JP4760274B2 - Map update device - Google Patents

Description

  The present invention relates to a map updating apparatus, and in particular, detects a mark such as a three-dimensional object around the host vehicle, and calculates a road shape using a travel locus of the host vehicle when the mark is detected with high accuracy. The present invention relates to a map update device capable of updating a map.

When the road information is corrected based on travel locus data obtained by traveling on an actual road, if the road to which the road information can be bidirectionally expressed can be expressed as one road shape, it is based on the travel locus data. A road information correction device that corrects road information generated in map matching for each traveling direction has been proposed (Patent Document 1).
JP 2005-121707 A

  However, in the above conventional technique, the road information is corrected based on the travel locus data obtained from the GPS device, so the correction accuracy of the road information is not always sufficient. In addition, the road shape, the number of lanes, etc. may change depending on the construction, etc. In this case, there is a problem that the travel locus cannot be detected with high accuracy and the road information cannot be corrected accurately. there were.

  The present invention has been made to solve the above-described problems, and updates map information such as road shape from a result of estimating the position of the own vehicle with high accuracy using a mark such as a three-dimensional object around the own vehicle. An object of the present invention is to provide a map updating device capable of

In order to achieve the above object, the present invention provides an own vehicle position detecting means for detecting a position of the own vehicle , a mark detecting means for detecting a position of a mark around the own vehicle as a relative position with respect to the own vehicle, and the own vehicle. A host vehicle position accuracy calculation unit that calculates the accuracy of the position of the host vehicle from variations in the position of the host vehicle detected by the position detection unit; a detection accuracy acquisition unit that acquires a detection accuracy of the mark detection unit; and the host vehicle Based on the accuracy of the position of the host vehicle calculated by the position accuracy calculation means and the detection accuracy of the mark detection means acquired by the detection accuracy acquisition means, the accuracy of the position of the mark detected by the mark detection means is calculated. storage and mark position accuracy calculation means, storage means for storing map information including the position of a road shape and a plurality of landmarks, the position and the storage means of the plurality of landmarks detected by the mark detecting means for Matching determination means and the position of the plurality of landmarks to determine whether matching has been calculated by the mark position accuracy calculation means for location of a plurality of landmarks are determined to be matched by the matching determination means When the accuracy of the position of the mark is compared with a predetermined threshold value, the accuracy of the position of the plurality of marks stored in the storage means is increased when it is determined that the accuracy of the position of the mark is increased. The position of the host vehicle is estimated from the updated positions of the plurality of landmarks, and the road shape stored in the storage means is changed to a road shape corresponding to the estimated position of the host vehicle. And updating means for updating.

  According to the present invention, the position of the host vehicle and the marks around the host vehicle are detected, and the position information of the detected plurality of marks is stored in the storage means. After the position information of the plurality of landmarks is stored in the storage means, it is determined whether or not the position information of the plurality of landmarks detected matches the position information of the plurality of landmarks stored in the storage means. The road shape is calculated on the basis of the position information of the plurality of landmarks determined to be and the detected position information of the own vehicle. Then, the map information is updated based on the calculated road shape. Thereby, the map information is updated to the latest information based on the detected position information of the plurality of landmarks.

  In the present invention, it is preferable to determine the accuracy of the position information of the mark stored in the storage means and update the position information of the mark stored in the storage means to position information with high accuracy. As a result, the position information of the mark such as the three-dimensional object around the host vehicle is always updated to the latest position information, and thus the map information can be updated more accurately.

  As described above, according to the present invention, the map information is updated by detecting a mark such as a three-dimensional object around the host vehicle and calculating the road shape using the traveling locus of the host vehicle when the mark is detected. Therefore, for example, even when a new road is created or when the road shape is temporarily changed due to road construction or the like, an effect that high-accuracy map information is always obtained can be obtained.

  For this reason, it is not necessary to provide a high-accuracy self-vehicle position measurement system, or to newly create or purchase a map, and cost reduction and robustness can be improved. In addition, the road edge can be estimated by registering buildings and other landmarks, etc., so that a static driving area can be provided on a map, stop lines, intersections, signs, road markings Etc. can accurately provide information related to vehicle motion control such as warning and stop control.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  As shown in FIG. 1, the map update device mounted on the vehicle of the present embodiment includes a host vehicle position sensor 10 composed of a GPS device that detects the current position of the host vehicle with a predetermined low accuracy, and a host vehicle. There is provided a landmark sensor 12 for detecting a mark such as a three-dimensional object (hereinafter referred to as a landmark) as a mark around the vehicle with higher accuracy than the vehicle position sensor. The landmark sensor 12 can be constituted by a laser radar capable of sensing a range within ± 30 ° in front of the host vehicle.

  The own vehicle position sensor 10 and the landmark sensor 12 are connected to a control device 14 constituted by a microcomputer or the like.

  Connected to the control device 14 is a storage device 16 for storing map information including road shapes and positions of a plurality of landmarks, relative position information indicating a relative positional relationship between a plurality of landmarks detected by a landmark sensor, and the like. ing. The control device 14 is connected to a display device 18 composed of an LCD or the like that displays a map based on the map information.

  As the GPS device, a general-purpose DGPS device having an accuracy of several meters can be used, and map information of the navigation system (only road information is acceptable in the initial state) can be used as the map information. In the initial state, no initial information may be included.

  From the GPS device, position data indicating the position of the host vehicle is output and input to the control device 14. The control device 14 updates the map information stored in the storage device according to the map information update routine described below, reads the map information stored in the storage device, and represents the map information based on the read map information. A map is displayed on the display device 18, and the position of the host vehicle is displayed on the displayed map.

  The laser radar is configured to include a light emitting element composed of a semiconductor laser that irradiates and scans with an infrared light pulse, and a light receiving element that receives an infrared light pulse reflected from a front landmark or the like. Installed on the grill or bumper. In this laser radar, based on the arrival time of the reflected infrared light pulse until it is received by the light receiving element with respect to the time point when the light is emitted from the light emitting element, the distance r from the own vehicle to the landmark ahead and the landmark Position information represented by polar coordinates (r, θ) indicating the existence direction θ can be detected.

  Note that a gyro sensor may be used as the vehicle position sensor instead of the GPS device, and a millimeter wave radar or a camera may be used as the landmark sensor instead of the laser radar. Moreover, you may use the structure which combined these arbitrarily.

  As landmarks, it is possible to detect electric poles, signs, traffic lights, traffic light pole parts, guard rail pillar parts, reflector guard rails, road markings, characteristic parts such as building walls, etc., and position detection with sensors Use easy landmarks. A landmark that can be detected about once every 20 to 30 m is used as a landmark, and an object such as a three-dimensional object that exists on both the left and right sides of the host vehicle is used as a landmark. For example, it is effective to improve accuracy to use a thing such as a utility pole or a pillar part of a sign as a landmark. Moreover, you may make it detect the pole of a guardrail, the curb located in the edge of a road, a white line, and a center line as a landmark.

  When a landmark that can be detected about once in 20 to 30 m is selected as a landmark, although it depends on the sensor performance, about three landmarks can be detected in one sensing, which is reasonable. If one or more landmarks are detected in one sensing, there is no section in which the traveling position accuracy of the host vehicle is reduced, and the estimation accuracy can be improved reliably.

  If there is a landmark on only one side of the moving object, the lateral position accuracy of the host vehicle is weak, so there is a limit to improving the lateral position accuracy. It is possible to improve.

  Next, FIG. 2 shows changes in the error variance of the landmark density and the vehicle position when a landmark sensor capable of detecting landmarks up to 70 m away is used. The landmark density and the vehicle position error average (true value) are shown in FIG. FIG. 3 shows the change in the difference between the detected value and the detected value. The Z axis indicates the vehicle traveling direction, and the X axis indicates the vehicle lateral direction.

  As shown in FIG. 2, when the landmarks are sparsely present (for example, when the density is 1/500 m to 1/100 m), the dispersion is small during the landmark observation, but in the non-observation state. Since it returns to the same variance as the vehicle location variance of the GPS device alone, there is not much change due to density. When the number of landmarks is 1 / 70m, the next landmark is detected before returning to the dispersion when the GPS device alone detects when the landmark is not observed. Compared to the case of 1 / 100m or more. Dispersion is reduced. In the case of 1 piece / 50 m or less, the landmark is always detected, and the dispersion tends to decrease as the density increases, but there is no dramatic change.

  In addition, as shown in FIG. 3, the error in the Z direction is an average of one landmark / 70 m or less, as long as even one landmark can be observed, compared to the case where a single GPS device is detected. The error is less than half, and there is not much change due to density.

  Regarding the error in the X direction, there is no significant difference between the case where the GPS device is detected alone and the case where the landmark is sparse.

  In the case of a density of 1/70 m (the unobserved state of the landmark exists, the next observed value is obtained before returning to the detected state of the GPS device alone), it is effective in reducing the lateral position estimation error .

  When two or more landmarks always exist in one sensing (1/30 m to 1/20 m), the estimation error tends to be further reduced.

  From the above, when a sensor having a detection range of 70 m is used, it is desirable that landmarks with a density of 1/30 m to 1/20 m exist to improve estimation accuracy. When the detection range of the sensor is very narrow, it is expected that the landmark density is better. When the target accuracy is set, it is desirable to derive the landmark interval that can achieve the sensor accuracy and the target accuracy, and select an appropriate landmark according to the result.

  Next, the landmark executed by the control device of the present embodiment is detected and the position information of the landmark is registered, and it is determined whether the registered landmark matches the detected landmark. A routine for updating the map information will be described with reference to FIG.

While the host vehicle is traveling, the host vehicle position information indicating the position of the host vehicle detected by the GPS device in step 96 is captured, and the host vehicle position information is distributed based on the host vehicle position information captured in step 98 P _vv (t) is calculated, and it is determined in step 100 whether or not a landmark has been detected by the landmark sensor. If no landmark is detected, the process waits for a predetermined time Δt from the current time t in step 94, and returns to step 96 when the current time reaches t + Δt. As a result, the vehicle position information is acquired and the variance is calculated every predetermined time Δt.

  FIG. 5 schematically shows changes in the dispersion of the landmark positions and changes in the dispersion of the own vehicle positions when traveling on the same traveling path a plurality of times. In the first run, the accuracy of the vehicle position is improved during landmark observation. In addition, the accuracy of the landmark position and the vehicle position is improved as the vehicle travels the same traveling path a plurality of times.

  When a camera is used as the landmark sensor, not only the landmark position information but also additional information related to the landmark height information, color, and size may be registered.

When the landmark is detected, the position of the landmark is captured in step 102, and in step 104, information indicating the relative positional relationship between the landmark detected by the landmark sensor and the landmark registered in the storage device is obtained. It is used to determine whether or not the landmark is matched. If the landmark is matched, the landmark position variance P _mm (t) is calculated in step 108, and the following (2) is calculated in step 110. According to the equation, the variance P (t) of the vehicle position and the landmark position is calculated.

  When the landmark is detected, the position variance of the landmark newly detected by the variance of the own vehicle and the error variance of the sensor is expressed by the following equation (1).

  Accordingly, if the off-diagonal term is the correlation between the vehicle position and the landmark position, the variance P (t) of the vehicle position and the landmark position is expressed by the following equation (2).

  In the next step 112, the accuracy of the landmark position information is determined by determining whether the variance P (t) of the vehicle position and the landmark position is less than the threshold Th, and the variance P (t) Is equal to or greater than the threshold value Th and the accuracy of the landmark position information is low, step 94 returns to step 96 after the elapse of a predetermined time Δt, and the vehicle position information is taken in as described above, and the accuracy of the landmark position information is obtained. Is high, after executing the process of updating the position information of the landmarks registered in step 114 and the process of updating the travel locus registered based on the vehicle position information, in step 116, Predict and predict road shapes (eg guardrail arrangement, curb arrangement, etc.) based on landmark location information and travel trajectory. Updating the map information based on the road edges as road shape.

  Next, the details of the matching process in step 104 will be described. In the present embodiment, landmark matching is performed by limiting the search range as shown in FIG. Although it is possible to search the entire range without limiting the search range, in reality, the vehicle position detected by the vehicle position sensor and the error range assumed for the vehicle position sensor are considered. In addition, it is preferable that the searchable range is a range in which the landmark can be detected from the performance of the mounted landmark sensor.

When a GPS device with an accuracy of 30 m is used as the own vehicle position sensor, as shown in FIG. 7 (1), the current position (X _t , Z _t ) of the own vehicle is used as a reference, and the long axis direction indicates the traveling direction. The facing elliptical region S is set as a range where the own vehicle may exist, that is, the own vehicle position error range. This own vehicle position error range can be represented by an ellipse having a length that is approximately the sum of the traveling distance of 70 m and the traveling distance for the second speed in the traveling direction, and a distance that is approximately 30 m in the lateral direction. Further, as shown in FIG. 7B, the search range W is the sum of the range in which the landmark can be detected from the vehicle position error range and the absolute position error of the registered information.

  When a sensor that detects information based on past history such as a gyro sensor is used as the vehicle position sensor, the vehicle's existence possibility range, that is, the vehicle position error, is determined from the error range corresponding to the time of position specification. Determine the range. This is because errors accumulate as time increases from the initial state.

In step 120, the vehicle position (Xt, Zt) and the vehicle position error range (σa, σb) in the X-axis and Z-axis directions are acquired. In step 122, k landmarks are detected, and each detected Landmark position information (r ₀ , θ ₀ ), (r ₂ , θ ₂ ),... (R _r , θ _r ) is acquired. Assume that the positions of the respective landmarks in the detected k landmark rows are (S ₁ , S ₂ , S ₃ ,... S _k ). In this case, an arbitrary number of landmarks may be detected.

  In the next step 124, the detection range (Fov = ± Φs, range = minRs to maxRs) stored in advance in the storage device is taken in, and the search range W on the absolute coordinates is determined.

In the next step 126, absolute positions (X _i , Z _i ), (X _{i + 1} , Z _{i + 1} ) · N of N landmarks existing within the search range W from the landmarks registered in the storage device. .. (X _{i + N} , Z _{i + N} ) are extracted, k registered landmarks are extracted from N registered landmarks in step 128, and the extracted k registered landmarks are registered landmarks A column. An example of a landmark row is shown in FIG.

In this case, when k registration landmarks are extracted from N registration landmarks, there are _N C _k combinations, and therefore _N C _k sets of registration landmark strings are extracted. The position of each landmark in the registered landmark row of each set is _defined as (M ₁ , M ₂ , M ₃ ,... M _k ).

In the next step 130, the positional relationship between the currently acquired landmark and the registered landmark, that is, the shape of the currently acquired landmark row and the shape of the registered landmark row is expressed by the following evaluation function. Use and compare. As the evaluation function, the Mahalanobis distance MD ² between the landmarks is used, and the jth set of registered landmark strings (M ₁ , M ₂ , M ₃₎ among the _N C _k registered landmark strings. ,... M _k ) and the Mahalanobis distance MD ² between corresponding landmarks between the detected landmark row (S ₁ , S ₂ , S ₃ ,... S _k ). An example of distance evaluation is shown in FIG.

In step 134, it is determined whether the calculated minimum value minE _j For _N C _k-number of sets of all, if not calculated minimum minE _j For _N C _k-number of sets all to step 132 The calculation of the evaluation function is repeated until the minimum value minE _j is calculated for all the _N C _k combinations. Thus, the minimum value is calculated for each set.

  When the landmarks are arranged at equal intervals, or when the difference between evaluation functions is unlikely to occur, such as when there is only one detected landmark, the landmark attributes (height, color, large) Etc.), landmark attributes may be detected by a sensor, and an evaluation function using these attributes may be obtained.

_{When the} minimum value minE _j is calculated for all of the _N C _k sets, in step 136, the number of registered landmark sequences that minimizes the minimum value minE _j is determined, and the set determined in step 138 is determined. It is determined that the landmark row matched with the detected landmark row is a landmark row.

  As described above, according to the present embodiment, it is possible to acquire high-accuracy road shape information (center and road edge) using a travel locus and update map information with high accuracy. Also, the landmark can always be updated.

It is a block diagram which shows embodiment of this invention. It is a diagram which shows the relationship between the density of a landmark, and own vehicle position dispersion | distribution. It is a diagram which shows the relationship between the density of a landmark, and the own vehicle position error. It is a flowchart which shows the routine which updates the map information of this Embodiment. It is a schematic diagram which shows the change of the dispersion | distribution of the dispersion | distribution of the position of the own vehicle when traveling on the same traveling path, and the dispersion | distribution of the position of a landmark. It is a flowchart which shows the routine which judges whether registration information and the present acquisition information matched. It is explanatory drawing explaining the procedure which judges whether registration information and the present acquisition information matched, (1) is a figure which shows the present own vehicle position and the own vehicle position error range, (2) is The figure which shows the search range based on sensor performance, (3) is a figure which shows the relationship between the registered landmark row | line | column and the landmark row | line | column acquired by detection, (4) is for demonstrating the distance evaluation between landmark row | line | columns. FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Own vehicle position sensor 12 Landmark sensor 14 Control apparatus 16 Memory | storage device

Claims (2)

Own vehicle position detecting means for detecting the position of the own vehicle;
Mark detection means for detecting a position of a mark around the host vehicle as a relative position with respect to the host vehicle;
Own vehicle position accuracy calculating means for calculating the accuracy of the position of the own vehicle from variations in the position of the own vehicle detected by the own vehicle position detecting means;
Detection accuracy acquisition means for acquiring detection accuracy of the mark detection means;
Based on the accuracy of the position of the own vehicle calculated by the own vehicle position accuracy calculating means and the detection accuracy of the mark detecting means acquired by the detection accuracy acquiring means, the position of the mark detected by the mark detecting means is calculated. Mark position accuracy calculating means for calculating accuracy,
Storage means for storing map information including road shapes and positions of a plurality of landmarks;
A matching determination unit that determines whether or not the positions of the plurality of landmarks detected by the landmark detection unit match the positions of the plurality of landmarks stored in the storage unit;
By comparing the mark position accuracy calculated by the mark position accuracy calculation means with a predetermined threshold for the positions of a plurality of marks determined to be matched by the matching determination means , the position of the mark When it is determined that the accuracy is increased, the positions of the plurality of landmarks stored in the storage unit are updated to the positions of the plurality of landmarks that are increased in accuracy, and the position of the host vehicle is updated from the updated positions of the plurality of landmarks. Updating means for estimating a position and updating the road shape stored in the storage means to a road shape according to the estimated position of the host vehicle;
A map update device. Before SL updating means, the accuracy of the position of the mark represented by the variation in the positions of a plurality of landmarks are determined to be matched by the matching determination means, and the position of the mark that has been calculated by the mark position accuracy calculation means The map updating apparatus according to claim 1, wherein a determination is made as to whether or not the accuracy of the position of the mark is increased by comparing a value using the accuracy of the mark with a predetermined threshold.

Images