JP2021012086A - Lane estimation apparatus - Google Patents

Abstract

To improve the accuracy of estimating a lane where a vehicle is positioned.SOLUTION: A traffic event statisticizing-server 40 performs: estimating plural sets of probability value for each lane indicating the probability of a vehicle positioned in individual lanes by using each piece of plural categories of detection information that can be used for estimating a lane where the vehicle is positioned; determining, for each of plural categories of detection information, a weight factor for the probability value of each lane in accordance with a coefficient of confidence of the probability value for each lane estimated from the detection information; calculating, for each lane, a total probability value produced by integrating products of plural probability values whose categories of detection information, the information that corresponds to a single lane and has been used for estimation, are mutually different, and an each weight factor corresponding to the detection information used for estimating the respective probability value; and estimating a lane where the vehicle is positioned, on the basis of the total probability value calculated for each lane.SELECTED DRAWING: Figure 1

Description

The present invention relates to a lane estimation device.

In Patent Document 1, the position of the own vehicle is estimated based on the positioning signal from the GNSS (Global Navigation Satellite System) satellite, and the position of the own vehicle is estimated based on the captured state of the GNSS satellite and the information of the sensor mounted in the own device. A technique for outputting the certainty of the result is disclosed. Specifically, the captured state of the GNSS satellite is good, the amount of fluctuation of the white line position is smaller than the threshold value due to the white line recognition of the camera, and there is no abnormal change due to the stability judgment of the acceleration / angular velocity sensor. When the deviation amount is smaller than the threshold value by the consistency judgment of the redundant acceleration / angular velocity sensor, the vehicle position information is determined. In other cases, the vehicle position information is undefined.

JP-A-2017-2233483

However, although Patent Document 1 describes that the capture state of the GNSS satellite is combined with the information of a sensor such as a camera to determine the certainty of the own vehicle position estimation result, the positioning signal from the GNSS satellite is used by the camera or the like. It is not described that it is used for estimating the position of the own vehicle in combination with the information of the sensor of. For example, when the lane in which the vehicle is located is estimated by using the positioning information from the GNSS satellite alone, the correct answer rate decreases depending on the situation, and the accuracy of the positioning information value cannot be guaranteed.

The present invention has been made in consideration of the above facts, and an object of the present invention is to obtain a lane estimation device capable of improving the accuracy of estimating the lane in which a vehicle is located.

The lane estimation device according to the invention according to claim 1 uses a plurality of types of detection information that can be used for estimating the lane in which the vehicle is located, and the vehicle is located in each lane. The first estimation unit estimates from the detection information the first estimation unit that estimates a plurality of sets of probability values for each lane representing the probability and the weighting coefficient for the probability value for each lane estimated by the first estimation unit. A determination unit that determines each of the plurality of types of detection information according to the reliability of the probability value for each lane, and a plurality of probabilities that correspond to the same lane and have different types of detection information used for estimation. A calculation unit that calculates the total probability value obtained by multiplying the value by each of the weighting coefficients corresponding to the detection information used for estimating each probability value for each lane, and the calculation unit for each lane. Includes a second estimation unit that estimates the lane in which the vehicle is located based on the total probability value calculated in 1.

When the probability value for each lane is estimated from the detection information available for estimating the lane in which the vehicle is located, the reliability of the probability value for each lane depends on the environment of the vehicle. However, the environmental conditions of the vehicle that affect the reliability of the probability value for each lane (for example, the presence or absence of a tunnel, whether the weather is rainy, etc.) are different for each type of detection information. Therefore, for example, under certain environmental conditions, the reliability of the probability value for each lane estimated from the first detection information is low, while the reliability of the probability value for each lane estimated from the second detection information is low. May show a high value.

Based on this, in the invention according to claim 1, the probability value for each lane estimated from each of a plurality of types of detection information that can be used for estimating the lane in which the vehicle is located is based on the detection information. 1 The weighting coefficient is determined for each of the plurality of types of detection information according to the reliability of the probability value for each lane estimated by the estimation unit. Then, the total probability obtained by multiplying a plurality of probability values corresponding to the same lane and different types of detection information used for estimation by weight coefficients corresponding to the detection information used for estimating each probability value is integrated. The value is calculated for each lane, and the lane in which the vehicle is located is estimated based on the total probability value calculated for each lane. As a result, the weighting coefficient for each of the plurality of types of detection information is switched according to the reliability of the probability value for each lane estimated from the detection information, so that the accuracy of estimating the lane in which the vehicle is located is improved. Can be made to.

The present invention has the effect that the accuracy of estimating the lane in which the vehicle is located can be improved.

It is a schematic block diagram of the traffic information processing system which concerns on embodiment. It is a flowchart which shows an example of a lane estimation process. It is a block diagram which shows the outline of a lane estimation process. It is a diagram which shows the histogram of the positioning information acquired by a GNSS sensor. It is a diagram which shows an example of the estimation result of the speed distribution for each lane. It is a flowchart which shows an example of the weighting coefficient determination processing which concerns on 1st Embodiment. It is a flowchart which shows an example of the weighting coefficient determination process which concerns on 2nd Embodiment. It is a diagram which shows an example of the map which defines the relationship between the reliability αB and the weighting coefficient b. It is a diagram which shows an example of the map which defines the relationship between the reliability αA and the weighting coefficient a2. It is a diagram which shows an example of the map which defines the relationship between the reliability αC and the weighting coefficient c2.

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

[First Embodiment]
As shown in FIG. 1, the traffic information processing system 10 according to the present embodiment includes an in-vehicle system 12 mounted on each of a plurality of vehicles, a plurality of collection servers 38, a traffic event statistics server 40, and traffic information distribution. Includes server 60 and. The in-vehicle system 12, the collection server 38, the traffic event statistics server 40, and the traffic information distribution server 60 can communicate with each other via the network 72.

The in-vehicle system 12 receives arbitrary information from a GNSS sensor 14 that receives a positioning signal from a GNSS satellite and acquires GNSS positioning information, a camera 16 that photographs the surroundings of the vehicle, and a vehicle speed sensor 18 that detects the vehicle speed of the vehicle. A display unit 20 that can be displayed and an in-vehicle ECU 22 are included. The GNSS sensor 14, the camera 16, the vehicle speed sensor 18, the display unit 20, and the vehicle-mounted ECU 22 are connected to each other via the bus 24.

The in-vehicle ECU 22 is a CPU (Central Processing Unit) 26, a memory 28 such as a ROM (Read Only Memory) or a RAM (Random Access Memory), and a non-volatile storage unit such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive). 30 and communication I / F (Inter Face) 32 are included. The CPU 26, the memory 28, the storage unit 30, and the communication I / F 32 are connected to each other so as to be able to communicate with each other via the internal bus 34.

The vehicle-mounted ECU 22 of the vehicle-mounted system 12 has vehicle information such as GNSS positioning information acquired by the GNSS sensor 14, image information captured by the camera 16, vehicle speed information detected by the vehicle speed sensor 18, and vehicle throttle and brake operation information. Is sent regularly. Vehicle information such as GNSS positioning information transmitted from the in-vehicle system 12 is received by any of the plurality of collection servers 38, and then transferred from the collection server 38 to the traffic event statistics server 40.

The GNSS positioning information acquired by the GNSS sensor 14, the image information taken by the camera 16, and the vehicle speed information detected by the vehicle speed sensor 18 are examples of a plurality of types of detection information.

The traffic event statistics server 40 includes a CPU 42, a memory 44 such as a ROM or RAM, a non-volatile storage unit 46 such as an HDD or SSD, and a communication I / F 48. The CPU 42, the memory 44, the storage unit 46, and the communication I / F 48 are communicably connected to each other via the internal bus 50. The traffic event statistics program 52 and the map information 54 are stored in the storage unit 46.

In the traffic event statistic server 40, the traffic event statistic program 52 is read from the storage unit 46 and expanded in the memory 44, and the traffic event statistic program 52 expanded in the memory 44 is executed by the CPU 42. The traffic event statistics processing described below is performed.

That is, the traffic event statistics server 40 is based on vehicle information such as GNSS positioning information transferred from the in-vehicle system 12 via the collection server 38, and the individual vehicles are on the road on which the individual vehicles are traveling. The process of estimating the lane in which the vehicle is located (the lane estimation process described later) is performed for each of the individual vehicles from which the vehicle information is transmitted. In the present embodiment, the traffic event statistics server 40 functions as an example of the first estimation unit, the determination unit, the calculation unit, and the second estimation unit by performing the lane estimation processing described later, and lane estimation. Functions as an example of a device.

In addition, the traffic event statistics server 40 estimates congestion for each lane of the road based on the result of lane estimation processing, estimates the traffic volume for each lane of the road, and estimates the speed distribution for each lane of the road. Then, processing such as predicting the time required to drive on the road is performed. Then, the traffic event statistics server 40 performs a process of transmitting a processing result such as a traffic jam estimation for each lane of the road to the traffic information distribution server 60.

The traffic information distribution server 60 includes a CPU 62, a memory 64 such as ROM and RAM, a non-volatile storage unit 66 such as HDD and SSD, and a communication I / F 68. The CPU 62, the memory 64, the storage unit 66, and the communication I / F 68 are connected to each other so as to be able to communicate with each other via the internal bus 70.

The traffic information distribution server 60 detects a traffic flow (traffic jam) on the road, detects a sign of traffic jam on the road, and is on the road based on a processing result such as a traffic jam estimation for each lane of the road received from the traffic event statistics server 40. Each of the above-mentioned information is generated as traffic information by detecting the position of the beginning and the end of the traffic jam, detecting or predicting the required time for each section of the road, and the like. Then, the traffic information distribution server 60 performs a process of distributing the generated traffic information to each vehicle. The traffic information delivered to each vehicle is displayed on the display unit 20 of the in-vehicle system 12.

Next, as the operation of the first embodiment, the lane estimation process executed by the traffic event statistics server 40 will be described with reference to FIG.

In step 100 of the lane estimation process, the traffic event statistics server 40 is included in the latest vehicle information transferred from the in-vehicle system 12 of a certain vehicle (hereinafter, tentatively referred to as "vehicle X") via the collection server 38. The GNSS positioning information acquired by the GNSS sensor 14 is read from. Note that FIG. 4 shows an example of variation in the position of the vehicle represented by the GNSS positioning information along the vehicle width direction. The numerical value on the horizontal axis in FIG. 4 represents the deviation from the actual position of the vehicle, and “A” in FIG. 4 represents the width of one lane.

In step 102, the traffic event statistic server 40 collates the GNSS positioning information read in step 100 with the map information 54, and is based on the GNSS positioning information by the first lane estimation logic that estimates the lane with the highest driving frequency. Estimated probability Ai for each lane (probability that vehicle X is located in each lane of the road on which vehicle X is traveling, i is a variable representing each lane) is calculated (see also FIG. 3). By this process, assuming that the total number of lanes on the road on which the vehicle X is traveling is i0, sets A1 to Ai0 of estimated probabilities of lanes 1 to i0 can be obtained.

Further, the traffic event statistics server 40 sets the reliability αA for lane estimation based on the GNSS positioning information, for example, based on the strength of the positioning signal received from the GNSS satellite by the GNSS sensor 14, as the signal strength increases. Calculate so that the value of reliability αA is high. The value of the reliability αA may be changed according to the number of GNSS satellites captured by the GNSS sensor 14.

In step 104, the traffic event statistics server 40 reads out the image information taken by the vehicle camera 16 from the latest vehicle information transferred from the vehicle-mounted system 12 of the vehicle X via the collection server 38.

In step 106, the traffic event statistics server 40 detects the lane boundary line from the image information read in step 104. Then, the traffic event statistics server 40 uses the second lane estimation logic that estimates the lane using the result of detecting the lane boundary line, and the estimation probability Bi (vehicle X is traveling) for each lane by the lane boundary line detection. Calculate the probability that vehicle X is located in each lane of the road (see also FIG. 3). By this process, assuming that the total number of lanes on the road on which the vehicle X is traveling is i0, sets B1 to Bi0 of estimated probabilities of lanes 1 to lane i0 can be obtained.

Further, the traffic event statistics server 40 has a high reliability αB for lane estimation by lane boundary detection, for example, based on an evaluation value for evaluating the certainty of the lane boundary detected from image information. It is calculated so that the value of the reliability αB increases as the value increases.

In step 108, the traffic event statistics server 40 reads out the vehicle speed information detected by the vehicle speed sensor 18 from the latest vehicle information transferred from the vehicle-mounted system 12 of the vehicle X via the collection server 38.

In step 110, the traffic event statistics server 40 collates the vehicle speed information read in step 108 with the speed distribution for each lane (an example is shown in FIG. 5) estimated in advance for the road on which the vehicle X is traveling. By the third lane estimation logic that selects the lane with the speed distribution closest to the vehicle speed of the vehicle X, the estimated probability Ci for each lane based on the speed distribution (the vehicle X is located in each lane of the road on which the vehicle X is traveling). Calculate the probability of doing so (see also FIG. 3). As for the speed distribution for each lane shown in FIG. 5, the traffic volume for each individual lane on the road Y can be obtained by performing the lane estimation process (FIG. 2) for each vehicle traveling on the road Y. Therefore, it can be obtained from the traffic volume for each lane and the vehicle speed of each vehicle. By this process, assuming that the total number of lanes on the road on which the vehicle X is traveling is i0, sets C1 to Ci0 of estimated probabilities of lanes 1 to i0 can be obtained.

Further, the traffic event statistics server 40 sets the reliability αC for lane estimation based on the speed distribution, for example, as the vehicle speed represented by the vehicle speed information deviates toward the low speed side (as the vehicle speed approaches the traffic jam). Calculate so that the value is low. The above-mentioned steps 102, 106, and 110 are examples of processing by the first estimation unit.

In step 112, the traffic event statistics server 40 performs the weighting coefficient determination process. Hereinafter, this weighting coefficient determination process will be described with reference to FIG. The weighting coefficient determination process is an example of the process by the determination unit.

In step 130, the traffic event statistics server 40 determines whether or not the reliability αB for lane estimation by lane boundary detection is equal to or greater than the threshold value. If the determination in step 130 is affirmed, the process proceeds to step 132. In step 132, the traffic event statistics server 40 sets the weighting coefficient for the estimated probability for each lane to (a, b, c) = (0,1,0). The weighting coefficient a is a weighting coefficient for the estimated probability Ai for each lane based on GNSS positioning information, the weighting coefficient b is a weighting coefficient for the estimated probability Bi for each lane by detecting the lane boundary line, and the weighting coefficient c is the speed distribution. It is a weighting coefficient with respect to the estimated probability Ci for each lane based on.

For example, when the weather around the vehicle X is fine, the evaluation value for evaluating the certainty of the lane boundary line detected from the image information is high, so that the lane estimation by the lane boundary line detection can be performed. The reliability αB exceeds the threshold value. In step 130, in such a case, the weight coefficient a is set to 0, the weight coefficient b is set to 1, and the weight coefficient c is set to 0. As a result, the total probability value Di, which will be described later, matches the estimated probability Bi for each lane by lane boundary detection in which the corresponding reliability αB is equal to or higher than the threshold value. When the process of step 132 is performed, the weighting coefficient setting process is terminated, and the process proceeds to step 114 of the lane estimation process.

On the other hand, for example, when the weather around the vehicle X is rainy, the evaluation value for evaluating the certainty of the lane boundary line detected from the image information is low, so that the lane estimation by the lane boundary line detection can be performed. The reliability αB drops below the threshold. In such a case, the determination in step 130 is denied and the process proceeds to step 134. In step 134, the traffic event statistics server 40 determines whether or not the reliability αA for lane estimation based on the GNSS positioning information is equal to or greater than the threshold value.

For example, when there is no obstacle such as a tunnel that shields the signal from the GNSS satellite around the vehicle X, the reliability αA for lane estimation based on the GNSS positioning information becomes equal to or higher than the threshold value. In such a case, the determination in step 134 is affirmed and the process proceeds to step 136.

In step 136, the traffic event statistics server 40 determines whether or not the reliability αc for lane estimation based on the speed distribution is equal to or greater than the threshold value. For example, when the vehicle X is traveling at a vehicle speed corresponding to a traffic jam or a vehicle speed close to a traffic jam, the reliability αC for lane estimation based on the speed distribution becomes less than the threshold value. In such a case, the determination in step 136 is denied and the process proceeds to step 138.

In step 138, the traffic event statistics server 40 sets the weighting coefficient for the estimated probability for each lane to the weighting coefficient (a, b, c) = (1,0,0). By setting the weighting coefficient a to 1, the weighting coefficient b to 0, and the weighting coefficient c to 0 in this way, the GNSS positioning information in which the total probability value Di described later has the corresponding reliability αA equal to or higher than the threshold value. It will match the estimated probability Ai for each lane based on. When the process of step 138 is performed, the weighting coefficient setting process is terminated, and the process proceeds to step 114 of the lane estimation process.

Further, for example, when the vehicle X is traveling at a vehicle speed corresponding to a free flow, the reliability αC for lane estimation based on the speed distribution becomes equal to or higher than the threshold value. In this case, the determination in step 136 is affirmed and the process proceeds to step 140. In step 140, the traffic event statistics server 40 sets the weighting coefficient for the estimated probability for each lane to the weighting coefficient (a, b, c) = (0.5,0,0.5).

By setting the weighting coefficient a to 0.5, the weighting coefficient b to 0, and the weighting coefficient c to 0.5 in this way, the total probability value Di described later has a corresponding reliability αA equal to or higher than the threshold value. It is matched with the average value of the estimated probability Ai for each lane based on the GNSS positioning information and the estimated probability Ci for each lane based on the speed distribution in which the corresponding reliability αC is equal to or higher than the threshold value. When the process of step 138 is performed, the weighting coefficient setting process is terminated, and the process proceeds to step 114 of the lane estimation process.

Further, for example, when there is an obstacle such as a tunnel that shields the signal from the GNSS satellite around the vehicle X, the reliability αA for lane estimation based on the GNSS positioning information becomes less than the threshold value. In such a case, the determination in step 134 is denied and the process proceeds to step 142. In step 142, the traffic event statistics server 40 determines whether or not the reliability αc for lane estimation based on the speed distribution is equal to or greater than the threshold value, as in step 136 described above. If the determination in step 142 is affirmed, the process proceeds to step 144.

In step 144, the traffic event statistics server 40 sets the weighting coefficient for the estimated probability for each lane to the weighting coefficient (a, b, c) = (0,0,1). By setting the weighting coefficient a to 0, the weighting coefficient b to 0, and the weighting coefficient c to 1 in this way, the total probability value Di described later becomes a velocity distribution in which the corresponding reliability αC is equal to or higher than the threshold value. It matches the estimated probability Ci for each lane based on. When the process of step 144 is performed, the weighting coefficient setting process is terminated, and the process proceeds to step 114 of the lane estimation process.

If the determination in step 142 is denied, the process proceeds to step 146. If the determination in step 142 is denied, the estimated probability Ai for each lane based on the GNSS positioning information, the estimated probability Bi for each lane based on the lane boundary detection, and the estimated probability Ci for each lane based on the speed distribution are all reliable. If you can't. Therefore, in step 146, the traffic event statistics server 40 outputs "unestimable" which means that it is difficult to accurately estimate the lane in which the vehicle X is located. When the process of step 146 is performed, the weighting coefficient setting process is terminated, and the process proceeds to step 114 of the lane estimation process.

Returning to FIG. 2 and continuing the description, in step 114, the traffic event statistics server 40 determines whether or not the result of the weighting coefficient determination process is “unpresumable”. If the determination in step 114 is affirmed, the lane estimation process ends without performing lane estimation. On the other hand, if the determination in step 114 is denied, the process proceeds to step 116. In step 116, the traffic event statistics server 40 sets the variable i to 1.

In step 118, the traffic event statistics server 40 calculates the total probability value Di of the lane i by the following equation (1).
Di ← Ai × a ＋ Bi × b ＋ Ci × c… (1)
According to the above equation (1), the detection information used for estimating each probability value corresponds to a plurality of probability values Ai, Bi, Ci corresponding to the same lane and different types of detection information used for estimation. The total probability value Di is calculated by multiplying each of the weighting coefficients a, b, and c.

In step 120, the traffic event statistics server 40 determines whether or not the variable i has reached the total number of lanes i0 of the road on which the vehicle X is traveling. If the determination in step 120 is denied, the process proceeds to step 122, and in step 122, the traffic event statistics server 40 increments the variable i by 1. When the process of step 122 is performed, the process returns to step 118.

As a result, steps 118 to 122 are repeated until the determination in step 120 is affirmed, and the total probability value Di is calculated for each lane of the road on which the vehicle X is traveling. Note that steps 116 to 112 are examples of processing by the calculation unit.

If the determination in step 120 is affirmed, the process proceeds to step 124. In step 124, the traffic event statistics server 40 compares the total probability value Di for each lane, and estimates the lane in which the total probability value Di indicates the maximum value as the lane in which the vehicle X is located (FIG. 3). See also). Note that step 124 is an example of processing by the second estimation unit.

Further, in step 126, the traffic event statistics server 40 feeds back the lane estimation result obtained in step 124 to the first lane estimation logic to the third lane estimation logic (see also FIG. 3). Specifically, the lane estimation result fed back to each lane estimation logic is used for adjusting processing parameters in each lane estimation logic as needed.

As described above, in the first embodiment, the traffic event statistics server 40 uses a plurality of types of detection information that can be used for estimating the lane in which the vehicle is located, and the vehicles are individually used. A plurality of sets of probability values (Ai, Bi, Ci) for each lane representing the probability of being located in a lane are estimated. Further, the weighting coefficient (a, b, c) for the probability value for each lane is determined for each of the plurality of types of detection information according to the reliability of the probability value for each lane estimated from the detection information. In addition, the total probability obtained by multiplying a plurality of probability values corresponding to the same lane and different types of detection information used for estimation by weighting coefficients corresponding to the detection information used for estimating each probability value. The value (Di) is calculated for each lane. Then, the lane in which the vehicle is located is estimated based on the total probability value calculated for each lane. This makes it possible to improve the accuracy of estimating the lane in which the vehicle is located.

Further, in the first embodiment, the GNSS positioning information acquired by the GNSS sensor 14, the image information taken by the camera 16, and the vehicle speed information detected by the vehicle speed sensor 18 are used as the plurality of types of detection information. The probability values for each lane estimated from these detection information differ from each other in the environmental conditions of the vehicle (for example, the presence or absence of a tunnel, whether the weather is rainy, etc.) that affect the reliability. Therefore, by combining these detection information, the lane is estimated by increasing the possibility that the probability value for each lane estimated from at least one detection information shows high reliability in various environments of the vehicle. The accuracy of the operation can be improved.

Further, in the first embodiment, it is determined whether or not the reliability αB, αA, αC of the lane estimation is equal to or higher than the threshold value, and the weighting coefficients a, b, and c are determined by the conditional branching based on the determination. The coefficients a, b, and c can be determined by a simple process.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. Since the second embodiment has the same configuration as the first embodiment, the same reference numerals are given to the respective parts to omit the description of the configuration. Hereinafter, the operation of the second embodiment will be described in the first embodiment. Only the parts that differ from are explained.

In the second embodiment, the weighting coefficient determining process shown in FIG. 7 is performed instead of the weighting coefficient determining process (FIG. 6) described in the first embodiment. In the weighting coefficient determination process according to the second embodiment, in step 150, the traffic event statistics server 40 is based on a map that predetermines the relationship between the reliability αB of lane estimation by lane boundary detection and the weighting coefficient b. The weighting coefficient b for the estimated probability Bi for each lane by lane boundary detection is calculated from the reliability αB. FIG. 8 shows an example of a map that defines the relationship between the reliability αB and the weighting coefficient b. The map shown in FIG. 8 defines the relationship between the reliability αB and the weighting coefficient b so that the weighting coefficient b increases in a quadratic curve as the reliability αB approaches 100%.

Further, in step 152, the traffic event statistics server 40 calculates the weighting coefficient a2 from the reliability αA based on a map that prescribes the relationship between the reliability αA of the lane estimation based on the GNSS positioning information and the weighting coefficient a2. .. FIG. 9 shows an example of a map that defines the relationship between the reliability αA and the weighting coefficient a2. The map shown in FIG. 9 defines the relationship between the reliability αB and the weighting coefficient b so that the weighting coefficient a2 increases in a quadratic curve as the reliability αA approaches 100%.

In step 154, the traffic event statistics server 40 estimates the probability for each lane based on the GNSS positioning information from the weighting coefficient b calculated in step 150 and the weighting coefficient a2 calculated in step 152 based on the following equation (2). The weighting coefficient a for Ai is calculated.
a = (1-b) × a2 ... (2)
By calculating the weighting coefficient a based on the above equation (2), when the value of the weighting coefficient b is high, the value of the weighting coefficient a is suppressed to be low, and when the value of the weighting coefficient b is low, the value of the weighting coefficient a2 is set. However, the value of the weighting coefficient a is determined so that the weighting coefficient a becomes a relatively high value and the sum (a + b) of the weighting coefficients a and b is 1 or less.

In step 156, the traffic event statistics server 40 calculates the weighting coefficient c2 from the reliability αC based on a map that prescribes the relationship between the reliability αC of the lane estimation based on the speed distribution and the weighting coefficient c2. FIG. 10 shows an example of a map that defines the relationship between the reliability αC and the weighting coefficient c2. The map shown in FIG. 10 defines the relationship between the reliability αB and the weighting coefficient b so that the weighting coefficient c2 increases in a quadratic curve as the reliability αC approaches 100%.

In step 158, the traffic event statistics server 40 speeds from the weighting coefficient b calculated in step 150, the weighting coefficient a calculated in step 154, and the weighting coefficient c2 calculated in step 156 based on the following equation (3). The weighting coefficient c for the estimated probability Ci for each lane based on the distribution is calculated.
c = (1-ab) × c2 ... (3)
By calculating the weighting coefficient c based on the above equation (3), when at least one of the weighting coefficients a and b is high, the weighting coefficient c is suppressed to be low and the weighting coefficients a and b are low. The value of the weighting coefficient c is determined so that the weighting coefficient c becomes a relatively high value and the total sum (a + b + c) of the weighting coefficients a, b, and c is 1 or less, although it sometimes depends on the value of the weighting coefficient c2. Will be done.

In step 160, the traffic event statistics server 40 determines whether or not the sum (a + b + c) of the weighting coefficients a, b, and c obtained by the above-described processing is equal to or greater than a predetermined value K. If the determination in step 160 is affirmed, the process proceeds to step 162, and in step 162, the traffic event statistics server 40 outputs the weighting coefficients a, b, and c. In this case, using the output weighting coefficients a, b, and c, the lane in which the vehicle X is located is estimated as described in the first embodiment.

If the determination in step 160 is denied, the process proceeds to step 164, and in step 164, the traffic event statistics server 40 outputs "unestimable". In this case, the lane in which the vehicle X is located is not estimated.

As described above, in the second embodiment, the weighting coefficient is based on a map that prescribes the relationship between the reliability αB, αA, αC and the weighting coefficient b, a2, c2 of the probability value for each lane estimated from the detection information. b, a2, and c2 are determined, and the weighting coefficient a is determined from the weighting coefficient a2 based on the equation (2), and the weighting coefficient c is determined from the weighting coefficient c2 based on the equation (3). As a result, the weighting factors b, a2, and c2 change according to the reliability αB, αA, and αC, so that the weighting factors b, a, and c can be determined so that the lane can be estimated more accurately.

In the above description, the mode in which the traffic event statistics server 40 performs the lane estimation based on the lane boundary line (step 106) has been described, but the present invention is not limited to this, and the lane estimation based on the lane boundary line is performed by the in-vehicle system 12. You may do so.

Further, in the above, as an example of the detection information that can be used for lane estimation, the GNSS positioning information acquired by the GNSS sensor 14, the image information taken by the camera 16, and the vehicle speed information detected by the vehicle speed sensor 18 have been described. , Not limited to this. For example, if the position of a point on the road where lanes branch or merge is known, it is possible to estimate the lane based on the frequency with which the throttle or brake of the vehicle is operated at that point. It is also possible to use the throttle and brake operation information of the vehicle as detection information for lane estimation.

10 Traffic information processing system 12 In-vehicle system 14 GNSS sensor 16 Camera 18 Vehicle speed sensor 38 Collection server 40 Traffic event statistics server 60 Traffic information distribution server

Claims (1)

Using each of the multiple types of detection information that can be used to estimate the lane in which the vehicle is located, a plurality of sets of probability values for each lane representing the probability that the vehicle is located in each lane are estimated. 1st estimation part and
The weighting coefficient for the probability value for each lane estimated by the first estimation unit is the detection information of the plurality of types according to the reliability of the probability value for each lane estimated by the first estimation unit from the detection information. The decision unit that decides for each of
A total probability value obtained by multiplying a plurality of probability values corresponding to the same lane and different types of detection information used for estimation by the weighting factors corresponding to the detection information used for estimating each probability value. With a calculation unit that calculates for each lane,
A second estimation unit that estimates the lane in which the vehicle is located based on the total probability value calculated by the calculation unit for each lane.
Lane estimation device including.