JP6916705B2 - Self-driving vehicle position detector - Google Patents

Description

The present invention relates to an autonomous vehicle position detection device capable of improving the detection accuracy of the vehicle position without damaging the multiplex system by a plurality of sensor units for detecting the vehicle position.

Conventionally, in this type of automatic driving, in-vehicle sensor units have been multiplexed for the purpose of ensuring safety and high reliability. Even if one sensor unit fails due to multiplexing of the sensor units, fail-safe can be realized by interpolating with the other sensor unit.

At that time, it is also conceivable to mutually utilize the multiplexed sensor units to improve the detection accuracy. However, when the multiplexed sensor units are mutually used, if one sensor unit fails, the reliability of the other sensor unit cannot be ensured and the multiplex system is impaired.

For example, when the locator unit and the front recognition means are mounted on the own vehicle, the locator unit estimates the position of the own vehicle on the road map data based on the positioning signal from GNSS ((Global Navigation Satellite System) hygiene). The front recognition means estimates the position of the own vehicle on the traveling lane from the traveling environment in front of the own vehicle acquired by a unit that combines a stereo camera unit or a monocular camera and a laser radar for the traveling environment in front of the own vehicle.

In this case, in the lane keeping control during automatic driving, the vehicle position detection accuracy is corrected by correcting the vehicle position on the road map detected by the locator unit with the vehicle position on the traveling lane estimated by the forward recognition means. Can be enhanced. However, if one of the two interdependent sensors fails, it becomes difficult to ensure safety and reliability in automatic operation using only the other sensor, and the multiplex system is impaired.

Therefore, for example, as disclosed in Patent Document 1 (Japanese Unexamined Patent Publication No. 2003-15742), a multiplex system in which sensor units for detecting the same event are completely independent without interdependence is always secured. It is necessary to keep it.

Japanese Unexamined Patent Publication No. 2003-15742

However, in the above-mentioned technology for estimating the position of the own vehicle, if the information acquired by each sensor unit can be mutually used without damaging the multiplex system of the sensor units, in other words, without interdependence, during automatic driving. It becomes possible to detect the position of the own vehicle in the vehicle with higher accuracy.

In view of the above circumstances, the present invention makes it possible to mutually use the information of each sensor unit without damaging the multiplex system of the sensor units that detect the same event, and detect the position of the own vehicle during automatic driving with high accuracy. It is an object of the present invention to provide a self-driving vehicle position detection device capable of automatic driving.

The present invention compares a plurality of sensor units that detect the same event on the traveling path of the own vehicle on which the own vehicle is traveling with the same event detected by each of the sensor units, and determines the reliability in which the difference is preset. When the reliability determination means determines that the vehicle is reliable when it is within the threshold value, and when the reliability determination means determines that the vehicle is reliable, the vehicle position is based on the same event detected by each of the sensor units. In the self-driving own vehicle position detecting device provided with the own vehicle position verification means for verifying the above, the same event is the road curvature of the own vehicle traveling path, and the reliability determination means is from the current time to a predetermined past time. When the difference in road curvature detected by each of the sensor units is within the reliability determination threshold value, each sensor unit is determined to be reliable, and the reliability determination means determines that the vehicle is reliable. , The lateral position of the own vehicle position in the vehicle width direction with reference to the detection position of the road curvature detected by one of the plurality of sensor units in the preset near-past time on the current time side of the predetermined past time. The vehicle lateral position calculation means for obtaining the vehicle lateral position deviation which is the difference from the vehicle is further provided, and the vehicle position verification means is the vehicle position on the road curvature detected by the other sensor unit. Is corrected by the own vehicle lateral position deviation obtained by the own vehicle lateral position calculation means, and the own vehicle position at the current time is recalculated based on the corrected own vehicle position.

According to the present invention, when it is determined that a plurality of sensor units have reliability, the position of the own vehicle detected by one sensor unit in the near past time close to the current time is obtained by another sensor unit. Since the position of the vehicle in the near past time is corrected by the lateral position deviation of the vehicle and the position of the vehicle in the current time is recalculated based on the corrected position of the vehicle, multiple sensor units for detecting the road curvature, which is the same event, are used. The position of the own vehicle can be obtained with high accuracy by mutually using the information of each sensor unit without damaging the system.

Schematic configuration of the vehicle position detection device Flowchart showing vehicle position detection routine Time chart showing the acquisition timing of camera information and map information Explanatory drawing showing a state in which the own vehicle is driven along the road curvature on the road map by automatic driving. Explanatory drawing showing a mode in which the road curvature is corrected by the camera curvature acquired from the camera information when the own vehicle is driven along the road curvature on the road map by automatic driving. Time chart showing the difference between map curvature and camera curvature

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. The own vehicle position detection device 1 shown in FIG. 1 is mounted on the own vehicle M (see FIG. 4). The own vehicle position detecting device 1 has a locator unit 11 and a camera unit 21 as sensor units for detecting the road curvature which is the same event, and the two units 11 and 21 are completely independent multiplex without being dependent on each other. It constitutes a system.

The locator unit 11 estimates the position of the own vehicle on the road map, while the camera unit 21 recognizes the lane marking that divides the left and right of the traveling lane of the own vehicle M, and uses the center of this lane marking as a reference. The lateral position deviation of the own vehicle M in the vehicle width direction is detected.

The locator unit 11 has a locator calculation unit 12 for estimating the position of the own vehicle and a high-precision road map database 18 as a storage means. The locator calculation unit 12, the vehicle horizontal position calculation unit 21d as the vehicle horizontal position calculation means described later, and the travel control unit 22 are composed of a well-known microcomputer provided with a CPU, RAM, ROM, and the like, and peripheral devices thereof. The ROM stores in advance fixed data such as a program executed by the CPU and a base map.

On the input side of the locator calculation unit 12, a front-rear acceleration sensor 13 that detects the front-rear acceleration acting on the own vehicle M, a wheel speed sensor 14 that detects the rotation speed of each of the front-rear, left-right, and left wheels, and an angular velocity or angular acceleration of the own vehicle M are applied. Sensors required for estimating the position (own vehicle position) of the own vehicle M, such as the gyro sensor 15 for detection and the GNSS receiver 16 for receiving positioning signals transmitted from a plurality of positioning satellites, are connected. ..

The locator calculation unit 12 includes a vehicle position estimation unit 12a and a map information acquisition unit 12b as a vehicle position estimation means as functions for estimating the vehicle position. Further, the high-precision road map database 18 is a large-capacity storage medium such as an HDD, and high-precision road map information (dynamic map) is stored. This high-precision road map information holds lane data (lane width data, lane center position coordinate data, lane travel direction angle data, speed limit, etc.) required for automatic driving, and this lane data Are stored at intervals of several meters in each lane on the road map.

The map information acquisition unit 12b acquires and acquires the route map information from the current location to the destination from the map information stored in the high-precision road map database 18 based on the destination set by the driver during automatic driving, for example. The route map information (lane data on the route map) is transmitted to the own vehicle position estimation unit 12a. The own vehicle position estimation unit 12a acquires the position coordinates of the own vehicle M based on the positioning signal received by the GNSS receiver 16, map-matches the position coordinates on the route map information, and maps the own vehicle position on the road map. Is estimated, the driving lane is specified, and the road curvature at the center of the driving lane stored in the road map data is acquired.

Further, in an environment where the vehicle position estimation unit 12a cannot receive an effective positioning signal from the positioning satellite due to a decrease in sensitivity of the GNSS receiver 16 such as traveling in a tunnel, the wheel detected by the wheel speed sensor 14 The vehicle position on the road map is estimated by switching to autonomous navigation that estimates the vehicle position based on the vehicle speed obtained based on the speed, the angular velocity detected by the gyro sensor 15, and the front-rear acceleration detected by the front-rear acceleration sensor 13.

On the other hand, the camera unit 21 is an in-vehicle camera (stereo) composed of a main camera 21a and a sub camera 21b which are fixed to the upper center of the front part of the vehicle interior and are arranged symmetrically with respect to the center in the vehicle width direction. It has a camera), an image processing unit (IPU) 21c, and a vehicle lateral position calculation unit 21d. The camera unit 21 performs predetermined image processing on the front traveling environment image information in front of the own vehicle M captured by both cameras 21a and 21b by the IPU 21c, and transmits the image processing to the own vehicle lateral position calculation unit 21d.

Based on the received forward driving environment image information of the own vehicle M, the own vehicle lateral position calculation unit 21d divides the left and right of the traveling path (own vehicle traveling path) on which the own vehicle M travels, and the road curvature [1 / m] and the width (vehicle width) between the left and right lane markings. There are various known methods for obtaining the road curvature and the vehicle width. For example, the road curvature is binarized by the brightness difference based on the forward driving environment image information to recognize the left and right lane markings, and the minimum two. The curvature of the left and right lane markings is obtained for each predetermined section using a curve approximation formula based on the multiplication method, and the vehicle width is calculated from the difference in curvature between the two lane markings.

Then, the road curvature at the center of the lane (referred to as "camera curvature" in this embodiment) is obtained based on the curvature of the left and right section lines and the lane width, and further, the side of the own vehicle M with the center of the lane as a reference. The position deviation, to be exact, the distance from the center of the lane to the center of the own vehicle M in the vehicle width direction (own vehicle lateral position deviation Xdiff) is calculated.

The own vehicle position estimated by the own vehicle position estimation unit 12a of the locator calculation unit 12 and the own vehicle lateral position deviation Xdiff obtained by the own vehicle lateral position calculation unit 21d of the camera unit 21 are read by the traveling control unit 22. The locator unit 11 and the camera unit 21 form a completely independent multiplex system.

The travel control unit 22 constantly compares the lateral position of the vehicle position estimated by the vehicle position estimation unit 12a from the center of the lane on the road map with the vehicle lateral position obtained by the vehicle lateral position calculation unit 21d. .. When the difference (absolute value) exceeds a preset threshold value, either the own vehicle position estimated by the own vehicle position estimation unit 12a or the own vehicle lateral position obtained by the own vehicle lateral position calculation unit 21d. It is determined that the reliability of the vehicle is low, the automatic operation is stopped, and the driver is notified to that effect via a notification means (not shown).

When the automatic driving is stopped, the travel control unit 22 executes lane keeping control for the vehicle M to travel along the center of the lane based on the vehicle lateral position obtained by the vehicle lateral position calculation unit 21d of the camera unit 21. At that time, if the camera unit 21 cannot recognize one or both of the left and right section lines and determines that it is difficult to continue the lane keeping control, the traveling control unit 22 stops the lane keeping control and manually. After notifying the driver that the vehicle will be switched to driving, the lane keeping control is stopped.

By the way, the locator unit 11 and the camera unit 21 have the lateral position information (own vehicle lateral position information) of the own vehicle M at the same time. Therefore, if the vehicle lateral position estimated by the locator unit 11 is corrected by the vehicle lateral position information obtained by the camera unit 21, the vehicle position can be estimated with high accuracy, but the sensor units are multiplexed. It becomes difficult to maintain the system.

Therefore, the above-mentioned own vehicle position estimation unit 12a is the own vehicle lateral position information before (past) the current own vehicle lateral position information obtained by the own vehicle lateral position calculation unit 21d, and is similarly obtained before (past). By correcting the position of the own vehicle, the detection accuracy of the position of the own vehicle is improved while maintaining the complete independence of the multiple systems.

Specifically, the own vehicle position obtained by the own vehicle position estimation unit 12a described above is obtained by the own vehicle position detection routine shown in FIG. In this routine, first, the own vehicle position information received by the GNSS receiver 16 is read in step S1, and then the own vehicle position is map-mapped to the route map information acquired by the map information acquisition unit 12b in step S2. After that, the process proceeds to step S3, and the road curvature (hereinafter referred to as “map curvature”) RMPU [1 / m] on the road map of the own vehicle position map-matched to the route map information is read.

Next, the process proceeds to step S4, and the road curvature (hereinafter, “camera curvature”) RCAM [1 / m] of the own vehicle traveling path obtained by the own vehicle lateral position calculation unit 21d of the camera unit 21 is read. In the present invention, the actual road curvature is the same event detected by the locator unit 11 and the camera unit 21, and the map curvature RMPU detected by each of the units 11 and 21 and the camera curvature RCAM detect the same event. Become.

Then, the process proceeds to step S5, and both curvatures RMPU (t) to RMPU (t-) obtained for each calculation period Δt (see FIG. 3) in the section from the present (t) to the predetermined long-past time (t-x'). x'), RCAM (t) to RCAM (t-x') are read, and their respective curvature differences ΔR (t) to ΔR (t-x') are calculated. At that time, assuming that the own vehicle M is traveling along the map curvature RMPU in the center of the lane, the lane in which the own vehicle M is presumed to be in the center of the lane and is traveling is determined by the image information from the camera unit 21. The lateral position is deviated from the acquired actual center of the lane (camera curvature RCAM) by the curvature difference ΔR. In the present embodiment, the far past time (t-x') is set 3 [sec] before the current time (t), but the present embodiment is not limited to this.

Next, the process proceeds to step S6, and the reliability determination preset as the absolute value | ΔR (t) | to | ΔR (t-x') | of each curvature difference ΔR (t) to ΔR (t-x') | Compare with the threshold ΔRo. The reliability determination threshold value ΔRo determines the reliability of the camera unit 21, and is ^{set to ± ΔRo = ± 3 ・ 10 -4} in the present embodiment, but the present invention is not limited to this. ..

As shown in FIG. 3, the map curvatures RMPU (t) to RMPU (t-x') and RCAM (t) to RCAM (t-x') are sequentially fed with the previous values at each calculation cycle Δt, and the map curvature RMPU. The latest road curvature is incorporated into (t) and RCAM (t). Therefore, this reliability determination is sequentially executed every calculation cycle Δt.

Then, when the absolute value | ΔR (t) | to | ΔR (t-x') | of all the curvature differences in the determination intervals (t) to (t-x') is equal to or less than the reliability determination threshold value ΔRo (| ΔR). (t) | ~ | ΔR (t-x') | ≤ΔRo), and the process proceeds to step S7. If | ΔR (t) | to | ΔR (t-x') | <ΔRo), the routine is exited.

Therefore, as shown by the solid line in FIG. 4, in the determination interval (t) to (t-x'), all the curvature differences ΔR (t) to ΔR (t-x') are within the reliability determination threshold ± ΔRo. If it fits, it is determined that the reliability of the camera unit 21 is high. On the other hand, if a curvature difference ΔR outside the reliability determination threshold ± ΔRo is detected within the determination interval (t) to (t-x') as shown by the alternate long and short dash line in the figure, it is determined to be NG. , Exit the routine as it is. The processing in steps S5 and S6 described above corresponds to the reliability determination means of the present invention.

By the way, when trying to accurately detect the position of the own vehicle, the current position of the own vehicle on the road map is deviated by the lateral position of the own vehicle M from the center of the lane obtained based on the image captured by the camera unit 21. Therefore, it should be corrected at this horizontal position. However, as described above, when the current position of the vehicle on the road map is corrected by the horizontal position detected by the camera unit 21, the control system is integrated and it becomes difficult to maintain a completely independent multiplex system. ..

Therefore, in the present embodiment, in steps S7 to S9, the own vehicle position is verified based on the map curvature RMPU and the camera curvature RCAM, and the correction of the own vehicle position is performed by the map curvature RMPU and the camera curvature detected in the past time (tx). It is based on RCAM. As a result, a completely independent multiplex system of both units 11 and 21 is maintained. The processing in steps S7 to S9 corresponds to the own vehicle position verification means of the present invention.

That is, first, in step S7, the own vehicle lateral position deviation Xdiff of the past time (tx) obtained by the own vehicle lateral position calculation unit 21d (hereinafter referred to as "near past time (tx)") is read. (tx) is the time based on the present time (t), which is on the present side of the above-mentioned far past time (t-x') (tx) <(t-x')). Incidentally, in the present embodiment, the near past time (t-x) is set 1 [sec] before the current time (t), but the present embodiment is not limited to this.

Then, when the process proceeds to step S8, the position of the own vehicle on the road map in the near past time (t-x) is corrected by the lateral position deviation Xdiff of the own vehicle. Since the curvature difference ΔR between the map curvature RMPU and the camera curvature RCAM is within the reliability determination threshold ± ΔRo during the time (t) to (t-x') (see FIG. 4), both curvatures RMPU and RCAM are reliable. Highly sexual. Therefore, the reliability of the vehicle lateral position deviation Xdiff from the center of the lane in the near past time (tx) detected by the camera unit 21 is also high, and the vehicle position (vehicle side) on the road map by this vehicle lateral position deviation Xdiff. If the position) is corrected, a higher in-house position accuracy can be obtained.

After that, the process proceeds to step S9, the vehicle position at the current time (t) is recalculated based on the vehicle position on the road map corrected in the near past time (t-x), and the routine is exited. As a result, as shown in FIG. 6, the target travel path to be traveled by the own vehicle M is corrected from the near past time (t-x), and the lateral position deviation of the own vehicle at the current time (t) becomes a smaller value.

As described above, in the present embodiment, the curvature difference ΔR (t) to ΔR (t-x') of both curvatures RMPU and RCAM in the predetermined determination interval (t) to (t-x') is the reliability determination threshold ±. If it is within ΔRo, it is judged that the reliability of the camera curvature RCAM obtained based on the image from the camera unit 21 is high, and the position of the vehicle on the road map in the near past time (tx) is the lateral position of the vehicle. Since the position of the own vehicle on the road map of the current time (t) is obtained by correcting with the deviation Xdiff, the information of each unit 11 and 21 is obtained without damaging the multiplex system of the locator unit 11 and the camera unit 21. Can be used interchangeably, the position of the own vehicle can be obtained with higher accuracy, and lane keeping control and the like during automatic driving can be controlled with higher accuracy.

The present invention is not limited to the above-described embodiment. For example, the sensor unit for detecting the road curvature in a multiplex system includes three or more types of the locator unit 11, the camera unit 21, and other sensor units. It may have been done. In this case, the road curvatures detected by each sensor unit are compared with each other, and if all the differences are within the reliability determination threshold ± ΔR, all the sensor units are determined to be reliable.

1 ... Own vehicle position detection device,
11 ... Locator unit,
12 ... Locator calculation unit,
12a ... Own vehicle position estimation unit,
12b ... Map information acquisition department,
13 ... Front-back accelerometer,
14 ... Wheel speed sensor,
15 ... Gyro sensor,
16 ... GNSS receiver,
18 ... High-precision road map database,
21 ... Camera unit,
21a ... Main camera,
21b ... Sub camera,
21c ... Image processing unit,
21d ... Own vehicle horizontal position calculation unit,
22 ... Travel control unit,
M ... own vehicle,
RCAM ... Camera curvature,
RMPU ... Map curvature,
Xdiff ... Vehicle lateral position deviation,
ΔR ... Curvature difference,
ΔRo… Reliability judgment threshold Δt… Calculation cycle

Claims (3)

Multiple sensor units that detect the same event on the vehicle's travel path, and
A reliability determination means for comparing the same events detected by each sensor unit and determining that the difference is within the preset reliability determination threshold value, and
An automatic driving vehicle position detecting device including a vehicle position verification means for verifying the vehicle position based on the same event detected by each sensor unit when the reliability determination means determines that the vehicle is reliable. In
The same event is the road curvature of the own vehicle traveling path.
When the difference in road curvature detected by each of the sensor units from the current time to a predetermined past time is within the reliability determination threshold value, the reliability determination means determines that each sensor unit is reliable.
When the reliability determination means determines that the road curvature is reliable, the detection position of the road curvature detected by one of the plurality of sensor units in the preset near-past time on the current time side of the predetermined past time. Further, it has a vehicle lateral position calculation means for obtaining a vehicle lateral position deviation which is a difference between the vehicle position and the vehicle width direction lateral position based on the above.
The own vehicle position verification means corrects the own vehicle position on the road curvature detected by the other sensor unit with the own vehicle lateral position deviation obtained by the own vehicle lateral position calculation means, and corrects the correction. An automatic driving vehicle position detecting device, which recalculates the vehicle position at the current time based on the vehicle position. The plurality of sensor units are a camera unit and a locator unit.
The camera unit has an in-vehicle camera and the vehicle lateral position calculation means.
The own vehicle lateral position calculation means acquires the central curvature between the lane markings that section the left and right sections as the road curvature based on the driving environment image information in front of the own vehicle captured by the in-vehicle camera, and also acquires the curvature between the lane markings as the road curvature. Calculate the lateral position deviation of the own vehicle with reference to the center,
The locator unit identifies the traveling lane by map-matching the own vehicle position obtained based on the positioning signal from the positioning satellite on the road map data, and also identifies the traveling lane of the traveling lane stored in the road map data. It has a vehicle position estimation means to acquire the road curvature,
The own vehicle position verification means is executed by the own vehicle position estimation means, and the own vehicle position verification means obtains the own vehicle position on the road curvature of the road map data by the own vehicle lateral position calculation means. The self-driving vehicle position detecting device according to claim 1, wherein the vehicle is corrected by the lateral position deviation of the vehicle. The vehicle position detection for automatic driving according to claim 1 or 2, wherein the predetermined past time is a time 3 seconds before the current time, and the near past time is a time 1 second before the current time. Device.

Images