Map updating device, map updating method and computer program for map updating

Abstract

This invention discloses a map updating apparatus, a map updating method, and a computer program for map updating. A map updating apparatus is provided that can improve the positional accuracy of features shown on a map. The map updating apparatus includes: a detection unit that detects the positions of multiple reference points corresponding to features on the road on which the vehicle travels, from surrounding data representing features around a vehicle; and an updating unit that updates the map by updating the position of each of the multiple reference points with a probability distribution that represents the probability of the reference point existing at each position, thereby increasing the probability that the reference point exists at the detected reference point's position.

Description

Technical Field

[0001] This disclosure relates to a map updating apparatus, a map updating method, and a computer program for map updating based on surrounding data representing features around a vehicle. Background Technology

[0002] In the high-precision maps referenced by autonomous driving systems for controlling vehicles, it is required to accurately represent information related to road features, such as lane markings and other features associated with vehicle movement. Therefore, a technique is proposed to collect data representing these features from vehicles actually traveling on roads.

[0003] For example, Patent Document 1 describes a reliability evaluation device for evaluating the reliability of values ​​(evaluation object values) related to land features contained in map information. The reliability evaluation device described in Patent Document 1 acquires multiple measurement values ​​for measuring the evaluation object values ​​using sensors. Furthermore, the reliability evaluation device described in Patent Document 1 selects a reliability evaluation method based on the variance of the measurement values, and evaluates the reliability of the evaluation object values ​​using the selected reliability evaluation method.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2011-017989 Summary of the Invention

[0007] Sensors mounted on a vehicle may not be able to adequately detect all features around the vehicle. That is, there may be situations where the data generated by the sensors can only detect a portion of the features around the vehicle. In such cases, if map information is updated only based on the detected features, inconsistencies in map information may sometimes occur, such as lane markings becoming discontinuous.

[0008] The purpose of this disclosure is to provide a map updating device that can improve the location accuracy of features shown on a map.

[0009] The map updating apparatus disclosed herein includes: a detection unit that detects the positions of multiple reference points corresponding to features on the road on which the vehicle travels from surrounding data representing features around the vehicle; and an updating unit that updates the map by increasing the probability distribution of the probability of the reference point existing at the detected reference point position, which is associated with each of the multiple reference points.

[0010] The map updating apparatus disclosed herein preferably further comprises: a generation unit for generating a distribution map of reference points, including a probability distribution of multiple reference points whose variance is less than a variance threshold; and a distribution unit for distributing the generated distribution map to vehicles.

[0011] In the map updating apparatus disclosed herein, preferably, the features are lane marking lines that divide the driving lanes, and the multiple reference points are points on the lane marking lines that are discrete at predetermined intervals with a predetermined location as a reference.

[0012] In the map updating apparatus disclosed herein, preferably, when a lane marking line is one of a pair of lane marking lines that divides a lane of interest among a plurality of lanes, and a reference point on one lane marking line is detected from surrounding data, but no reference point on the other lane marking line is detected, the updating unit considers that a reference point on the other lane marking line was detected at a position a distance away from the position of one lane marking line from the width associated with that lane, and updates the probability distribution associated with the reference point on the other lane marking line.

[0013] In the map updating apparatus disclosed herein, it is preferable that the updating unit updates the probability distribution associated with the reference point when it determines that there is an intentional difference between the position of a reference point detected from a predetermined number or more of the surrounding data representing the features around the vehicle and the probability distribution associated with that reference point.

[0014] The map updating method disclosed herein includes: detecting the positions of multiple reference points corresponding to the features on the road where the vehicle is traveling from surrounding data representing the features around the vehicle; and updating the map by using a probability distribution associated with each of the multiple reference points, representing the probability of the existence of the reference point at each position, in a way that increases the probability of the reference point existing at the detected reference point's position.

[0015] The computer program for map updating that is stored in a non-transitory computer-readable medium disclosed herein causes a processor to perform the following: detect the positions of multiple reference points corresponding to features on the road on which the vehicle is traveling from surrounding data representing features around the vehicle; and update the map in such a way that the probability distribution of the probability of the presence of the reference point at each of the multiple reference points is associated with each of the reference points, thereby increasing the probability of the reference point being present at the detected reference point's location.

[0016] According to the map updating apparatus disclosed herein, the positional accuracy of features shown on a map can be improved. Attached Figure Description

[0017] Figure 1 This is a hardware structure diagram of the map update device.

[0018] Figure 2 This is a functional block diagram of the processor in the map updating device.

[0019] Figure 3 This is a schematic diagram illustrating the summary of the updated reliability distribution.

[0020] Figure 4 This is a flowchart of the map update process.

[0021] (Symbol Explanation)

[0022] 1: Map updating device; 141: Detection unit; 142: Update unit; 143: Generation unit; 144: Distribution unit. Detailed Implementation

[0023] Hereinafter, with reference to the accompanying drawings, a map updating apparatus capable of improving the positional accuracy of features shown on a map will be described in detail. The map updating apparatus uses peripheral data representing features around a vehicle to update a map stored in a storage device. In the map, for each of multiple reference points corresponding to features on the road, a probability distribution representing the probability of the presence of that reference point at each location is associated. The map updating apparatus detects the location of a reference point from the peripheral data. Then, the map updating apparatus updates the probability distribution associated with that reference point in a manner that increases the probability of the reference point existing at the detected location.

[0024] Figure 1 This is a hardware structure diagram of the map updating device 1. The map updating device 1 has a communication interface 11, a storage device 12, a memory 13, and a processor 14.

[0025] Communication interface 11 is an example of a communication unit, having interface circuitry for connecting map update device 1 to a communication network. Communication interface 11 is configured to communicate with other devices via the communication network. Specifically, communication interface 11 sends data received from other devices via the communication network to processor 14. Additionally, communication interface 11 sends data received from processor 14 to other devices via the communication network.

[0026] Storage device 12 is an example of a storage unit, having a storage device such as a hard disk drive or a non-volatile semiconductor memory device. Storage device 12 stores a map including multiple reference points corresponding to features on the road.

[0027] Reference points are points set up to correspond to geographical features in order to indicate their location. For example, in a map, reference points are set at predetermined locations such as intersections to indicate the location of lane markings. Furthermore, in a map, points on lane markings discrete at predetermined intervals (e.g., 10 meters) based on these predetermined locations are set as multiple reference points. The predetermined interval is not limited to a fixed interval; the distance between a reference point and the predetermined location can be determined in advance.

[0028] Furthermore, the storage device 12 stores a probability distribution (hereinafter referred to as the "reliability distribution") representing the probability of the existence of a reference point at each location. The reliability distribution can be set as a normal distribution corresponding to the location on a two-dimensional plane along the road surface. Alternatively, the reliability distribution can also be a normal distribution corresponding to the location in three-dimensional space.

[0029] The memory 13 includes both volatile and non-volatile semiconductor memory. The memory 13 temporarily stores various data used in processing executed by the processor 14, such as data received via the communication interface 11. Additionally, the memory 13 stores various application programs, such as a map updater that updates a map stored in the storage device 12.

[0030] Processor 14 has one or more CPUs (Central Processing Units) and their peripheral circuitry. Processor 14 may also have other arithmetic circuitry such as logic units or numerical arithmetic units.

[0031] Figure 2 This is a functional block diagram of the processor 14 in the map update device 1.

[0032] In the processor 14 of the map update device 1, there are functional blocks including a detection unit 141, an update unit 142, a generation unit 143, and a distribution unit 144. These units of the processor 14 are functional modules installed by a computer program that executes on the processor 14. The computer program that implements the functions of each unit of the processor 14 may also be provided in the form of a computer-readable portable recording medium such as a semiconductor memory, magnetic recording medium, or optical recording medium. Alternatively, these units of the processor 14 may be installed as separate integrated circuits, microprocessors, or firmware in the map update device 1.

[0033] The detection unit 141 detects the positions of multiple reference points corresponding to the features on the road on which the vehicle is traveling, based on the surrounding data of features representing the area around the vehicle (not shown).

[0034] The vehicle is equipped with a surrounding camera that captures images of the vehicle's surroundings and outputs surrounding data. The ECU (Electronic Control Unit) installed in the vehicle acquires the surrounding data from the surrounding camera and sends the surrounding data to the map update device 1 via a communication network including a wireless base station.

[0035] Vehicles can also record surrounding data onto a computer-readable portable recording medium. The map update device 1 is able to acquire surrounding data by reading the recording medium using a medium reading device (not shown) connected to the communication interface 11.

[0036] The detection unit 141 detects the positions of road features by inputting surrounding data into a recognizer that has been pre-learned to identify road features such as lane markings. For each feature shown on the map, the detection unit 141 calculates the distance from the location where the reliability for that feature is the highest (i.e., the location that is the average value in the reliability distribution) to the location of the feature detected from the surrounding data. Then, the detection unit 141 identifies the features detected from the surrounding data that have the smallest calculated distance among the features shown on the map, that are below a predetermined distance threshold, and that are of the same type as the features detected from the surrounding data.

[0037] The recognizer can be configured, for example, as a convolutional neural network (CNN) with multiple convolutional layers connected in series from the input side to the output side. By using images of road features that are to be detected as teacher data in advance, the CNN is trained using a predetermined learning method such as backpropagation of errors, thereby acting as a recognizer for detecting road features.

[0038] In addition, the detection unit 141 detects the positions of the lane markings at predetermined locations and the positions of the lane markings at predetermined intervals based on the positions of vehicles corresponding to the surrounding data.

[0039] The updating unit 142 updates the map stored in the storage device 12 by associating the reliability distribution with each of the multiple reference points, in a way that increases the probability that the reference point exists at the location of the detected reference point.

[0040] The update unit 142 updates the reliability distribution using the maximum likelihood estimation method. The update unit 142 divides the area containing the map containing the probability of a reference point existing into multiple zones, and accumulates the number of times a reference point is detected for each zone. Then, the update unit 142 creates a reliability distribution representing the probability of a reference point existing for each zone by dividing the number of times a reference point is detected for each zone by the total number of times a reference point is detected in all zones. When creating a reliability distribution corresponding to a position on a two-dimensional plane, the zones are set in a grid pattern. Regarding the probability of a reference point existing at the location of a newly detected reference point, the reliability distribution calculated based on the detection of the reference point is higher than the reliability distribution before the update.

[0041] Alternatively, the update unit 142 can also update the reliability distribution using Bayesian updates. That is, the update unit 142 divides a predetermined range around a reference point within a region on the map into multiple zones, and sets a reliability threshold for the presence of a reference point in each zone. As the initial reliability value for each zone, either the same value can be set for all zones, or a higher value can be set for zones with a higher probability of the reference point existing. When the location of a reference point is detected from the surrounding data, the update unit 142 updates the reliability of each zone in a way that increases the reliability of the zone including the detected reference point's location. Alternatively, the update unit 142 can also update the reliability of each zone in a way that increases the reliability of each zone within a predetermined range from the location of the reference point shown in the surrounding data. In this case, the update unit 142 can also increase the rate of increase in reliability for zones closer to the reference point's location. The update unit 142 calculates the updated reliability distribution for the location of the reference point by approximating the reliability of each zone using a normal distribution, and stores it in the storage device 12.

[0042] Alternatively, the update unit 142 may set multiple candidates for the reliability distribution of the location of each reference point. In this case, each candidate can be set as a normal distribution represented by the average value and variance-covariance matrix of the location. Furthermore, the update unit 142 calculates the posterior probability of each candidate for the location of the reference point shown in the surrounding data, and uses this posterior probability as the prior probability of each candidate for the next update. The update unit 142 uses the normal distribution corresponding to the candidate with the largest prior probability as the reliability distribution of the location of the reference point.

[0043] Figure 3 This is a schematic diagram illustrating the summary of the updated reliability distribution.

[0044] In the surrounding data SD, lane marking lines LL1, LL2, and LL3 are represented. The detection unit 141 detects reference points DRP2 from the surrounding data SD at predetermined intervals from predetermined locations within the lane marking lines LL1.

[0045] In map M, which includes lane marking lines LL1, LL2, and LL3, reference points RP1 and RP2 are set at predetermined intervals within lane marking line LL1. A reliability distribution is assigned to each reference point 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 of the road, and the vertical axis represents the probability of detecting the reference point at that position.

[0046] exist Figure 3 In the example, the reference point DRP2 detected from the surrounding data SD is located to the right of the reference point RP2 set in the map M. The map updating device 1 updates the reliability distribution PD21 associated with RP2 to a reliability distribution PD22 such that the probability of the reference point RP2 existing at the location of the reference point DRP2 increases.

[0047] The update unit 142 can also determine whether there is an intentional discrepancy between the position of a reference point detected from multiple peripheral data sources and the reliability distribution associated with the reference point. In this case, if the update unit 142 determines that there is an intentional discrepancy between the position of the detected reference point and the probability distribution associated with the reference point, it updates the reliability distribution associated with that reference point.

[0048] The generation unit 143 generates a distribution map that includes reference points among multiple reference points stored in the map stored in the storage device 12, whose variance in the associated reliability distribution is smaller than a variance threshold pre-stored in the memory 13. In the reliability distribution corresponding to a two-dimensional plane or three-dimensional space, a variance threshold can be set for each direction of the variance. The generation unit 143 may also store the generated distribution map in the storage device 12.

[0049] The distribution unit 144 distributes the distribution map generated by the generation unit 143 to the vehicle via the communication interface 11 and the communication network. The vehicle's autonomous driving system uses the distributed distribution map to perform autonomous driving control of the vehicle.

[0050] Figure 4 This is a flowchart of the map update process. Whenever the processor 14 of the map update device 1 receives surrounding data of the object being processed, it executes... Figure 4The map update process is shown. Additionally, the processor 14 of the map update device 1 can also perform this process whenever it receives two or more predetermined amounts of surrounding data. Figure 4 The map update process is shown below.

[0051] First, the detection unit 141 of the processor 14 detects the positions of multiple reference points corresponding to the ground features on the road on which the vehicle is traveling from the surrounding data representing the ground features around the vehicle (step S1).

[0052] Next, the updating unit 142 of the processor 14 updates the map update process by associating the reliability distribution with each of the multiple reference points in a way that increases the probability that the reference point exists at the location of the detected reference point (step S2).

[0053] By performing map update processing in this way, map update device 1 can improve the positional accuracy of features shown on the map.

[0054] In a situation where the detection unit 141 detects the position of a reference point on one of a pair of lane marking lines that divide a lane of interest into multiple lanes from surrounding data, but does not detect the position of a reference point on the other lane marking line, the update unit 142 updates the probability distribution associated with the reference point on the other lane marking line, assuming that a reference point on the other lane marking line is detected at a position away from the position of the one lane marking line from the width associated with that lane. The width can also be pre-stored in the storage device 12 associated with the lane. When updating the probability distribution using Bayesian updates, the update unit 142 can also express the likelihood of a reference point detected on the other lane marking line as a normal distribution with the position away from the width of the reference point detected on the one lane marking line as the average, and the variance in a direction orthogonal to the other lane marking line as more than half of the width. By updating the map in this way, even if the location of a reference point on one side's lane marking line is not detected, the location and width of the reference point on the other side's lane marking line can be used to improve the accuracy of the location of features shown on the map.

[0055] It is intended that those skilled in the art understand that various changes, substitutions, and modifications can be made without departing from the spirit and scope of this disclosure.

Claims

1. A map updating device, comprising: The detection unit, based on surrounding data representing features around the vehicle, detects the positions of multiple reference points corresponding to features on the road the vehicle is traveling on; and The updating unit updates the probability distribution associated with each of the plurality of reference points, representing the probability of the existence of the reference point at each location, in a way that increases the probability of the reference point existing at the detected location. The aforementioned features are the lane marking lines used to delineate driving lanes, and the plurality of reference points are points on the lane marking lines that are discrete at predetermined intervals, using predetermined locations as references. When the updating unit detects the position of a reference point on one of the lane marking lines that divides a lane of interest into multiple lanes from the surrounding data, and does not detect the position of a reference point on the lane marking line of the other lane, it considers that a reference point on the lane marking line of the other lane was detected at a position away from the position of the lane marking line of the first lane marking line, within a width associated with that lane. The update unit then updates the probability distribution associated with the reference point on the other lane marking line in a manner that uses the position of the reference point detected on the lane marking line of the first lane marking line as the average position away from that width, and uses at least half of that width as the variance in a direction orthogonal to the lane marking line of the other lane marking line. If the updating unit determines that there is an intentional difference between the position of the reference point detected from multiple surrounding data representing the ground features around the vehicle and the probability distribution associated with the reference point, the updating unit updates the probability distribution associated with the reference point.

2. The map updating device according to claim 1, wherein, It also has: The generation unit generates a distribution map comprising reference points from the plurality of reference points whose variance in the associated probability distribution is less than a variance threshold; and The distribution unit distributes the generated distribution map to the vehicles.

3. A map updating method, comprising: From the surrounding data of features representing the area around the vehicle, the positions of multiple reference points corresponding to features on the road the vehicle is traveling on are detected. The probability distribution associated with each of the plurality of reference points, representing the probability of the existence of the reference point at each location, is updated in such a way that the probability of the reference point existing at the location of the detected reference point is increased. The aforementioned features are the lane marking lines used to delineate driving lanes, and the plurality of reference points are points on the lane marking lines that are discrete at predetermined intervals, using predetermined locations as references. In the update, if the location of a reference point on one of the lane marking lines that divides a lane of interest into multiple lanes is detected from the surrounding data, and the location of a reference point on the lane marking line of the other lane is not detected, it is considered that the reference point on the lane marking line of the other lane was detected at a position away from the position of the lane marking line of the first lane marking line, which is a width associated with that lane. The probability distribution associated with the reference point on the other lane marking line is updated in such a way that the probability distribution associated with the reference point on the other lane marking line is a normal distribution with the average position of the reference point detected on the lane marking line of the first lane marking line away from the width, and the variance in a direction orthogonal to the lane marking line of the second lane marking line being at least half of the width. In the update, if it is determined that the position of the reference point detected from a predetermined number or more of the surrounding data representing the ground features around the vehicle has an intentional difference between the position of the reference point and the probability distribution associated with the reference point, the probability distribution associated with the reference point is updated.

4. A non-transitory computer-readable medium storing a computer program for map updating, the computer program for map updating causing a computer to perform: From the surrounding data of features representing the area around the vehicle, the positions of multiple reference points corresponding to features on the road the vehicle is traveling on are detected. The probability distribution associated with each of the plurality of reference points, representing the probability of the existence of the reference point at each location, is updated in such a way that the probability of the reference point existing at the location of the detected reference point is increased. The aforementioned features are the lane marking lines used to delineate driving lanes, and the plurality of reference points are points on the lane marking lines that are discrete at predetermined intervals, using predetermined locations as references. In the update, if the location of a reference point on one of the lane marking lines that divides a lane of interest into multiple lanes is detected from the surrounding data, and the location of a reference point on the lane marking line of the other lane is not detected, it is considered that the reference point on the lane marking line of the other lane was detected at a position away from the position of the lane marking line of the first lane marking line, which is a width associated with that lane. The probability distribution associated with the reference point on the other lane marking line is updated in such a way that the probability distribution associated with the reference point on the other lane marking line is a normal distribution with the average position of the reference point detected on the lane marking line of the first lane marking line away from the width, and the variance in a direction orthogonal to the lane marking line of the second lane marking line being at least half of the width. In the update, if it is determined that the position of the reference point detected from a predetermined number or more of the surrounding data representing the ground features around the vehicle has an intentional difference between the position of the reference point and the probability distribution associated with the reference point, the probability distribution associated with the reference point is updated.

Images