JP7277412B2 - Target tracker - Google Patents

Description

The present disclosure relates to technology for tracking a target using target position information obtained from a plurality of sensors.

By intelligently combining detection results obtained by a plurality of types of sensors that detect targets, performance improvement and functional realization of systems that use information on detected targets have been attempted. In such a system, when tracking a target, time-series observation information of instantaneous values of the target position detected by the sensor is used to improve the accuracy of target position estimation. However, it is known that near the limit of the detection area of the sensor, the tracking accuracy of the target decreases due to an increase in observation errors and false detections.

In Patent Document 1, in order to suppress such a decrease in tracking accuracy, a plurality of sensors are arranged so that part of their detection areas overlap each other, so that observation is performed near the boundary of the detection areas of the sensors. Techniques for providing redundancy have been proposed.

Japanese Patent Application Publication No. 2019-519051

However, as a result of detailed studies by the inventors, it was found that the above-described prior art has the following problems. In other words, when the multiple sensors that make up the system include sensors with moving detection areas, such as in-vehicle sensors, it is not always possible to provide redundancy near the boundaries of the detection areas, resulting in observation errors. There is a problem that a sufficient suppressing effect cannot be obtained.

One aspect of the present disclosure provides a technique for improving target tracking accuracy near the limit of the detection area of a sensor.

One aspect of the present disclosure is a target tracking device, which includes an information acquisition unit (5: S110), a reliability setting unit (5: S140), an estimated value update unit (5: S190), and an area setting unit (5: S130) and a reliability adjustment unit (5: S180).

The information acquisition unit acquires sensor information including position information of the sensors and detection areas of the sensors, and target information including the positions of the detected targets from the plurality of sensors. The reliability setting unit uses the position of the target that is estimated in each processing cycle as an estimated value, the position of the target that is predicted from the estimated value in the previous processing cycle as a predicted value, and uses Using the position of the target indicated in the target information as the observed value, the reliability of the predicted value and the observed value is set. The estimated value updating unit uses the corresponding value pair, which is the predicted value and the observed value representing the same target, and the reliability set by the reliability setting unit for the corresponding value pair to update the object associated with the corresponding value pair. Update target estimates. The area setting unit sets a connecting area that includes at least a part of the boundary of the detection area indicated by the sensor information and has lower detection accuracy than other areas in the detection area. The reliability adjustment unit adjusts the reliability of the predicted value forming the pair of the observed value and the corresponding value so as to increase the reliability when the observed value is in the connecting area.

According to such a configuration, it is possible to suppress a decrease in target position estimation accuracy and target tracking accuracy due to an increase in sensor detection error, that is, observation value error.

1 is a block diagram showing the configuration of a traffic control system; FIG. 2 is a block diagram showing the configuration of an in-vehicle terminal; FIG. It is a block diagram which shows the structure of an infrastructure sensor. It is a block diagram which shows the structure of a server. 4 is a flowchart of tracking processing executed by a server; FIG. 10 is an explanatory diagram regarding setting of a connecting area; It is explanatory drawing which shows the relationship of the observed value in a connection area, a predicted value, and an estimated value. 7 is a graph showing results of calculation of estimated values by simulation with and without a connecting area. FIG. 11 is an explanatory diagram showing another method of setting a connecting area; FIG. 10 is an explanatory diagram showing a simple setting method for connecting areas;

Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.
[1. overall structure]
A traffic control system 1 shown in FIG. 1 integrates observation information acquired from multiple sensors existing in a traffic environment to estimate target positions, movements, etc., and notifies each terminal in the traffic environment of the estimation results. This will improve the accuracy of automated driving and driving assistance.

The traffic control system 1 includes an in-vehicle terminal 2 , an infrastructure sensor 3 and a server 5 . The server 5 corresponds to a target object tracking device.
[1-1. In-vehicle terminal]
The in-vehicle terminal 2 is mounted in a vehicle. As shown in FIG. 2 , the vehicle-mounted terminal 2 includes an in-vehicle communication device 21 , an external communication device 22 , and a processing section 23 . The in-vehicle communication device 21 acquires various information via an in-vehicle network of a vehicle (hereinafter referred to as own vehicle) in which the in-vehicle terminal 2 is mounted. The external communication device 22 communicates with the server 5 by wireless communication. The processing unit 23 includes a microcomputer having a CPU and memories such as ROM and RAM, and executes processing for realizing various functions.

The in-vehicle terminal 2 includes an information collecting unit 231 and an integrated information providing unit 232 as functional blocks representing functions realized by processing executed by the processing unit 23 .
The information collecting unit 231 repeatedly collects target information, sensor information, and vehicle state information via the in-vehicle communication device 21 and transmits the information to the server 5 via the external communication device 22 .

The target object information is observation information about the target object detected by the perimeter monitoring sensor group 100 mounted on the vehicle. The peripheral monitoring sensor group 100 may include a millimeter wave radar 101, a lidar 102, an ultrasonic sensor 103, a camera 104, and the like. Hereinafter, these components belonging to the surroundings monitoring sensor group 100 are also simply referred to as sensors 101-104. The information included in the target information differs depending on the sensors 101 to 104 that are information generation sources, but in any case, at least the position of the detected target is included. The target information may include, in addition to the position of the target, the distance to the target, the moving speed and moving direction of the target, the size of the target, the type of the target, and the like. Note that the position of the target, the distance to the target, and the moving speed and moving direction of the target may be represented by relative values with the in-vehicle terminal 2 as a reference.

The sensor information includes performance information regarding the performance of the individual sensors 101 to 104 and mounting information representing the mounting state with respect to the own vehicle. The mounting information includes the height from the road surface, the straight direction of the vehicle, and the mounting angle with respect to the horizontal direction. The performance information includes sensor type, detection distance range, and detection angle range. The detection distance range is the range of distances in which the target can be detected with acceptable accuracy under certain conditions. The detection angle range is the range of angles in which the target can be detected with acceptable accuracy under certain conditions. For example, an illuminance value, a reflectance, or the like may be used as the constant condition. Also, as the allowable accuracy, a detection probability of a certain value (for example, 95%) or more may be used. Here, the sensor front direction is the center direction of the detection area A determined by the detection distance range and the detection angle range, and the attachment angle represents the orientation of the sensor front direction. Also, the detection angle range is expressed with the front direction of the sensor as a reference (that is, 0°).

The vehicle state information includes at least the position and traveling direction of the vehicle detected by the position sensor 110 mounted on the vehicle. Position sensor 110 may be a device that obtains information using GNSS (ie, Global Navigation Satellite System). That is, by combining the vehicle state information and the mounting information included in the sensor information, it is possible to specify the sensor position and the sensor front direction in the global coordinate system.

The integrated information providing unit 232 receives integrated information about targets and the like existing around the own vehicle from the server 5 via the external communication device 22, and automatically transmits the received integrated information via the internal communication device 21. Provided to each ECU that executes vehicle control related to driving and driving support.

[1-2. Infrastructure sensor]
The infrastructure sensor 3 is installed on roadside buildings and the like in addition to traffic infrastructure such as road signs and traffic lights. As shown in FIG. 3 , the infrasensor 3 includes a sensor body 31 , a communication device 32 and a processing section 33 .

The sensor main body 31 includes, for example, a camera 311, a radar 312, and the like. The sensor main body 31 is configured to be able to change the orientation of the camera 311 and the radar 312 in the front direction of the sensor (that is, the orientation of the sensor main body 31 ) according to an instruction from the processing unit 33 . Below, the camera 311 and the radar 312 are also simply referred to as sensors 311 and 312 .

The communication device 32 performs communication with the server 5 via a wired communication network or a wireless communication network.
The processing unit 33 includes a microcomputer having a CPU and memories such as ROM and RAM, and executes processing for realizing various functions. The infra-sensor 3 includes an information collection unit 331 and an attitude control unit 332 as functional blocks representing functions realized by processing executed by the processing unit 33 .

The information collecting unit 331 repeatedly collects target information from the sensor main body 31 and transmits the collected target information to the server 5 via the communication device 32 together with the sensor information and the attitude information.
The target information and sensor information are basically the same as the target information and sensor information collected by the information collection unit 231 of the vehicle-mounted terminal 2 . However, the mounting information included in the sensor information includes the direction of the front direction of the sensor when the sensor is in a preset reference posture (hereinafter referred to as the reference front direction). The attitude information includes the attitude of the sensor main body 31 determined by the attitude control section 332, that is, the orientation of the sensor front direction with respect to the reference front direction. That is, by combining the mounting information included in the sensor information and the orientation information, it is possible to identify the sensor position and the sensor front direction in the global coordinate system.

If the infrasensor 3 is configured so that the orientation of the sensor main body 31 cannot be changed, the orientation control section 332 and the orientation information may be omitted.
Here, the case where target object information is generated in the infrastructure sensor 3 and provided to the server 5 has been described, but instead of the target object information, raw data obtained from the sensor main body 31 is provided to the server 5 . may In this case, the server 5 requires a function of generating target information from raw data acquired from the infrastructure sensor 3 .

Hereinafter, the sensors 101 to 104 belonging to the surrounding monitoring sensor group 100 and the sensors 311 and 312 belonging to the sensor main body 31 are collectively referred to as field sensors S, and when it is necessary to identify individual field sensors S, Si is used. show. i is a positive integer. In FIG. 1, any one of the sensors 101 to 104 belonging to the surrounding monitoring sensor group 100 mounted on the vehicle is designated as S1, S2, and one of the sensors 311, 312 belonging to the sensor body 31 of the infrastructure sensor 3 is designated as S3. . Also, detection areas of the field sensors S1 to S3 are represented by A1 to A3.

[1-3. server]
The server 5 acquires information from a plurality of field sensors Si present in the traffic environment and integrates the acquired information, thereby realizing highly accurate and robust detection compared to detection by a single sensor. Integrated information, which is the detection result, is provided to the vehicle-mounted terminal 2 .

As shown in FIG. 4 , the server 5 includes a communication device 51 , a processing section 52 and a storage section 53 .
The communication device 51 performs communication with the in-vehicle terminal 2 and the infrastructure sensor 3 .

The processing unit 52 includes a microcomputer having a CPU and memories such as ROM and RAM, and executes processing for realizing various functions.
The server 5 includes an information generating unit 521 and an information distributing unit 522 as functional blocks representing functions realized by processing executed by the processing unit 52 .

The information generation unit 521 uses sensor information and target information acquired from a plurality of field sensors S via the communication device 51 to generate integrated information that integrates information on targets existing in the target area. The target area is an area to be processed by the server 5 and is assigned to the server 5 in advance. Integrated information includes at least target position information. The details of the tracking process for tracking the target and sequentially updating the position information will be described later.

The information distribution unit 522 distributes the integrated information generated by the information generation unit 521 to the vehicle-mounted terminals 2 existing in the target area via the communication device 51 .
The storage unit 53 stores at least a map of the target area and time-series information on the positions of targets detected within the target area within a certain period of time in the past.

[2. tracking process]
Next, the tracking process executed by the server 5 to implement part of the functions of the information generation unit 521 will be described with reference to the flowchart of FIG.

This process is repeatedly executed at a constant cycle while the server 5 is running.
In S110, the server 5 acquires target information and sensor information received via the communication device 51 from the previous processing cycle to the current processing cycle.

In subsequent S120, the server 5 generates a sensor map showing the position and detection area of the field sensor S on the map showing the target area based on the sensor information acquired in S110.

In subsequent S130, the server 5 sets a connecting area CX where the detection performance of the field sensor S is degraded based on the detection area A of the field sensor S shown in the sensor map. For example, as shown in FIG. 6, the connecting area CX may be the range from the inside X[m] to the outside Y[m] with reference to the boundary of the detection area. Further, if the connecting area CX overlaps a non-connecting area of the detection area of another field sensor S, the overlapping portion is defined as a non-connecting area. The non-connecting area here refers to a range obtained by removing the connecting area CX from the detection area A. FIG.

In subsequent S140, the server 5 obtains predicted values for targets detected up to the previous processing cycle (hereinafter referred to as tracked targets) and targets indicated in the target information acquired in S110 (hereinafter referred to as observed targets). Set the confidence level for each observation that is the position of the target. The reliability is set based on at least one of the types of sensors used for position detection, specs, external environment (that is, the environment of the target area), types and speeds of tracking targets, and the like.

For example, if the field sensor S that generated the observed value is the lidar 102, which is easily affected by the weather, the reliability of the observed value is set high for fine weather and low for rain or snow. good too. Further, the higher the degree of matching with the predicted value, the higher the reliability of the observed value may be set. Further, in the case of the predicted value, for example, the longer the elapsed time since the tracked target associated with the predicted value was first recognized as the target, the higher the reliability may be set. Further, when the type of tracked target associated with the predicted value is a target whose position is easy to predict, such as a running vehicle, the reliability of the predicted value may be set high.

In subsequent S150, the server 5 selects any one of the tracking targets for which the processing of S160 to S190, which will be described later, has not yet been performed, as the target target.
In subsequent S160, the server 5 determines whether or not there is an observation target having a history connection with the target target. The presence or absence of history connection can be determined, for example, by the distance between the target target and the observation target, but the present invention is not limited to this, and a known method for determining the presence or absence of history connection is used. be able to. A combination of a predicted value and an observed value determined to have a history connection is hereinafter referred to as a corresponding value pair.

In S170, the server 5 determines whether or not the observed value is within the connecting area CX based on the predicted value of the target object and the observed value forming the corresponding value pair. If it is not within the connection area CX, the process proceeds to S190.

In S180, the server 5 adjusts the reliability of the predicted value of the target target so that the reliability is increased, and advances the process to S190.
In S190, the server 5 calculates the position (that is, the estimated value) of the target object according to the corresponding value pair and the confidence level set in S140 or the confidence level adjusted in S180 for the corresponding value pair. Update. Specifically, for example, the estimated value may be calculated by weighted addition of corresponding value pairs weighted by reliability. Alternatively, one of the corresponding value pairs with higher reliability may be used as the estimated value as it is. If the predicted value of the target object and the observed value forming the corresponding value pair are not extracted, the predicted value may be used as the estimated value as it is. A target target for which an observation target having a history connection has not been extracted for a predetermined maximum number of consecutive times may be regarded as lost and deleted from the tracked target.

Subsequently, in S200, the server 5 determines whether or not all tracked targets have been selected as target targets in the previous S150. If the server 5 determines that there are unselected tracking targets, the process returns to S150, and if it determines that all the tracking targets have been selected, the process proceeds to S210.

In S210, the server 5 updates the tracking target and terminates the process. Specifically, for already registered tracking targets, the estimated values updated in S190 are used to calculate predicted values to be used in the next processing cycle. An observation target for which history connection has not been confirmed with any tracking target is registered as a new tracking target, and a predicted value to be used in the next processing cycle is calculated based on the observed value.

In the tracking process, S110 corresponds to the information acquisition unit, S130 corresponds to the area setting unit, S140 corresponds to the reliability setting unit, S180 corresponds to the reliability adjustment unit, and S190 corresponds to the estimated value updating unit. corresponds to

[3. Operation example]
FIG. 7 shows a case where the detection areas A1 and A2 of the field sensors S1 and S2 do not overlap. When the target moves from the detection area A1 to the detection area A2, the accuracy of the observed value decreases near the boundary between the detection areas A1 and A2 and in the range that does not belong to any of the detection areas A1 and A2. Such an area becomes the connection area CX. In FIG. 7, the connecting area CX is indicated by a rectangle for convenience.

In the connecting area CX, the predicted value of the tracking target indicated by the white triangle in FIG. 7 is higher than the observed value of the target target indicated by the black square in FIG. The reliability is set higher than the observed value of the target indicated by the black square mark. Therefore, in the connecting area CX, the estimated value is calculated with more emphasis on the predicted value than on the observed value with reduced positional accuracy.

As shown in FIG. 8, when the connecting area CX is not set, when the target moves across the detection areas A1 and A2, the detection error of the observed value increases near the boundary of the detection areas. Estimates calculated using small observations will deviate significantly from the actual position. As a result, when the target straddles the detection areas A1 and A2, the last observed value detected by the field sensor S1 and the first observed value detected by the field sensor S2 are significantly different. may become. As a result, although the targets detected in both detection areas A1 and A2 are the same, they are recognized as different targets. On the other hand, when the connecting area CX is set, the reliability is adjusted in the connecting area CX so that the predicted value is emphasized. Therefore, unless the movement of the target changes significantly, the estimated value does not greatly deviate from the true value, and the tracking accuracy of the target is maintained.

[4. effect]
According to the embodiment detailed above, the following effects are obtained.
(4a) In the present embodiment, a connecting area CX, which is an area where the accuracy of the observed value is lowered, is set, and in the connecting area CX, adjustment is made so that the reliability of the predicted value increases. Therefore, in the connecting area CX, it is possible to suppress a decrease in target position estimation accuracy and target tracking accuracy due to an increase in sensor detection error. Furthermore, it is possible to suppress erroneous detection and non-detection of the target in the connecting area CX.

[5. Other embodiments]
Although the embodiments of the present disclosure have been described above, the present disclosure is not limited to the above-described embodiments, and various modifications can be made.

(5a) In the above embodiment, a predetermined range near the boundary of the detection area A of the field sensor S is set as the connecting area CX, but the present disclosure is not limited to this. For example, as shown in FIG. 9, all detection areas A other than the non-connecting areas set in the above embodiment may be set as connecting areas CX.

(5b) In the above embodiment, the connecting areas CX are individually set for each detection area A, but the present disclosure is not limited to this. For example, as shown in FIG. 10, a connecting area CX may be set on a straight line connecting two field sensors S1 and S2. In this case, as the connection area CX, a rectangular shape having an area center on the above straight line and formed by a side parallel to the vector V12 and having a side length E1 and a side perpendicular to the vector AB and having a side length E2. area can be set. If the center of the connecting area CX is separated from the field sensor S1 by a distance D1, the distance D1 may be set so as to satisfy the formula (1). However, D is the distance between the field sensors S1 and S2, R1 is the detection distance of the field sensor S1, and R2 is the detection distance of the field sensor S2.

D1/D=R1/(R1+R2) (1)
(5c) In the above embodiment, the connecting area CX is set regardless of the state of the tracked target, but the present disclosure is not limited to this. For example, when there is a tracking target that may move across the detection areas A1 and A2 of two adjacent field sensors S1 and S2, the connecting area CX is set only between the two field sensors S1 and S2. may be set. For example, when the movement vector indicating the direction of movement of the tracked target and the position vector indicating the direction of viewing the field sensor S2 from the field sensor S1 have a certain degree of parallelism, the tracked target is located in the detection area A1. , A2. The degree of parallelism between the movement vector and the position vector may be determined, for example, by determining whether the inner product of the normalized two vectors is equal to or greater than a certain value.

(5d) processing units 23, 33, 52 and techniques described in this disclosure by configuring a processor and memory programmed to perform one or more functions embodied by a computer program; It may be implemented by a provided dedicated computer. Alternatively, the processing units 23, 33, 52 and techniques described in this disclosure may be implemented by a dedicated computer provided by configuring the processor with one or more dedicated hardware logic circuits. Alternatively, the processing units 23, 33, 52 and techniques described in this disclosure are configured with a processor and memory and one or more hardware logic circuits programmed to perform one or more functions. It may also be implemented by one or more dedicated computers configured in combination with processors. Computer programs may also be stored as computer-executable instructions on a computer-readable non-transitional tangible storage medium. The method of realizing the function of each part included in the processing units 23, 33, 52 does not necessarily include software, and all the functions are realized using one or more pieces of hardware. good too.

(5e) A plurality of functions possessed by one component in the above embodiment may be realized by a plurality of components, or a function possessed by one component may be realized by a plurality of components. . Also, a plurality of functions possessed by a plurality of components may be realized by a single component, or a function realized by a plurality of components may be realized by a single component. Also, part of the configuration of the above embodiment may be omitted. Moreover, at least part of the configuration of the above embodiment may be added or replaced with respect to the configuration of the other above embodiment.

(5f) In addition to the server 5 as the target tracking device described above, a system having the target tracking device as a component, a program for causing a computer to function as the target tracking device, a semiconductor memory storing this program, etc. The present disclosure can also be realized in various forms such as a non-transitional substantive recording medium, a position estimation method, and the like.

DESCRIPTION OF SYMBOLS 1... Traffic control system, 2... Vehicle-mounted terminal, 3... Infrastructure sensor, 5... Server, 51... Communication device, 52... Processing part, 53... Storage part, 521... Information generation part, 522... Information distribution part, A... Detection area, CX... connecting area, S... field sensor.

Claims (4)

an information acquisition unit (5: S110) configured to acquire, from a plurality of sensors, sensor information including position information of the sensors and detection areas of the sensors, and target information including the positions of detected targets; ,
The position of the target estimated in each processing cycle is set as an estimated value, the position of the target predicted from the estimated value in the previous processing cycle is set as a predicted value, and the object acquired by the information acquisition unit is set as a predicted value. a reliability setting unit (5: S140) configured to set the reliability of each of the predicted value and the observed value using the position of the target indicated in the target information as the observed value;
Using the corresponding value pair, which is the predicted value and the observed value representing the same target, and the reliability set by the reliability setting unit for the corresponding value pair, the an estimated value updating unit (5: S190) configured to update the estimated value of the target;
an area setting unit (5: S130) configured to set a connecting area including at least part of a boundary of the detection area indicated in the sensor information and having lower detection accuracy than other areas in the detection area; and,
a reliability adjustment unit ( 5: S180) and
A target tracking device comprising a The target tracking device according to claim 1,
The target tracking device, wherein the reliability adjustment unit is configured to adjust the reliability of the predicted value for which the observed value forming the corresponding value pair does not exist so that the reliability increases. The target tracking device according to claim 1 or claim 2,
The reliability setting unit determines the position of the observed value in the connecting area, the degree of matching between the observed value and the predicted value, the type and speed of the target associated with the predicted value, and the target The reliability is calculated using at least one of the elapsed time since the first detection, the type and specifications of the sensor that generated the observed value, and the external environment when the observed value was obtained. A target tracker configured to set. The target tracking device according to any one of claims 1 to 3,
The area setting unit is configured to set the connecting area between the detection areas of two adjacent sensors.
Target tracker.

Images