Map update device, map update method, and computer program for updating map
The map updating device enhances map accuracy by detecting reference points from surrounding data and updating probability distributions, addressing inconsistencies in feature detection by sensors.
Patent Information
- Application Number
- JP2021128538
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-08-04
- Publication Date
- 2024-11-06
- Estimated Expiration
- 2041-08-04
AI Technical Summary
Sensors on vehicles may fail to detect all features around the vehicle, leading to inconsistencies in map information, such as discontinuous lane markings.
A map updating device that detects reference points from surrounding data, updates a probability distribution associated with each point to increase the likelihood of the point's existence, and generates a distribution map with reduced variance, using methods like maximum likelihood estimation or Bayesian updating.
Improves the accuracy of feature positions on maps by refining the probability distributions associated with detected reference points, ensuring consistent representation of road features.
Smart Images

Figure 0007578078000001 
Figure 0007578078000002 
Figure 0007578078000003
Abstract
Description
[Technical field]
[0001] The present disclosure relates to a map updating device, a map updating method, and a computer program for updating a map, which update a map based on surrounding data that represents features around a vehicle. [Background technology]
[0002] High-precision maps that an autonomous driving system refers to in order to control the autonomous driving of a vehicle are required to accurately represent information about features related to vehicle travel, such as lane markings that are provided on roads or around roads. Therefore, technology has been proposed that collects data representing features from vehicles that actually travel on roads.
[0003] For example, Patent Document 1 describes a reliability evaluation device that evaluates the reliability of a value (evaluation target value) related to a feature included in map information. The reliability evaluation device described in Patent Document 1 obtains measurement values for measuring the evaluation target value multiple times using a sensor. The reliability evaluation device described in Patent Document 1 then selects a reliability evaluation method according to the variance of the measurement values, and evaluates the reliability of the evaluation target value using the selected reliability evaluation method. [Prior art documents] [Patent documents]
[0004] [Patent Document 1] JP 2011-017989 A Summary of the Invention [Problem to be solved by the invention]
[0005] A sensor mounted on a vehicle cannot always properly detect all features around the vehicle. In other words, the data generated by the sensor may only detect some of the features around the vehicle. In this case, if the map information is updated based on only the detected features, inconsistencies in the map information, such as discontinuity of lane markings, may occur.
[0006] An object of the present disclosure is to provide a map updating device capable of improving the accuracy of the positions of features shown on a map. [Means for solving the problem]
[0007] The map updating device according to the present disclosure includes a detection unit that detects the positions of a plurality of reference points corresponding to features on a road along which the vehicle is traveling from surrounding data representing features around the vehicle, and an update unit that updates a probability distribution associated with each of the plurality of reference points, the probability of the reference point being present for each position, so as to increase the probability that the reference point is present at the position of the detected reference point.
[0008] It is preferable that the map updating device according to the present disclosure further includes a generation unit that generates a delivery map including, among the multiple reference points, reference points whose variance in an associated probability distribution is smaller than a variance threshold, and a distribution unit that distributes the generated delivery map to the vehicle.
[0009] In the map updating device according to the present disclosure, it is preferable that the features are lane markings that divide lanes, and that the multiple reference points are points on the lane markings that are discrete at predetermined intervals with a predetermined point as a base point.
[0010] In the map updating device according to the present disclosure, when the position of a reference point on one of a pair of lane markings that divide a lane of interest among a plurality of lanes is detected from the surrounding data, but the position of a reference point on the other lane marking that divides the lane is not detected, it is preferable that the update unit updates the probability distribution associated with the reference point on the other lane marking, assuming that a reference point on the other lane marking has been detected at a position spaced from the position of the one lane marking by the width associated with the lane.
[0011] In the map updating device of the present disclosure, it is preferable that when the updating unit determines that the position of a reference point detected from a predetermined number or more of the surrounding data among the positions of the reference points respectively detected from the multiple surrounding data has a significant difference from the probability distribution associated with the reference point, the updating unit updates the probability distribution associated with the reference point.
[0012] The map generation method according to the present disclosure includes detecting the positions of a plurality of reference points corresponding to features on a road along which the vehicle is traveling from surrounding data representing features around the vehicle, and updating a probability distribution associated with each of the plurality of reference points, the probability of the reference point being present at each position, so as to increase the probability that the reference point is present at the position of the detected reference point.
[0013] The computer program for updating a map according to the present disclosure causes a computer processor to detect the positions of a plurality of reference points corresponding to features on a road on which the vehicle is traveling from surrounding data representing features around the vehicle, and update a probability distribution associated with each of the plurality of reference points, which indicates the probability of the reference point being present at each position, so that the probability of the reference point being present at the position of the detected reference point is increased.
[0014] According to the map updating device according to the present disclosure, it is possible to improve the accuracy of the positions of features shown on a map. [Brief description of the drawings]
[0015] [Figure 1] FIG. 2 is a hardware configuration diagram of the map updating device. [Diagram 2] FIG. 2 is a functional block diagram of a processor included in the map updating device. [Diagram 3] FIG. 13 is a schematic diagram for explaining an overview of updating of a reliability distribution. [Figure 4] 13 is a flowchart of a map update process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0016] A map updating device capable of improving the accuracy of the positions of features shown on a map will be described in detail below with reference to the drawings. The map updating device updates a map stored in a storage device using surrounding data showing features around a vehicle. In the map, a probability distribution indicating the probability of the reference point being present for each position is associated with each of a plurality of reference points corresponding to features on a road. The map updating device detects the positions of the reference points from the surrounding data. Then, the map updating device updates the probability distribution associated with the reference points so that the probability of the reference points being present at the positions of the detected reference points is increased.
[0017] 1 is a diagram showing a hardware configuration of the map updating device 1. The map updating device 1 includes a communication interface 11, a storage device 12, a memory 13, and a processor .
[0018] The communication interface 11 is an example of a communication unit, and has an interface circuit for connecting the map updating device 1 to a communication network. The communication interface 11 is configured to be able to communicate with other devices via the communication network. That is, the communication interface 11 passes data and the like received from other devices via the communication network to the processor 14. In addition, the communication interface 11 transmits data and the like received from the processor 14 to other devices via the communication network.
[0019] The storage device 12 is an example of a storage unit, and includes a storage device such as a hard disk device or a non-volatile semiconductor memory device. The storage device 12 stores a map including a plurality of reference points corresponding to features on roads.
[0020] A reference point is a point that is set corresponding to a feature to indicate the position of the feature. For example, on a map, a reference point is set at a predetermined point such as an intersection to indicate the position of a lane marking. Also, on the map, points on the lane marking that are separated at a predetermined interval (e.g., 10 m) from the predetermined point are set as a plurality of reference points. The predetermined interval is not limited to a fixed interval, and it is sufficient that the distance from a certain reference point to a predetermined point is determined in advance.
[0021] The storage device 12 also stores, for each reference point, a probability distribution (hereinafter referred to as a "reliability distribution") indicating the probability that the reference point exists for each position. The reliability distribution can be a normal distribution corresponding to a position on a two-dimensional plane along the road surface. The reliability distribution may also be a normal distribution corresponding to a position in a three-dimensional space.
[0022] The memory 13 includes a volatile semiconductor memory and a non-volatile semiconductor memory. The memory 13 temporarily stores various data used in processing by the processor 14, such as data received via the communication interface 11. The memory 13 also stores various application programs, such as a map update program for updating the map stored in the storage device 12.
[0023] The processor 14 includes one or more central processing units (CPUs) and their peripheral circuits, and may further include other arithmetic circuits such as a logic arithmetic unit or a numeric arithmetic unit.
[0024] FIG. 2 is a functional block diagram of the processor 14 included in the map updating device 1. As shown in FIG.
[0025] The processor 14 of the map updating device 1 has, as functional blocks, a detection unit 141, an update unit 142, a generation unit 143, and a distribution unit 144. Each of these units in the processor 14 is a functional module implemented by a computer program executed on the processor 14. A computer program that realizes the functions of each unit of the processor 14 may be provided in a form recorded on a computer-readable portable recording medium such as a semiconductor memory, a magnetic recording medium, or an optical recording medium. Alternatively, each of these units in the processor 14 may be implemented in the map updating device 1 as an independent integrated circuit, a microprocessor, or firmware.
[0026] The detection unit 141 detects the positions of a plurality of reference points corresponding to features on the road on which the vehicle (not shown) is traveling from surrounding data that indicates features around the vehicle.
[0027] The vehicle is equipped with a surrounding camera that captures the situation around the vehicle and outputs surrounding data. An ECU (Electronic Control Unit) equipped in the vehicle acquires the surrounding data from the surrounding camera and transmits the surrounding data to the map updating device 1 via a communication network including a wireless base station.
[0028] The vehicle may record the surrounding data in a portable computer-readable recording medium. The map updating device 1 can obtain the surrounding data by reading the recording medium using a medium reading device (not shown) connected to the communication interface 11.
[0029] The detection unit 141 detects the positions of features on the road by inputting the surrounding data to a classifier that has been trained in advance to identify features on the road, such as lane markings. For each feature shown on the map, the detection unit 141 calculates the distance from the position where the reliability of the feature is maximum, i.e., the average value in the reliability distribution, to the position of the feature detected from the surrounding data. Then, the detection unit 141 associates the feature detected from the surrounding data with the feature that has the smallest calculated distance among each feature shown on the map, that is equal to or smaller than a predetermined distance threshold, and that is the same type as the feature detected from the surrounding data.
[0030] The classifier can be, for example, a convolutional neural network (CNN) having multiple convolution layers connected in series from the input side to the output side. Images including features on roads to be detected are used as training data in advance, and the CNN is trained by a predetermined learning method such as the backpropagation method, so that the CNN operates as a classifier that detects features on roads.
[0031] In addition, for multiple reference points on the lane markings, the detection unit 141 detects, as the positions of the reference points, the positions of the lane markings detected at a specified point and the positions of the lane markings detected at points spaced apart from the specified point by a specified interval, based on the vehicle position corresponding to the surrounding data.
[0032] The update unit 142 updates the reliability distribution associated with each of the multiple reference points in the map stored in the storage device 12 so as to increase the probability that the reference point exists at the position of the detected reference point.
[0033] The update unit 142 updates the reliability distribution by a maximum likelihood estimation method. The update unit 142 divides the area included in the map where the reference point may exist into a plurality of sections, and accumulates the number of times the reference point is detected for each section. The update unit 142 then creates a reliability distribution that indicates the probability that the reference point exists for each section by dividing the number of times the reference point is detected for each section by the total number of times the reference point is detected in each section. When creating a reliability distribution corresponding to a position on a two-dimensional plane, the sections are set in a lattice pattern. The probability that the reference point exists at the position of the newly detected reference point is higher in the reliability distribution calculated in response to the detection of the reference point than in the reliability distribution before the update.
[0034] The update unit 142 may also update the reliability distribution by Bayesian updating. That is, for each of a plurality of sections of a predetermined range around a reference point in an area included in a map, a reliability that the reference point exists in the section is set. The same value may be set as an initial value of the reliability in each section, or a higher value may be set for a section with a higher possibility of the reference point existing. When the position of the reference point is detected from the peripheral data, the update unit 142 updates the reliability of each section so that the reliability of the section including the detected position of the reference point is increased. Alternatively, the update unit 142 may update the reliability of each section so that the reliability of each section within a predetermined range from the position of the reference point represented in the peripheral data is increased. At this time, the update unit 142 may increase the rate of increase of the reliability for a section closer to the position of the reference point. The update unit 142 calculates an updated reliability distribution for the position of the reference point by approximating the reliability of each section with a normal distribution, and stores the updated reliability distribution in the storage device 12.
[0035] Alternatively, the update unit 142 may set multiple candidates of the reliability distribution for the position of each reference point. In this case, each candidate may be a normal distribution represented by the average value and variance-covariance matrix of the position. The update unit 142 then calculates the posterior probability of each candidate for the position of the reference point represented in the peripheral data, and sets the posterior probability as the prior probability of each candidate for the next update. The update unit 142 sets the normal distribution corresponding to the candidate with the maximum prior probability as the reliability distribution for the position of the reference point.
[0036] FIG. 3 is a schematic diagram for explaining an overview of updating the reliability distribution.
[0037] The surrounding data SD indicates lane markings LL1, LL2, and LL3. The detection unit 141 detects, from the surrounding data SD, points on the lane marking LL1 that are spaced apart from a predetermined point by a predetermined interval as reference points DRP2.
[0038] Of lane markings LL1, LL2, and LL3 included in map M, reference points RP1 and RP2 are set at a predetermined interval on lane marking LL1. A reliability distribution is associated with each of reference points RP1 and RP2, and the prior distribution corresponding to reference point RP2 is represented by reliability distribution PD21. In the reliability distribution, the horizontal axis represents the left-right position on the road, and the vertical axis represents the probability that a reference point will be detected at that position.
[0039] 3, the reference point DRP2 detected from the peripheral data SD is located to the right of the reference point RP2 set on the map M. The map updating device 1 updates the reliability distribution PD21 associated with RP2 to a reliability distribution PD22 that increases the probability that the reference point RP2 exists at the position of the reference point DRP2.
[0040] The update unit 142 may determine whether or not the position of the reference point detected from a predetermined number or more of the peripheral data among the positions of the reference points detected from the multiple peripheral data is significantly different from the reliability distribution associated with the reference point. In this case, when it is determined that the position of the detected reference point is significantly different from the probability distribution associated with the reference point, the update unit 142 updates the reliability distribution associated with the reference point.
[0041] The generating unit 143 generates a delivery map including reference points whose variance in the associated reliability distribution is smaller than a variance threshold stored in advance in the memory 13, among a plurality of reference points included in the map stored in the storage device 12. In the reliability distribution corresponding to a two-dimensional plane or three-dimensional space, the variance threshold may be set for each direction of variance. The generating unit 143 may store the generated delivery map in the storage device 12.
[0042] The distribution unit 144 distributes the delivery map generated by the generation unit 143 to the vehicle via the communication interface 11 and the communication network. The autonomous driving system of the vehicle uses the distributed delivery map to control the autonomous driving of the vehicle.
[0043] Fig. 4 is a flowchart of the map update process. The processor 14 of the map update device 1 executes the map update process shown in Fig. 4 every time it receives surrounding data to be processed. The processor 14 of the map update device 1 may also execute the map update process shown in Fig. 4 every time it receives a predetermined number of surrounding data items equal to or greater than two.
[0044] First, the detection unit 141 of the processor 14 detects the positions of a plurality of reference points corresponding to features on the road on which the vehicle is traveling from surrounding data representing features around the vehicle (step S1).
[0045] Next, the update unit 142 of the processor 14 updates the reliability distribution associated with each of the multiple reference points so as to increase the probability that the reference point exists at the position of the detected reference point (step S2), and terminates the map update process.
[0046] By performing the map update process in this manner, the map updating device 1 can improve the accuracy of the positions of features shown on the map.
[0047] There are cases where the detection unit 141 detects the position of a reference point on one of a pair of lane lines that divide a lane of interest from the surrounding data, and does not detect the position of a reference point on the other lane line that divides the lane. According to a modified example, the update unit 142 updates the probability distribution associated with the reference point on one lane line, assuming that a reference point on the other lane line has been detected at a position spaced apart from the position of the one lane line by the width associated with the lane. The width may be stored in advance in the storage device 12 in association with the lane. When updating the probability distribution by Bayesian updating, the update unit 142 may express the likelihood of the reference point that is assumed to have been detected on the other lane line as a normal distribution in which the mean is the position spaced apart by the width from the position of the reference point detected on one lane line, and 1 / 2 or more of the width is the variance in a direction perpendicular to the other lane line. By updating the map in this manner, even if the position of a reference point on one lane marking cannot be detected, the position and width of the detected reference point on the other lane marking can be used to improve the accuracy of the positions of features shown on the map.
[0048] It should be understood that those skilled in the art can make various changes, substitutions, and alterations thereto without departing from the spirit and scope of the present disclosure. [Explanation of symbols]
[0049] 1 Map update device 141 Detection unit 142 Update Department 143 Generation part 144 Distribution Department
Claims
1. a detection unit that detects positions of a plurality of reference points corresponding to features on a road along which the vehicle is traveling from surrounding data representing features around the vehicle; an update unit that updates a probability distribution associated with each of the plurality of reference points, the probability of the reference point being present for each position, so that the probability of the reference point being present at the position of the detected reference point is increased; Equipped with the feature is a lane marking that marks a lane, and the plurality of reference points are points on the lane marking that are spaced apart at predetermined intervals from a predetermined point; When the position of a reference point on one of a pair of lane markings that divide a lane of interest among a plurality of lanes is detected from the surrounding data, and the position of a reference point on the other lane marking that divides the lane is not detected, the update unit updates the surrounding data so that a probability distribution associated with the reference point on the other lane marking becomes a normal distribution having a mean at a position spaced apart by the width associated with the lane from the position of the reference point detected on the one lane marking by the lane width and a variance in a direction perpendicular to the other lane marking by at least half of the width, and so that the probability that the reference point is present at a position spaced apart by the width from the position of the one lane marking becomes high. Map update device.
2. a generation unit that generates a distribution map including reference points, among the plurality of reference points, whose variance in the associated probability distribution is smaller than a variance threshold; The map updating device according to claim 1 , further comprising: a distribution unit that distributes the generated distribution map to a vehicle.
3. 3. The map updating device according to claim 1, wherein when it is determined that a position of the reference point detected from a predetermined number or more of the peripheral data among the positions of the reference point detected from the plurality of peripheral data has a significant difference from a probability distribution associated with the reference point, the updating unit updates the probability distribution associated with the reference point.
4. Detecting positions of a plurality of reference points corresponding to features on a road along which the vehicle is traveling from surrounding data representing features around the vehicle; updating a probability distribution associated with each of the plurality of reference points, the probability of the reference point being present for each position, so that the probability of the reference point being present at the position of the detected reference point is increased; the feature is a lane marking that marks a lane, and the plurality of reference points are points on the lane marking that are spaced apart at predetermined intervals from a predetermined point; In the updating, if the position of a reference point on one of a pair of lane markings that divide a lane of interest among a plurality of lanes is detected from the surrounding data, and the position of a reference point on the other lane marking that divides the lane is not detected, the update is performed so that a probability distribution associated with the reference point on the other lane marking becomes a normal distribution having a mean at a position spaced apart by the width associated with the lane from the position of the reference point detected on the one lane marking and a variance in a direction perpendicular to the other lane marking that is at least half of the width, and so that the probability that the reference point is located at a position spaced apart by the width from the position of the one lane marking becomes high. How to update maps.
5. Detecting positions of a plurality of reference points corresponding to features on a road along which the vehicle is traveling from surrounding data representing features around the vehicle; A computer program for updating a map, the computer program causing a processor of a computer to execute the following: updating a probability distribution associated with each of the plurality of reference points, the probability of the reference point being present for each position, so that the probability of the reference point being present at the position of the detected reference point is increased; the feature is a lane marking that marks a lane, and the plurality of reference points are points on the lane marking that are spaced apart at predetermined intervals from a predetermined point; In the updating, if the position of a reference point on one of a pair of lane markings that divide a lane of interest among a plurality of lanes is detected from the surrounding data, and the position of a reference point on the other lane marking that divides the lane is not detected, the update is performed so that a probability distribution associated with the reference point on the other lane marking becomes a normal distribution having a mean at a position spaced apart by the width associated with the lane from the position of the reference point detected on the one lane marking and a variance in a direction perpendicular to the other lane marking that is at least half of the width, and so that the probability that the reference point is located at a position spaced apart by the width from the position of the one lane marking becomes high. A computer program for updating maps.
Citation Information
Patent Citations
Device for extracting parallel lines
JP1993303638A
Device for detecting traveling path
JP1998040379A
Sensor fusion system and vehicle control unit using it
JP2005165421A
Positioning device
JP2008002906A
Road shape acquisition device
JP2008003253A
