WO2018168956A1

WO2018168956A1 - Own-position estimating device

Info

Publication number: WO2018168956A1
Application number: PCT/JP2018/010057
Authority: WO
Inventors: 健司三宅; 雄一南口; 謙太高橋; 稔岡田; 和美伊佐治; 竜巳杉山
Original assignee: 株式会社デンソー
Priority date: 2017-03-16
Filing date: 2018-03-14
Publication date: 2018-09-20
Also published as: WO2018168961A1

Abstract

An own-position estimating device (10) is provided with: a map information acquiring unit (109) which acquires map information including lane information of lanes along which a vehicle is capable of traveling; an own-position measuring unit (101), a vehicle movement amount measuring unit (102) and a dead reckoning (106), serving as a position calculating unit which calculates a host vehicle absolute position, which is the absolute position of a host vehicle, on the basis of navigation signals received from a plurality of navigation satellites; and a position estimating unit (108) which estimates a corrected host vehicle position, which is the position of the host vehicle after correction, on the basis of the map information and the host vehicle absolute position. The position estimating unit (108) estimates the corrected host vehicle position by superimposing a degree of certainty of the map information and a degree of certainty of the host vehicle absolute position.

Description

Self-position estimation device

Cross-reference of related applications

This application consists of Japanese Patent Application No. 2017-051066 filed on March 16, 2017, Japanese Patent Application No. 2017-248745 filed on December 26, 2017, and December 26, 2017. Japanese Patent Application No. 2017-248744 filed in Japan and claims the benefit of its priority, the entire contents of which are incorporated herein by reference .

This disclosure relates to a self-position estimation apparatus.

2. Description of the Related Art A self-localization device described in Patent Document 1 below is known as an apparatus for estimating a self-position of a vehicle. The self-localization device described in Patent Document 1 below uses existing road infrastructure such as white lines and road signs on the road for position calculation using GPS (Global Positioning System), inertial devices, and vehicle speed pulses. This will increase the orientation accuracy.

Specifically, the azimuth angle of the white line reflected in the image captured using the camera is calculated, and the difference between the azimuth angle of the white line stored in the azimuth database by the Kalman filter and the azimuth angle of the white line calculated from the image is calculated. Based on this, error estimation is performed.

JP 2008-249639 A

In Patent Document 1, since an image captured using a camera is used, an error cannot be estimated when an image cannot be clearly obtained, such as when the weather is bad. In particular, if position estimation at the lane level is required, the technique described in Patent Document 1 cannot cope with it. In hard driving support and automatic driving, it is necessary to specify the lane and the driving position in the lane, so that more accurate self-position estimation is required.

This disclosure aims to provide a self-position estimation device capable of highly accurate position estimation at the lane level.

The present disclosure is a self-position estimation device, and includes a map information acquisition unit (109) that acquires map information including lane information on which a vehicle can travel, and a navigation signal received from a plurality of navigation satellites. Based on the position calculation unit (101, 102, 106) that calculates the absolute position of the own vehicle and the map information and the absolute position of the own vehicle, the corrected own vehicle position that is the corrected position of the own vehicle is estimated. A position estimation unit (108). The position estimation unit estimates the corrected vehicle position by superimposing the accuracy of the map information and the accuracy of the vehicle absolute position.

In the present disclosure, since the certainty of the map information and the certainty of the absolute position of the own vehicle are overlapped, the corrected own vehicle position can be estimated with higher accuracy than when only the own vehicle absolute position is used.

The reference numerals in parentheses described in the “Summary of the Invention” and “Claims” indicate the correspondence with the “Mode for Carrying Out the Invention” to be described later. It does not indicate that the “claims” are limited to the “modes for carrying out the invention” described below.

FIG. 1 is a block configuration diagram illustrating a functional configuration of the self-position estimation apparatus according to the embodiment. FIG. 2 is a diagram for describing self-position estimation according to the present embodiment. FIG. 3 is a diagram for describing self-position estimation according to the present embodiment. FIG. 4 is a diagram for describing self-position estimation according to the present embodiment. FIG. 5 is a diagram for explaining the lane change probability in the present embodiment. FIG. 6 is a diagram for explaining the relationship between the lane change probability and the reliability of the lane center line. FIG. 7 is a diagram for explaining the relationship between the vehicle position probability distribution and the map probability distribution. FIG. 8 is a diagram for explaining the superposition of the vehicle position probability distribution and the map probability distribution when the lane change probability is low. FIG. 9 is a diagram for explaining the superposition of the vehicle position probability distribution and the map probability distribution when the lane change probability is high. FIG. 10 is a diagram for explaining the case where the lateral deviation probability distribution is taken into account. FIG. 11 is a diagram for explaining the case where the line type probability distribution is taken into account. FIG. 12 is a diagram for explaining the deviation of the estimated position when the vehicle position estimation according to the present embodiment is not performed. FIG. 13 is a diagram for explaining an estimated position when the vehicle position is estimated according to the present embodiment. FIG. 14 is a diagram for explaining the deviation of the estimated position when the vehicle position estimation according to the present embodiment is not performed. FIG. 15 is a diagram for explaining an estimated position when the vehicle position is estimated according to the present embodiment. FIG. 16 is a diagram for explaining the relationship between the estimated position correction amount and the estimated error correction amount. FIG. 16 is a diagram for explaining the relationship between the estimated position correction amount and the estimated error correction amount. FIG. 17 is a diagram for explaining the relationship between the initialization of the correction amount integrated value and the vehicle position candidate. FIG. 19 is a diagram for explaining the relationship between the initialization of the correction amount integrated value and the vehicle position candidate. FIG. 20 is a diagram for explaining the relationship between the initialization of the correction amount integrated value and the vehicle position candidate. FIG. 21 is a diagram for explaining the transition of the number of vehicle position candidates. FIG. 22 is a diagram for explaining the transition of the number of vehicle position candidates. FIG. 23 is a diagram for explaining the transition of the number of vehicle position candidates. FIG. 24 is a flowchart illustrating the process described with reference to FIGS. 16 to 23. FIG. 25 is a flowchart illustrating the processing described with reference to FIGS. 16 to 23.

Hereinafter, the present embodiment will be described with reference to the accompanying drawings. In order to facilitate the understanding of the description, the same constituent elements in the drawings will be denoted by the same reference numerals as much as possible, and redundant description will be omitted.

The self-position estimation apparatus 10 according to the present embodiment will be described with reference to FIG. The self-position estimation apparatus 10 is configured as a computer including a calculation unit such as a CPU, a storage unit such as a RAM and a ROM, and an interface unit for exchanging data with various sensors as hardware components. Subsequently, functional components of the self-position estimation apparatus 10 will be described.

The self-position estimation apparatus 10 includes a self-position measurement unit 101, a vehicle momentum measurement unit 102, a white line recognition unit 103, a surrounding environment measurement unit 104, a route information acquisition unit 105, a dead reckoning 106, and a position estimation unit 108. A map information acquisition unit 109, a travel lane estimation unit 110, and a map information storage unit 120.

The self-position measuring unit 101 is a part for measuring the position of the own vehicle by GNSS (Global Navigation Satellite System). The self-position measuring unit 101 calculates the own vehicle absolute position, which is the absolute position of the own vehicle, based on navigation signals received from a plurality of navigation satellites. The self-position measuring unit 101 outputs the calculated own vehicle absolute position to the dead reckoning 106 and the position estimating unit 108. The self-position measuring unit 101 corresponds to the position calculating unit of the present disclosure.

The vehicle momentum measuring unit 102 is a part that receives signals from sensors such as an acceleration sensor, a vehicle speed sensor, and a gyro sensor and measures the amount of movement of the vehicle. The vehicle momentum measuring unit 102 outputs information on the momentum such as the vehicle speed, the azimuth angle, the yaw rate, and the acceleration to the dead reckoning 106 and the position estimating unit 108. The vehicle momentum measurement unit 102 corresponds to the position calculation unit of the present disclosure.

The white line recognition unit 103 is a part for recognizing a white line that divides a lane using image data captured by the camera. The white line recognition unit 103 outputs information on the presence or absence of a white line and information on the type of white line to the position estimation unit 108.

The surrounding environment measurement unit 104 is a part that measures the weather and satellite arrangement information. The surrounding environment measurement unit 104 outputs weather and satellite arrangement information to the position estimation unit 108.

The route information acquisition unit 105 is a part that acquires the destination of the vehicle and the route to the destination from the navigation system. The route information acquisition unit 105 outputs information indicating the destination and the route to the travel lane estimation unit 110.

The dead reckoning 106 is based on the absolute position of the own vehicle output from the self-position measuring unit 101 and the information on the momentum output from the vehicle momentum measuring unit 102. This is the part to calculate. The dead reckoning 106 outputs the calculated position of the host vehicle to the position estimation unit 108. The dead reckoning 106 corresponds to the position calculation unit of the present disclosure. The self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106 as a position calculating unit have a function of calculating the own vehicle absolute position, which is the absolute position of the own vehicle, based on navigation signals received from a plurality of navigation satellites. Plays.

The map information acquisition unit 109 is a part that acquires map information including lane information on which the vehicle can travel. The map information acquisition unit reads the map information stored in the map information storage unit 120 and outputs the read map information to the position estimation unit 108 and the traveling lane estimation unit 110.

The position estimation unit 108 is a part that estimates a corrected vehicle position that is a corrected position of the vehicle based on the map information and the vehicle absolute position. The position estimation unit 108 estimates the corrected vehicle position by superimposing the accuracy of the map information and the accuracy of the vehicle absolute position. As the probability of the map information and the accuracy of the absolute position of the vehicle, a probability distribution representing the probability may be used, or a numerical value representing the certainty may be used.

An example of a method for estimating the corrected vehicle position of the position estimation unit 108 will be described with reference to FIG. In FIG. 2, three lanes L1, L2, and L3 are set. A solid white line SLa is provided on the left side in the traveling direction of the lane L1. A broken line white line BLa is provided between the lane L1 and the lane L2. A broken line white line BLb is provided between the lane L2 and the lane L3. A solid white line SLb is provided on the right side in the traveling direction of the lane L3. The lane center line L1c is a line indicating the center of the lane L1. The lane center line L2c is a line indicating the center of the lane L2. The lane center line L3c is a line indicating the center of the lane L3.

In the example shown in FIG. 2, the map probability distribution PDm of the lane center line L1c, the lane center line L2c, and the lane center line L3c is regarded as the likelihood of the map information. First, it is assumed that the host vehicle is at the host vehicle absolute position Pa. For convenience of explanation, it is assumed that the host vehicle is traveling along the lane L1 from the host vehicle absolute position Pa.

The position estimation unit 108 estimates the corrected vehicle position Pb by superimposing the vehicle position probability distribution PDca and the map information probability distribution PDm at the vehicle absolute position Pa at the first estimation timing. The corrected host vehicle position Pb is corrected to the lane center line L1c side by the distance d1 as compared with the case where the correction is not performed from the host vehicle absolute position Pa.

The position estimation unit 108 estimates the corrected vehicle position Pc by superimposing the vehicle position probability distribution PDcb at the corrected vehicle position Pb and the map information probability distribution PDm at the next estimation timing. The corrected vehicle position Pc is corrected to the lane center line L1c side by the distance d2 as compared to the case where no correction is made from the corrected vehicle position Pb.

The map probability distribution is not limited to the probability distribution of the lane center line, but a probability distribution indicating the certainty of the map information is used. Further, a map probability distribution offset by the driver's bag or road shape may be used. The road shape includes information such as the road width and the presence or absence of an adjacent lane.

In the example shown in FIG. 3, the map probability distribution PDmA offset from the lane center line L1c, the lane center line L2c, and the lane center line L3c to the left in the traveling direction is assumed to be the accuracy of the map information.

The position estimation unit 108 estimates the corrected vehicle position PbA by superimposing the vehicle position probability distribution PDcaA and the map information probability distribution PDmA at the vehicle absolute position PaA at the first estimation timing. The corrected host vehicle position PbA is corrected to the lane center line L1c side by the distance d1A as compared to the case where the correction is not performed from the host vehicle absolute position PaA.

The position estimation unit 108 estimates the corrected vehicle position PcA by superimposing the vehicle position probability distribution PDcbA at the corrected vehicle position PbA and the map information probability distribution PDmA at the next estimation timing. The corrected host vehicle position PcA is corrected to the lane center line L1c side by a distance d2A as compared to the case where no correction is made from the corrected host vehicle position PbA.

Compared to the case where the map probability distribution PDm of the lane center line L1c, the lane center line L2c, and the lane center line L3c described with reference to FIG. 2 is used, in the example described with reference to FIG. The corrected vehicle position is estimated to approach the left side.

The position estimation unit 108 estimates at least one of the accuracy of the map information and the accuracy of the absolute position of the host vehicle, changes the relative weight of the map information and the absolute position of the host vehicle, and estimates the corrected host vehicle position. can do.

In the example shown in FIG. 4, the probability of the map probability distribution PDmB of the lane center line L1c, the lane center line L2c, and the lane center line L3c is relative to the probability of the own vehicle position probability distributions PDcaB, PDcbB, and PDccB. The weighting is changed to be higher.

The position estimation unit 108 estimates the corrected vehicle position PbB by superimposing the vehicle position probability distribution PDcaB and the map information probability distribution PDmB at the vehicle absolute position PaB at the first estimation timing. The corrected host vehicle position PbB is corrected to the lane center line L1c side by the distance d1B as compared with the case where the correction is not performed from the host vehicle absolute position PaB. The distance d1B is longer than the distance d1 shown in FIG. 2, and the corrected vehicle position PbB is closer to the lane center line L1c side than the corrected vehicle position Pb.

The position estimation unit 108 estimates the corrected vehicle position PcB by superimposing the vehicle position probability distribution PDcbB at the corrected vehicle position PbB and the map information probability distribution PDmB at the next estimation timing. The corrected host vehicle position PcB is corrected to the lane center line L1c side by a distance d2B as compared to the case where no correction is made from the corrected host vehicle position PbB. The distance d2B is longer than the distance d2 shown in FIG. 2, and the corrected vehicle position PcB is closer to the lane center line L1c side than the corrected vehicle position Pc.

Thus, by increasing the probability of the map probability distribution PDmB and increasing the center of the mountain of the map probability distribution PDmB, it is possible to perform correction with an increased convergence speed to the lane center line L1c.

By the way, if the host vehicle continues to travel in the lane L1 without performing a lane change, the estimation of the corrected host vehicle position described with reference to FIG. 4 is highly accurate. However, in fact, the host vehicle may change lanes, and even in that case, it is necessary to increase the accuracy of the corrected host vehicle position.

Accordingly, the position estimation unit 108 uses a lane change probability (hereinafter, also referred to as “LC probability”) that is a probability indicating the degree to which the host vehicle is estimated to move to a lane different from the currently traveling lane, and uses a map. The certainty of information can be varied. More specifically, the position estimation unit 108 can decrease the probability of the map information when the lane change probability is increased, and can increase the probability of the map information when the lane change probability is decreased. Further, the position estimation unit 108 calculates the lane change probability based on the lane deviation amount that is the deviation amount of the traveling state of the host vehicle with respect to the extended state of the lane in which the host vehicle is traveling. The position estimation unit 108 determines the amount of lane deviation as an azimuth angle deviation that is a difference between the azimuth angle of the lane and the traveling direction of the host vehicle, the yaw rate of the host vehicle, the turning angle of the host vehicle, the steering angle of the host vehicle, It can be calculated based on at least one of the surrounding environment information of the place where the host vehicle is traveling.

As shown in FIG. 5A, if the azimuth deviation θ, which is the difference between the azimuth angle in the lane direction and the azimuth angle in the traveling direction of the host vehicle, increases, it is determined that the LC probability is high. On the other hand, as shown in FIG. 5B, when the azimuth deviation, which is the difference between the azimuth angle in the lane direction and the azimuth angle in the traveling direction of the host vehicle, is small or zero, it is determined that the LC probability is low. To do. The relationship between the azimuth deviation and the LC probability is as shown in FIG. 5C as an example.

The position estimation unit 108 estimates the corrected vehicle position based on the calculated lane change probability, the lane information, and the vehicle absolute position. The position estimation unit 108 increases the contribution degree of the lane information when the lane change probability does not indicate the lane change of the own vehicle, and the lane information indicates that the lane change probability indicates the lane change of the own vehicle. Perform correction to reduce the contribution. By increasing the contribution of the lane information when the lane change probability does not indicate the lane change of the own vehicle, even if the own vehicle absolute position deviates from the actual own vehicle position, the actual lane information is corrected by the lane information correction. The corrected vehicle position can be estimated to be close to the vehicle position. On the other hand, if the lane change probability indicates the lane change of the host vehicle, the position estimation is made to be in the lane before the lane change despite actually changing the lane by reducing the lane information contribution degree. Such an erroneous estimation of the corrected vehicle position can be avoided.

As shown in FIG. 6, the position estimation unit 108 calculates the own vehicle absolute position as the own vehicle position probability distribution. The position estimation unit 108 calculates the lane center position in the lane information as a lane center probability distribution. The position estimation unit 108 estimates the corrected vehicle position by superimposing the vehicle position probability distribution and the lane center probability distribution.

7, the position estimation unit 108 increases the reliability of the lane center probability distribution when the lane change probability does not indicate the lane change of the host vehicle. When the lane change probability indicates the lane change of the host vehicle, the position estimation unit 108 estimates the corrected host vehicle position by reducing the reliability of the lane center probability distribution.

If the lane change probability does not indicate the lane change of the host vehicle, the position estimation unit 108 determines that the lane keep is being performed, increases the reliability of the lane center line, and narrows the lane center probability distribution. If the lane change probability indicates the lane change of the host vehicle, the position estimation unit 108 determines that the lane change is being performed, reduces the reliability of the lane center line, and widens the lane center probability distribution.

FIG. 8 is a diagram for explaining superposition of probability distributions during lane keeping. As shown in FIG. 8A, the own vehicle position probability distribution and the lane center probability distribution are calculated. Since it is determined that the lane is being maintained, the reliability of the lane center line is high, and the lane center probability distribution is narrowed. As shown in FIG. 8B, the corrected vehicle position probability distribution is obtained by superimposing the vehicle position probability distribution and the lane center probability distribution. Since the reliability of the lane center line is increased, the corrected vehicle position is corrected so as to be close to the lane center, and the change in the lateral position is suppressed.

FIG. 9 is a diagram for explaining superposition of probability distributions during a lane change. As shown in FIG. 9A, the own vehicle position probability distribution and the lane center probability distribution are calculated. Since it is determined that the lane is being changed, the reliability of the lane center line is low, and the lane center probability distribution is widened. As shown in FIG. 9B, the probability distribution of the corrected vehicle position is obtained by superimposing the vehicle position probability distribution and the lane center probability distribution. Since the reliability of the lane center line is lowered, the corrected vehicle position is corrected to be close to the vehicle position, and the change in the lateral position is maintained.

The concept of obtaining the corrected vehicle position by superimposing such probability distributions can be applied to reflect other factors in the corrected vehicle position. FIG. 10 shows an example in which the probability distribution of lateral deviation is reflected. As shown in FIG. 10A, a lateral deviation probability distribution is calculated in addition to the own vehicle position probability distribution and the lane center probability distribution. The lateral deviation probability distribution is a probability distribution indicating which of the lane the vehicle tends to locate based on the white line recognition result of the white line recognition unit 103. As shown in FIG. 10 (B), in addition to the vehicle position probability distribution and the lane center probability distribution, the corrected vehicle position and the corrected amount taking the lateral deviation into account are obtained by superimposing the lateral deviation probability distribution. Can be calculated.

FIG. 11 shows an example in which the probability distribution of line types is reflected. As shown in FIG. 11A, in addition to the vehicle position probability distribution and the lane center probability distribution, a line type probability distribution is calculated. The line type probability distribution is a probability distribution indicating, for example, whether the lane is a continuous line or a broken line based on the white line recognition result of the white line recognition unit 103. As shown in FIG. 11B, in addition to the own vehicle position probability distribution and the lane center probability distribution, the corrected own vehicle position and correction amount taking the line type into account are obtained by superimposing the line type probability distribution. Can be calculated.

The position estimation unit 108 outputs the estimated corrected vehicle position to the travel lane estimation unit 110. The travel lane estimation unit 110 is a part that estimates a travel lane that is a lane in which the host vehicle is traveling in the map information using the host vehicle absolute position or the corrected host vehicle position.

With reference to FIGS. 12 to 15, estimation of the corrected vehicle position by the position estimation unit 108 and estimation of the travel lane by the travel lane estimation unit 110 will be described. In FIGS. 12 to 15, the actual running line 21 that is the actual running state is indicated by a solid line, and the measurement line 22 that is the locus of the absolute position of the vehicle by the self-position measuring unit 101 and dead reckoning 106 is indicated by a broken line. The traveling lane 23 estimated with a dashed-dotted line is shown.

FIG. 12 shows an example when the position estimation unit 108 does not estimate the corrected vehicle position. In the actual running line 21, the lane is changed to the right lane after running in the left lane. However, since the measurement line 22 has an error that shifts to the left as a whole, the estimation of the travel lane 23 is deviated, and it is estimated that the vehicle is traveling without changing the lane in the left lane.

FIG. 13 shows an example in which the position estimation unit 108 estimates the corrected vehicle position. In the area 30, the vehicle is traveling in the left lane without changing lanes. Therefore, the measurement line 22 is corrected and arranged so as to approach the lane center. The traveling lane estimation unit 110 estimates that the host vehicle is traveling in the left lane in the region 30.

FIG. 14 shows an example in which the corrected vehicle position is not estimated by the position estimating unit 108 at the time of a lane change. In the region 31, the vehicle is traveling with a lane change and a lane change from the left lane to the right lane. On the other hand, FIG. 15 shows an example of the case where the corrected vehicle position is estimated by the position estimation unit 108 at the time of a lane change. Since the lane change in the region 31 of FIG. 14 is reflected, the measurement line 22 is corrected and arranged so as to approach the lane center of the left lane. This correction continues even if the lane curves. Therefore, the traveling lane estimation unit 110 estimates that the host vehicle is traveling in the right lane in the region 32.

The correspondence relationship between the present embodiment described above and the present disclosure will be described. The self-position measuring unit 101, the vehicle motion measuring unit 102, and the dead reckoning 106 of the present embodiment correspond to a position calculating unit of the present disclosure.

As described above, the self-position estimation apparatus 10 according to the present embodiment is based on the map information acquisition unit 109 that acquires map information including lane information that the vehicle can travel and the navigation signals received from a plurality of navigation satellites. A self-position measuring unit 10, a vehicle motion measuring unit 102, a dead reckoning 106 as a position calculating unit for calculating the absolute position of the own vehicle, which is the absolute position of the vehicle, and the corrected self based on the map information and the own vehicle absolute position. And a position estimation unit 108 that estimates the corrected vehicle position that is the position of the vehicle. The position estimation unit 108 estimates the corrected vehicle position by superimposing the accuracy of the map information and the accuracy of the vehicle absolute position. Since the certainty of the map information and the certainty of the absolute position of the own vehicle are superimposed, the corrected own vehicle position can be estimated with higher accuracy than when only the absolute position of the own vehicle is used.

Further, in the present embodiment, the position estimation unit 108 changes at least one of the certainty of the map information and the certainty of the absolute position of the own vehicle, changes the relative weighting of the map information and the absolute position of the own vehicle, and performs correction. The own vehicle position can be estimated. By changing the relative weights of the map information and the absolute position of the vehicle, the influence of more reliable information can be increased and the accuracy of position estimation can be improved.

Further, in the present embodiment, the position estimation unit 108 changes the probability of the map information using a lane change probability that is a probability indicating a degree that the host vehicle is estimated to move to a lane different from the currently traveling lane. Can be made. More specifically, the position estimation unit 108 can maintain or decrease the probability of the map information when the lane change probability increases, while increasing the probability of the map information when the lane change probability decreases. When the host vehicle does not change lanes, it is possible to estimate the position of the host vehicle along the actual lane by increasing the probability of the map information. On the other hand, when the host vehicle makes a lane change, the influence of excessive map information is eliminated by maintaining or reducing the likelihood of the map information, and the position of the host vehicle reflecting the lane change is estimated. Can do. In this way, by reflecting the lane change probability in the estimation of the corrected vehicle position, it is determined that the lane has been changed to the next lane even though it is in the lane keep state, or the lane change state is not changed. Can be reduced.

Further, in the present embodiment, the position estimation unit 108 can calculate the lane change probability based on a lane deviation amount that is a deviation amount of the traveling state of the host vehicle with respect to the extended state of the lane in which the host vehicle is traveling. . More specifically, the position estimation unit 108 determines the amount of lane deviation as an azimuth angle deviation that is a difference between the azimuth angle of the lane and the traveling direction of the host vehicle, the yaw rate of the host vehicle, the turning angle of the host vehicle, It can be calculated based on at least one of the steering angle of the host vehicle and the surrounding environment information of the place where the host vehicle is traveling.

Further, in the present embodiment, the position estimation unit 108 represents the probability of the map information with a map probability distribution, the probability of the own vehicle absolute position with the own vehicle position probability distribution, and the corrected own vehicle position with the map probability distribution. And the vehicle position probability distribution can be estimated. The map probability distribution includes a probability distribution of a lane in which the vehicle is traveling or a lane center position.

In the present embodiment, the position estimation unit 108 can estimate the corrected vehicle position by reflecting the line type information obtained by the camera. The position estimation unit 108 can estimate the corrected vehicle position by reflecting the peripheral information from the camera.

The position estimation unit 108 can estimate the corrected vehicle position by reflecting the blinker information. For example, when the winker is on the right, the possibility of a lane change to the right lane is high, so the lane change probability can be increased to cope with it.

The position estimation unit 108 can also estimate the corrected vehicle position by reflecting the driver's state. The position estimation unit 108 can also estimate the corrected vehicle position by reflecting the operation by the driver. By reflecting the state and operation of the driver, the possibility of a lane change can be accurately estimated.

The position estimation unit 108 according to the present embodiment can execute reliability adjustment control that reduces the reliability of the corrected vehicle position according to the correction amount for correcting the vehicle absolute position to the corrected vehicle position. .

As the amount of correction increases, the difference between the vehicle's absolute position and the corrected vehicle position increases. Therefore, if the position close to the vehicle's absolute position before correction is the true vehicle position, the corrected vehicle There is a possibility that the true vehicle position will not enter within a predetermined reliability from the position. Therefore, by reducing the reliability of the corrected vehicle position in accordance with the correction amount, it is possible to increase the possibility that the true vehicle position falls within a predetermined reliability from the corrected vehicle position.

The position estimation unit 108 in this embodiment can reduce the reliability of the corrected vehicle position according to the correction amount integrated value obtained by integrating the correction amounts.

By using the correction amount integrated value, it is possible to cope with a case where the absolute position of the vehicle is continuously corrected to the corrected vehicle position along the time series. Even when the correction is continuously performed and the deviation between the absolute position of the vehicle and the corrected vehicle position becomes large, the reliability of the corrected vehicle position is lowered according to the correction amount integrated value. Thus, it is possible to increase the possibility that the true vehicle position will enter the predetermined reliability from the corrected vehicle position.

This will be described with reference to FIG. In FIG. 16, the own vehicle absolute positions 40 (a) and 40 (b) and the reliability 50 (a) and 50 (b) corresponding to each are shown. It is assumed that the vehicle's absolute position 40 (b) is reached after a predetermined time from the vehicle's absolute position 40 (a). As the own vehicle absolute position after the next predetermined time, the self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106, which are position calculating units, calculate the own vehicle absolute position 40 (c2). The position estimation unit 108 calculates the corrected host vehicle position 40 (c1) as the corrected host vehicle position.

When the host vehicle absolute position 40 (c2) is corrected to the corrected host vehicle position 40 (c1), the estimated position correction amount increases, so the reliability adjustment control is executed so that the estimated error correction amount also increases. An example of the relationship between the estimated position correction amount and the estimated error correction amount is shown in FIG.

As shown in FIG. 17, when the estimated position correction amount increases, the lateral position correction amount and the turning angle correction amount also increase, so that the estimated error correction amount is also set to be large.

In FIG. 16, when the reliability of the corrected vehicle position 40 (c2) is within the range of the reliability 50 (c1) when the reliability adjustment control is not executed, the reliability adjustment control is executed. The reliability is expanded to 50 (c2).

The position estimation unit 108 in the present embodiment can initialize the correction amount integrated value when the correction amount integrated value does not exceed the integration threshold value provided corresponding to the predetermined elapsed time. When the correction amount integrated value does not exceed the integration threshold during the predetermined elapsed time, it is possible to suppress the accumulation of lateral errors by initializing the correction amount integrated value.

Referring to FIG. 18, one mode for initializing the correction amount integrated value will be described. FIG. 18A shows changes in the position of the host vehicle and changes in reliability. FIG. 18B shows the transition of the lane change probability. FIG. 18C shows the transition of the correction amount integrated value. FIG. 18D shows the change in the number of own vehicle position candidates.

FIG. 18 shows the own vehicle absolute positions 40 (a) and 40 (b) and the corresponding degrees of reliability 50 (a) and 50 (b). It is assumed that the vehicle absolute position 40 (b) is reached at a time Tb after a predetermined time from the vehicle absolute position 40 (a) corresponding to the time Ta. The self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106, which are position calculation units, calculate the own vehicle absolute position 40 (c2) as the own vehicle absolute position at time Tc, which is the next predetermined time. The position estimation unit 108 calculates the corrected host vehicle position 40 (c1) as the corrected host vehicle position.

When the host vehicle absolute position 40 (c2) is corrected to the corrected host vehicle position 40 (c1), the estimated position correction amount increases, so the reliability adjustment control is executed so that the estimated error correction amount also increases. The execution contents of the reliability adjustment control are the same as those described with reference to FIGS. 16 and 17, and the reliability 50 (c1) is expanded to the reliability 50 (c2).

As the estimated error correction amount increases at time Tc, the correction amount integrated value exceeds the integration threshold. The position estimation unit 108 initializes the correction amount integrated value, and generates two corrected vehicle position candidates. At time Td, the corrected host vehicle position 40 (d1) and the corresponding reliability 50 (d1), and the corrected host vehicle position 40 (d2) and the corresponding reliability 50 (d2) in the adjacent lane are generated. To do.

Referring to FIG. 19, another aspect of initializing the correction amount integrated value will be described. FIG. 19A shows changes in the position of the host vehicle and changes in reliability. FIG. 19B shows the transition of the lane change probability. FIG. 19C shows the transition of the correction amount integrated value. FIG. 19D shows the change in the number of vehicle position candidates.

FIG. 19 shows the own vehicle absolute positions 40 (a) and 40 (b) and the corresponding degrees of reliability 50 (a) and 50 (b). It is assumed that the vehicle absolute position 40 (b) is reached at a time Tb after a predetermined time from the vehicle absolute position 40 (a) corresponding to the time Ta. The self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106, which are position calculation units, calculate the own vehicle absolute position 40 (c2) as the own vehicle absolute position at time Tc, which is the next predetermined time. The position estimation unit 108 calculates the corrected host vehicle position 40 (c1) as the corrected host vehicle position.

In the case of the example shown in FIG. 19, the correction amount integrated value does not exceed the integration threshold even at time Tc. As described above, when the predetermined time has elapsed and the correction amount integrated value does not exceed the integration threshold value, the position estimation unit 108 initializes the correction amount integrated value. Since the correction amount integrated value does not exceed the integration threshold value, the corrected vehicle position candidate remains one. At time Td, the corrected vehicle position 40 (d1) and the corresponding reliability 50 (d1) are generated.

Referring to FIG. 20, another aspect of initializing the correction amount integrated value will be described. FIG. 20A shows the transition of the position of the host vehicle and the transition of the reliability. FIG. 20B shows the transition of the lane change probability. FIG. 20C shows the transition of the correction amount integrated value. FIG. 20D shows the transition of the number of vehicle position candidates.

FIG. 20 shows the own vehicle absolute positions 40 (a) and 40 (b) and the corresponding reliability 50 (a) and 50 (b). It is assumed that the vehicle absolute position 40 (b) is reached at a time Tb after a predetermined time from the vehicle absolute position 40 (a) corresponding to the time Ta. The self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106, which are position calculation units, calculate the own vehicle absolute position 40 (c2) as the own vehicle absolute position at time Tc, which is the next predetermined time. The position estimation unit 108 calculates the corrected host vehicle position 40 (c1) as the corrected host vehicle position.

In the case of the example shown in FIG. 20, the correction amount integrated value does not exceed the integration threshold even at time Tc. In this way, when the correction amount integrated value does not exceed the integration threshold even when the predetermined time has elapsed, the position estimation unit 108 initializes the correction amount integrated value. Furthermore, when the own vehicle absolute position calculated by a calculation method different from the calculation method for calculating the own vehicle absolute position, for example, a calculation method such as line type detection, GNSS, or lateral position detection is the reliability 50 (c3), The reliability 50 (c3) can be prioritized. At time Td, the corrected vehicle position 40 (d2) and the corresponding reliability 50 (d2) are generated.

The position estimation unit 108 in the present embodiment, the vehicle absolute position calculated by a calculation method different from the vehicle absolute position calculation method used for the reliability adjustment control, the corrected vehicle position, and the corrected vehicle When the vehicle position specified by the predetermined reliability from the position is different, the correction amount integrated value can be initialized. By initializing the correction amount integrated value, it is possible to give priority to the own vehicle absolute position calculated by a calculation method different from the calculation method used for the reliability adjustment control.

Subsequently, another aspect of the reliability adjustment control will be described. In the reliability adjustment control, the position estimation unit 108 according to the present embodiment is a hypothesis in which the first corrected host vehicle position for correcting the host vehicle absolute position by the first correction amount and the host vehicle absolute position are different from the first correction amount. It is possible to hold the second corrected vehicle position corrected by the second correction amount based on the above.

By using the first correction amount and the second correction amount that are different from each other, it is possible to hold the first corrected vehicle position and the second corrected vehicle position with different correction degrees. Even if the deviation between the own vehicle absolute position and the corrected own vehicle position becomes large, tracking based on a plurality of hypotheses can be performed, and the first corrected own vehicle position and the second corrected own vehicle position. Therefore, the possibility that the true vehicle position enters within a predetermined reliability can be increased.

Specific description will be given with reference to FIG. FIG. 21A shows the transition of the position of the host vehicle and the transition of the reliability. FIG. 21B shows the transition of the lane change probability. FIG. 21C shows the transition of the number of vehicle position candidates. FIG. 21D shows the change in the number of lane candidates.

FIG. 21 shows the absolute positions 40 (a) and 40 (b) of the own vehicle and the corresponding reliability 50 (a) and 50 (b). It is assumed that the vehicle absolute position 40 (b) is reached at a time Tb after a predetermined time from the vehicle absolute position 40 (a) corresponding to the time Ta.

Since the lane change probability exceeds the predetermined probability at time Tb, the position estimation unit 108 starts the reliability adjustment control. Specifically, the number of vehicle position candidates is set to two, and the number of lane candidates is also set to two. At time Tc, the first corrected vehicle position 40 (c1) is generated on the assumption that the vehicle is traveling in the left lane in the traveling direction, and is held together with the reliability 50 (c1). At the same time, the second corrected vehicle position 40 (c2) is generated on the assumption that the vehicle is traveling in the right lane in the traveling direction, and is held together with the reliability 50 (c2).

Since the corrected first vehicle position 40 (c1) is generated by correcting the absolute vehicle position with respect to the information specifying the position of the left lane in the traveling direction, the corrected amount is This is the first correction amount. Since the second corrected host vehicle position 40 (c2) corrects the host vehicle absolute position with respect to the information specifying the position of the right lane in the traveling direction, the corrected host vehicle position is generated. The second correction amount is based on a hypothesis different from the first correction amount.

When the lateral difference between the first corrected host vehicle position and the second corrected host vehicle position does not exceed the lateral threshold, the position estimating unit 108 in the present embodiment is configured to perform the first corrected host vehicle position or the second corrected vehicle position. The corrected vehicle position can be rejected. When the lateral difference between the first corrected host vehicle position and the second corrected host vehicle position does not exceed the lateral threshold, at least one of the first corrected host vehicle position and the second corrected host vehicle position is rejected By doing so, tracking based on a plurality of hypotheses can be eliminated, and the load of the plurality of tracking can be reduced.

In this embodiment, the position estimation unit 108 executes the reliability adjustment control when the lane change probability, which is the probability that the vehicle is estimated to move to a lane different from the currently running lane, exceeds a predetermined probability. Can start. Since the reliability adjustment control is started when it is estimated that the host vehicle moves to a lane different from the lane in which the host vehicle is currently traveling, the host vehicle after the first correction is made in accordance with the possibility that the host vehicle changes the lane. Since the vehicle position and the second corrected vehicle position are calculated, it is possible to suppress unnecessary tracking based on a plurality of hypotheses.

Referring to FIG. 22, an example of rejecting the second corrected vehicle position while leaving the first corrected vehicle position will be described. In FIG. 22, the own vehicle absolute positions 40 (a), 40 (b) and the corresponding reliability 50 (a), 50 (b) are shown. It is assumed that the vehicle absolute position 40 (b) is reached at a time Tb after a predetermined time from the vehicle absolute position 40 (a) corresponding to the time Ta.

As shown in FIG. 22A, at time Tc, the lateral deviation of the second corrected vehicle position 40 (c2) with respect to the first corrected vehicle position 40 (c1) is small, and the horizontal direction The difference does not exceed the horizontal threshold. Accordingly, the position estimation unit 108 rejects the second corrected host vehicle position 40 (c2). At time Td, the first corrected vehicle position 40 (d1) and the reliability 50 (d1) are maintained.

The position estimation unit 108 in the present embodiment, the vehicle absolute position calculated by a calculation method different from the vehicle absolute position calculation method used for the reliability adjustment control, the first corrected vehicle position, and the first vehicle position When the corrected vehicle position is different from the corrected vehicle position by the predetermined reliability, the first corrected vehicle position is rejected and the vehicle absolute position calculation method used for the reliability adjustment control is calculated. If the vehicle absolute position calculated by a different calculation method from the vehicle position and the vehicle position specified by a predetermined reliability from the second corrected vehicle position and the second corrected vehicle position are different, 2 The vehicle position after correction can be rejected. By rejecting the corrected vehicle position away from the vehicle absolute position calculated by a calculation method different from the calculation method used for the reliability adjustment control, it is possible to reduce the load of multiple tracking.

Referring to FIG. 23, an example of rejecting the first corrected vehicle position while leaving the second corrected vehicle position will be described. In FIG. 23, the own vehicle absolute positions 40 (a) and 40 (b) and the corresponding reliability 50 (a) and 50 (b) are shown. It is assumed that the vehicle absolute position 40 (b) is reached at a time Tb after a predetermined time from the vehicle absolute position 40 (a) corresponding to the time Ta.

Since the lane change probability exceeds the predetermined probability at time Tb, the position estimation unit 108 starts the reliability adjustment control. Specifically, the number of vehicle position candidates is set to two and the number of lane candidates is also set to two. At time Tc, the first corrected vehicle position 40 (c1) is generated on the assumption that the vehicle is traveling in the left lane in the traveling direction, and is held together with the reliability 50 (c1). At the same time, the second corrected vehicle position 40 (c2) is generated on the assumption that the vehicle is traveling in the right lane in the traveling direction, and is held together with the reliability 50 (c2).

As shown in FIG. 23A, at time Tc, a calculation method different from the calculation method for calculating the own vehicle absolute position, for example, a self-calculation method calculated by a calculation method such as line type detection, GNSS, or lateral position detection. When the vehicle absolute position has a reliability of 50 (c3), the reliability of 50 (c3) can be prioritized. The position estimation unit 108 leaves the second corrected host vehicle position 40 (c2) close to the reliability 50 (c3). Based on the second corrected host vehicle position 40 (c2), the position estimation unit 108 generates a corrected host vehicle position 40 (d2) and a corresponding reliability 50 (d2) at time Td.

The operation of the self-position estimation apparatus 10 will be described with reference to the flowchart shown in FIG. In step S101, the self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106, which are position calculating units, calculate the own vehicle absolute position.

In step S102 following step S101, the map information acquisition unit 109 acquires surrounding map data. In step S103 following step S102, the position estimation unit 108 calculates the corrected vehicle position in consideration of the lane shape.

In step S104 following step S103, the correction amount calculated in step S103 is added to the correction amount integrated value. In step S105 subsequent to step S104, it is determined whether the correction amount integrated value exceeds the threshold value. If the correction amount integrated value exceeds the threshold value, the process proceeds to step S107. If the correction amount integrated value does not exceed the threshold value, the process proceeds to step S106.

In step S106, the correction amount integrated value is initialized on condition that the predetermined condition is satisfied. In step S107, it is determined whether the own vehicle absolute position is calculated by another method different from step S101. If the own vehicle absolute position is calculated by another method, the process proceeds to step S108. If the own vehicle absolute position has not been calculated by another method, the process proceeds to step S110.

In step S108, an estimation error is calculated. In step S109 following step S108, the correction amount integrated value is initialized. In step S110, it is determined whether the correction amount integrated value exceeds a threshold value. If the correction amount integrated value exceeds the threshold value, the process proceeds to step S111. If the correction amount integrated value does not exceed the threshold value, the process proceeds to step S113.

In step S111, the estimated position candidates are increased. In step S112 following step S111, the correction amount integrated value is initialized.

In step S113, it is determined whether or not there are a plurality of estimated position candidates. If there are a plurality of estimated position candidates, the process proceeds to step S114. If there are not a plurality of estimated position candidates, the process returns. In step S114, rejection determination is executed.

The operation of the self-position estimation apparatus 10 will be described with reference to the flowchart shown in FIG. In step S201, the self-position measuring unit 101, the vehicle momentum measuring unit 102, and the dead reckoning 106, which are position calculating units, calculate the absolute position of the own vehicle.

In step S202 following step S201, the map information acquisition unit 109 acquires surrounding map data. In step S203 following step S202, a lane change probability is calculated.

In step S204 following step S203, it is determined whether or not there is a single estimated position candidate. If there is a single estimated position candidate, the process proceeds to step S205. If there is a single estimated position candidate, the process proceeds to step S208.

In step S205, it is determined whether or not the lane change probability exceeds a threshold value. If the lane change probability exceeds the threshold, the process proceeds to step S206. If the lane change probability does not exceed the threshold value, the process proceeds to step S207.

In step S206, an estimated position candidate with a modified correction parameter is created. In step S207, position correction based on the lane shape is executed.

In step S208 following step S206 and step S207, position correction based on the lane shape is executed for all estimated position candidates. In step S209 following step S208, it is determined whether the lateral movement distance per unit travel distance exceeds a threshold value. If the lateral movement distance per unit travel distance exceeds the threshold value, the process proceeds to step S210. If the lateral movement distance per unit travel distance does not exceed the threshold value, the process proceeds to step S211.

In step S210, the estimated position candidate is rejected. In step S211, it is determined whether the own vehicle absolute position is calculated by another method different from step S201. If the own vehicle absolute position is calculated by another method, the process proceeds to step S212. If the vehicle's absolute position has not been calculated by another method, the process returns. In step S212, the estimated position candidate is rejected.

The embodiment has been described above with reference to specific examples. However, the present disclosure is not limited to these specific examples. Those in which those skilled in the art appropriately modify the design of these specific examples are also included in the scope of the present disclosure as long as they have the features of the present disclosure. Each element included in each of the specific examples described above and their arrangement, conditions, shape, and the like are not limited to those illustrated, and can be changed as appropriate. Each element included in each of the specific examples described above can be appropriately combined as long as no technical contradiction occurs.

Claims

A self-position estimation device,
A map information acquisition unit (109) for acquiring map information including lane information in which the vehicle can travel;
A position calculation unit (101, 102, 106) for calculating the absolute position of the host vehicle, which is the absolute position of the host vehicle;
A position estimation unit (108) for estimating a corrected host vehicle position that is a corrected position of the host vehicle based on the map information and the host vehicle absolute position;
The position estimation unit is a self-position estimation device that estimates the corrected vehicle position by superimposing the accuracy of the map information and the accuracy of the vehicle absolute position.
The self-position estimation apparatus according to claim 1,
The position estimation unit varies at least one of the probability of the map information and the accuracy of the vehicle absolute position, and changes the relative weighting of the map information and the vehicle absolute position to change the corrected vehicle Self-position estimation device that estimates vehicle position.
The self-position estimation apparatus according to claim 2,
The position estimation unit uses a lane change probability, which is a probability indicating a degree that the own vehicle is estimated to move to a lane different from the currently running lane, and varies the probability of the map information. Estimating device.
The self-position estimation apparatus according to claim 3,
The position estimation unit maintains or decreases the probability of the map information when the lane change probability increases, and increases the probability of the map information when the lane change probability decreases.
The self-position estimation apparatus according to claim 3 or 4,
The said position estimation part is a self-position estimation apparatus which calculates the said lane change probability based on the amount of lane deviation which is the amount of deviation of the driving state of the said vehicle with respect to the extending | stretching state of the lane where the said host vehicle is drive | working.
The self-position estimation apparatus according to claim 5,
The position estimation unit determines the amount of lane deviation, an azimuth angle deviation that is a difference between an azimuth angle of the lane and an azimuth angle of the traveling direction of the host vehicle, a yaw rate of the host vehicle, a turning angle of the host vehicle, A self-position estimating device that calculates based on at least one of a steer angle of the host vehicle and surrounding environment information of a place where the host vehicle is traveling.
The self-position estimation apparatus according to any one of claims 1 to 6,
The position estimation unit represents the likelihood of the map information by a map probability distribution, the probability of the absolute position of the vehicle by a vehicle position probability distribution, and the corrected vehicle position is represented by the map probability distribution and the map probability distribution. Self-position estimation device that estimates by superimposition with the own vehicle position probability distribution.
The self-position estimation apparatus according to claim 7,
The said map probability distribution is a self-position estimation apparatus containing the probability distribution of the lane which the said vehicle is drive | working, or a lane center position.
The self-position estimation apparatus according to any one of claims 1 to 8,
The position estimation unit is a self-position estimation apparatus that estimates the corrected vehicle position by reflecting line type information obtained by a camera.
The self-position estimation apparatus according to any one of claims 1 to 8,
The said position estimation part is a self-position estimation apparatus which estimates the said corrected vehicle position reflecting the surrounding information by a camera.
The self-position estimation apparatus according to any one of claims 1 to 8,
The position estimation unit is a self-position estimation apparatus that estimates the corrected vehicle position by reflecting turn signal information.
The self-position estimation apparatus according to any one of claims 1 to 8,
The said position estimation part is a self-position estimation apparatus which estimates the said corrected vehicle position reflecting a driver's state.
The self-position estimation apparatus according to any one of claims 1 to 8,
The said position estimation part is a self-position estimation apparatus which estimates the said corrected vehicle position reflecting the operation by a driver.
The self-position estimation apparatus according to any one of claims 1 to 13,
The position estimation unit executes reliability adjustment control for reducing reliability of the corrected vehicle position according to a correction amount for correcting the vehicle absolute position to the corrected vehicle position. apparatus.
The self-position estimation apparatus according to claim 14,
The said position estimation part is a self-position estimation apparatus which reduces the reliability of the said vehicle position after correction | amendment according to the correction amount integrated value which integrated | accumulated the said correction amount.
The self-position estimation apparatus according to claim 15,
The position estimation unit is a self-position estimation apparatus that initializes the correction amount integrated value when the correction amount integrated value does not exceed an integration threshold value provided corresponding to a predetermined elapsed time.
The self-position estimation apparatus according to claim 15,
The position estimation unit is configured to calculate the vehicle absolute position calculated by a calculation method different from the vehicle absolute position calculation method used for the reliability adjustment control, the corrected vehicle position, and the corrected vehicle. A self-position estimation device that initializes the correction amount integrated value when a vehicle position that is specified by a predetermined reliability differs from a position.
The self-position estimation apparatus according to claim 14,
In the reliability adjustment control, the position estimation unit is a hypothesis that the first corrected self-vehicle position for correcting the own vehicle absolute position by a first correction amount and the own vehicle absolute position different from the first correction amount. A self-position estimating apparatus that holds a second corrected self-vehicle position that is corrected by a second correction amount based on.
The self-position estimation apparatus according to claim 18, wherein
When the lateral difference between the first corrected host vehicle position and the second corrected host vehicle position does not exceed a lateral threshold, the position estimation unit or the second corrected host vehicle position or the second Self-position estimation device that rejects the corrected vehicle position.
The self-position estimation apparatus according to claim 18, wherein
The position estimator includes a vehicle absolute position calculated by a calculation method different from the vehicle absolute position calculation method used in the reliability adjustment control, the first corrected vehicle position, and the first vehicle position. When the corrected vehicle position is different from the corrected vehicle position by the predetermined reliability, the first corrected vehicle position is rejected,
From the own vehicle absolute position calculated by a calculation method different from the calculation method of the own vehicle absolute position used for the reliability adjustment control, the second corrected own vehicle position, and the second right after own vehicle position. A self-position estimation device that rejects the second corrected self-vehicle position when the self-vehicle position specified by a predetermined reliability is different.
The self-position estimation apparatus according to claim 18, wherein
The position estimator starts execution of the reliability adjustment control when a lane change probability, which is a probability indicating a degree that the own vehicle moves to a lane different from the currently running lane, exceeds a predetermined probability. A self-position estimation device.