher relative speed with the own vehicle, so that the influence is greater when it comes into contact with the own vehicle. . Therefore, a high priority is set for the white line B associated with the oncoming vehicle C1. For example, the priority weight of the white line B is set to 5. Thus, the target selection unit 35 sets a higher priority as the relative speed between the object associated with the road boundary and the host vehicle increases. As a result, the priority weight of each runway boundary is 8 for the white line B, 7 for the curb A1, and 4 for the curb A2. Thereby, a priority is set in order of white line B, curb A1, and curb A2.

  FIG. 6 shows a situation where the host vehicle, the oncoming vehicle, and the parallel vehicle are running across the white line. In this case, the white line B1 on the right side of the host vehicle V is associated with the oncoming vehicle C1, and the white line B2 on the left side of the host vehicle V is associated with the parallel running vehicle C2. And since the relative speed of the oncoming vehicle C1 is faster than that of the parallel running vehicle C2, the influence is greater when it comes into contact. Therefore, a high priority is set for the white line B1 on the right side associated with the oncoming vehicle C1. For example, the priority weight of the white line B1 associated with the oncoming vehicle C1 is set to 5, and the priority weight of the white line B2 associated with the parallel running vehicle C2 is set to 3. Moreover, since the relative position with the own vehicle is near, a high priority is set to the white line B1 and the white line B2. For example, the priority weights of the white line B1 and the white line B2 are set to 2, and the priority weights of the curb A1 and the curb A2 are set to 1. When these priority weights are added, the priority weight of the white line B1 is 7, the priority weight of the white line B2 is 5, and the priority weight of the curb A1 and the curb A2 is 1. Thereby, a priority is set in order of white line B1, white line B2, curb stone A1, curb stone A2.

  FIG. 7 shows a situation in which a pedestrian is present across a curb in addition to the situation of FIG. In this case, the white line B on the right side of the host vehicle V is associated with the oncoming vehicle C1, and the curb A1 on the left side of the host vehicle V is associated with the pedestrian D2. Then, the influence of the own vehicle on each object is compared in consideration of the attributes of the object of the vehicle and the pedestrian, and a high priority is set to the curb A1 associated with the pedestrian D2 where the influence when contacting is greater. Is done. For example, the priority weight of the curb A1 associated with the pedestrian D2 is set to 8. As described above, the target selection unit 35 sets the priority according to the attribute of the object associated with the road boundary. For example, a higher priority is set for an object that requires safety. As a result, the priority weight of each track boundary is 15, the priority weight of the curb A1 is 15, the priority weight of the white line B is 8, and the priority weight of the curb A2 is 4. Thereby, a priority is set in order of curb A1, white line B, curb A2.

  When the priority is set in this way, in step S60, the target selection unit 35 selects a target. In particular, the target selection unit 35 selects a road boundary such as a white line or a curb as a target. At this time, first, a target that is within a predetermined range from the host vehicle is selected, and if it is determined in step S80 that the average error is equal to or greater than the predetermined value, the target is based on the priority. Select. For example, the target with the lowest priority is removed and the target is selected again.

  In step S70, the self-position estimating unit 36 performs matching between the target position data of the target selected in step S60 and the target position information included in the map information 41, and performs self-position estimation. When the map information acquisition unit 34 acquires the target position information including the white line information and the road boundary information from the map information 41, the self-position estimation unit 36 selects the target position information and the target position information selected in step S60. The target position data of the target is compared and collated. Then, the position on the map that is most consistent is obtained. The method for estimating the position on the map is not particularly limited as long as it is widely known as a self-position estimation method.

  In step S80, the self-position estimation unit 36 determines whether or not the average value of the error between the position of the selected target and the position of the target on the map is less than a predetermined value as a result of the collation. The predetermined value is set to 5 cm, for example, when the self-position estimation result is used in automatic operation control. If the average value of the errors is less than the predetermined value, the position on the map obtained by collation is estimated as the own position of the host vehicle. On the other hand, when the average value of the errors is equal to or greater than the predetermined value, the process returns to step S60, and a target is selected based on the priority. For example, the target with the lowest priority is removed and selected again, and matching is performed again in step S70. As a result, when the error is less than a predetermined value, the position on the map at that time is estimated as the own position of the host vehicle. The processes in steps S60 to S80 are repeatedly executed until the average error value becomes less than a predetermined value.

  In step S90, the self-position estimating unit 36 outputs the self-position estimation result estimated in step S80. This output value is used, for example, to display the position of the host vehicle on a map of the navigation device, or to calculate a route from the position of the host vehicle to the destination and perform route guidance. When used in automatic driving control, it is used for setting the position of the host vehicle on a map and performing steering control and vehicle speed control of the host vehicle along a route on the map. When the self-position estimation result is output in this way, the self-position estimation process according to the present embodiment ends.

[Effect of the embodiment]
As described above in detail, in the self-position estimation apparatus and method according to the present embodiment, a plurality of road boundaries existing on the left and right sides of the host vehicle are extracted from the target position data, and the plurality of extracted road boundaries are extracted. A priority is set for each of the two, and a road boundary with a high priority is selected. Then, the target position of the host vehicle is estimated by collating the target position data of the selected road boundary with the target position information included in the map information. Thereby, even if it cannot collate with map information without inconsistency with all the targets, since it collates only to the target with high priority, the self-position estimation accuracy can be improved.

  In the self-position estimation apparatus and method according to the present embodiment, the priority is set higher as the relative position between the vehicle boundary and the object associated with the road boundary and the host vehicle becomes closer. Thereby, since a self-position can be estimated from the target which exists near the own vehicle with a high detection accuracy, a self-position can be estimated more correctly.

  Furthermore, in the self-position estimation apparatus and method according to the present embodiment, the priority is set higher as the height of the road boundary or the object associated with the road boundary increases. Thereby, since the priority of the three-dimensional object is set higher than that of the flat object, it is possible to estimate the self-position with high accuracy with respect to the three-dimensional object having a large influence on the own vehicle.

  In the self-position estimation apparatus and method according to the present embodiment, the priority is set higher as the relative speed between the object associated with the road boundary and the host vehicle becomes faster. Thereby, since the priority of the center line existing between the oncoming vehicle and the host vehicle is set high, it is possible to estimate the self-position with high accuracy with respect to the center line, and the possibility of contacting the oncoming vehicle Can be reduced.

  Furthermore, in the self-position estimation apparatus and method according to the present embodiment, the priority is set according to the attribute of the object associated with the road boundary. Thereby, since the priority of a pedestrian can be set higher than roadside objects, such as a guardrail, the self-position estimation in consideration of the attribute of an object, such as giving priority to human life, can be performed.

  The above-described embodiment is an example of the present invention. For this reason, the present invention is not limited to the above-described embodiment, and even if it is a form other than this embodiment, as long as it does not depart from the technical idea of the present invention, it depends on the design and the like. Of course, various modifications are possible.

V Vehicle 1 Ambient sensor group 3 Self-position estimation device 4 Storage device 5 Vehicle sensor group 11, 13 Camera 31 Target position detection unit 32 Movement amount estimation unit 33 Target position storage unit 34 Map information acquisition unit 35 Target selection unit 36 Self-position estimation unit 41 Map information 51 GPS receiver 52 Accelerator sensor 53 Steering sensor 54 Brake sensor 55 Vehicle speed sensor 56 Acceleration sensor 57 Wheel speed sensor 58 Other sensors

Claims (6)

A target detection sensor is provided that detects a relative position between the target vehicle and the target that is mounted on the host vehicle and exists around the host vehicle, and the relative position detected by the target detection sensor is moved by the host vehicle. The target position data is moved by an amount and accumulated as target position data. The accumulated target position data is collated with map information including target position information of the target existing on the map, and the current position of the host vehicle A self-position estimation method by a self-position estimation device that estimates a self-position,
Extracting a plurality of road boundaries existing on the left and right of the host vehicle from the accumulated target position data, setting a priority for each of the extracted plurality of road boundaries, and selecting a high-priority road boundary, A self-position estimation method, comprising: collating target position data on a selected road boundary with target position information included in the map information to estimate the self-position of the host vehicle.   2. The self-position estimation method according to claim 1, wherein the priority is set higher as a relative position between the running road boundary or an object associated with the running road boundary and the host vehicle becomes closer.   3. The self-position estimation method according to claim 1, wherein the priority is set higher as a height of the road boundary or an object associated with the road boundary becomes higher.   The self-position estimation according to any one of claims 1 to 3, wherein the priority is set higher as a relative speed between an object associated with the road boundary and the host vehicle becomes faster. Method.   5. The self-position estimation method according to claim 1, wherein the priority is set according to an attribute of an object associated with the road boundary. A target detection sensor is provided that detects a relative position between the target vehicle and the target that is mounted on the host vehicle and exists around the host vehicle, and the relative position detected by the target detection sensor is moved by the host vehicle. The target position data is moved by an amount and accumulated as target position data. The accumulated target position data is collated with map information including target position information of the target existing on the map, and the current position of the host vehicle A self-position estimation device for estimating a self-position,
Extracting a plurality of road boundaries existing on the left and right of the host vehicle from the accumulated target position data, setting a priority for each of the extracted plurality of road boundaries, and selecting a high-priority road boundary, A self-position estimation apparatus comprising a controller that collates target position data of a selected road boundary with target position information included in the map information and estimates the self-position of the host vehicle.

Images