Vehicle control system and vehicle control procedure

Abstract

Vehicle control system comprising: an integration unit that estimates, based on information from a sensor that detects information about the external area of a self-vehicle, information about a position and a speed of a target object present in an external area and errors of the position and speed, where the integration unit determines an error of a detection result from the detection result of a sensor detecting an external area of a vehicle in accordance with a characteristic of the sensor, a correlation between detection results of several sensors is determined and correlated detection results are integrated and the errors of the position and velocity of the target object are calculated.

Description

Technical field

[0001] The present invention relates to a vehicle control system that estimates a state of an object using information about the object detected by various types of sensors. background

[0002] The background of the related art includes the following prior art. In PTL (JP 2018-97765 A), when a radar target indicating an object detected by a radar and an image target indicating an object detected by an image pickup device are generated from the same object, a fusion target is generated by integrating the radar target and the image target. Then, calculation is performed using the position of the image target used to generate the fusion target in a width direction of the own vehicle as a lateral position and using a moving speed in the width direction of the own vehicle as a lateral speed.PTL 1 discloses an object detection device that, when no fusion target is generated by detecting a radar target and no image target is detected, generates a preliminary fusion target by the lateral position and lateral velocity of the image target used to generate the fusion target and a radar target detected by a radar target detecting unit (see abstract). Citation listPatent literature

[0003] PTL 1: JP 2018-97765 A Summary of the inventionTechnical problem

[0004] In the technique disclosed in PTL 1, a constant value is used for an observation error of a sensor (radar and image pickup device). Thus, grouping of a target object with a position estimated from multiple sensor values may be erroneously performed, and one object may be erroneously detected as multiple objects. In addition, the observation value of the sensor with high accuracy is not selected, and the detection result of the sensor is not integrated, although the trend of the error varies depending on the type of sensor. Thus, the overall detection accuracy may be lowered. Furthermore, the error of the sensor varies depending on the environment of the external area, and the influence of the external area is not considered. Solution to the problem

[0005] A representative example of the invention disclosed in this application is as follows. A vehicle control system includes an integration unit that estimates information about a position and a speed of a target object present in an external area and errors of the position and speed based on information from a sensor that detects information about the external area of a host vehicle. The integration unit estimates an error of a detection result from the detection result of a sensor that detects an external area of a vehicle in accordance with a characteristic of the sensor, determines a correlation between detection results of multiple sensors, and integrates correlated detection results and calculates the errors of the position and speed of the target object. Advantageous effects of the invention

[0006] According to one aspect of the present invention, it is possible to accurately obtain an error of an observation value of a sensor and improve the accuracy of a grouping process. Objects, configurations, and effects other than those described above will be clarified by the descriptions of the following embodiments. Brief description of the drawings Fig. 1 is a configuration diagram illustrating a vehicle control system according to an embodiment of the present invention. Fig. 2 is a flowchart illustrating an entirety of the integration processing in the present embodiment. Fig. 3 is a flowchart of a prediction update process in step S2. Fig. 4 is a diagram illustrating a process in step S2. Fig. Figure 5 is a diagram illustrating a grouping process (S3) in the related field. Fig. Figure 6 is a diagram illustrating the grouping process (S3). Description of the embodiments

[0007] An embodiment will be described below with reference to the drawings.

[0008] Fig. 1 is a configuration diagram illustrating a vehicle control system according to an embodiment of the present invention.

[0009] The vehicle control system in the present embodiment includes a self-vehicle movement detection sensor D001, an external area detection sensor group D002, a positioning system D003, a map unit D004, an input communication network D005, a sensor detection integration device D006, an autonomous driving plan determination device D007, and an actuator group D008. The self-vehicle movement detection sensor D001 includes a gyro sensor, a wheel speed sensor, a steering angle sensor, an acceleration sensor, and the like mounted on the vehicle, and measures a yaw rate, a wheel speed, a steering angle, an acceleration, and the like representing the movement of the self-vehicle.The external area detection sensor group D002 detects a vehicle, a person, a white line of a road, a sign, and the like outside the own vehicle and recognizes information about the vehicle, person, the white line, the sign, or the like. A position, a speed, and a type of object of an object such as a vehicle or a person are detected. The shape of the white line of the road including the position is recognized. For the statement, the position and content of a sign are recognized. As the external area detection sensor group D002, sensors such as a radar, a camera, and a sonar are used. The configuration and number of sensors are not particularly limited. The positioning system D003 measures the position of the own vehicle. As an example of the positioning system D003, there is a satellite positioning system.The map unit D004 selects and outputs map information around the own vehicle. The input communication network D005 acquires information from various information acquisition devices and transmits the information to the sensor detection integration device D006. The control unit area network (CAN), Ethernet, wireless communication, and the like are used as the input communication network D005. The CAN is a network generally used in a vehicle system. The sensor detection integration device D006 acquires information about the movement of the own vehicle, sensor object information, sensor road information, positioning information, and map information from the input communication network D005.The sensor detection integration device D006 then integrates the information components as the surroundings of the own vehicle and outputs the surroundings of the own vehicle to the autonomous driving plan determination device D007. The autonomous driving plan determination device D007 receives the information from the input communication network D005 and the surroundings of the own vehicle from the sensor detection integration device D006. The autonomous driving plan determination device plans and determines how the own vehicle should be moved and outputs command information to the actuator group D008. The actuator group D008 actuates the actuators in accordance with the command information.

[0010] The sensor detection integration device D006 in the present embodiment includes an information storage unit D009, a sensor object information integration unit D010, and a self-vehicle surrounding information integration unit D011. The information storage unit D009 stores information (e.g., sensor data measured by the external area detection sensor group D002) from the input communication network D005 and provides the information to the sensor object information integration unit D010 and the self-vehicle surrounding information integration unit D011. The sensor object information integration unit D010 acquires the sensor object information from the information storage unit D009 and integrates the information of the same object detected by multiple sensors as the same information.Subsequently, the sensor object information integration unit outputs the integration result as integration object information to the own-vehicle surrounding information integration unit D011. The own-vehicle surrounding information integration unit D011 acquires the integration object information and the own-vehicle movement information, the sensor road information, the positioning information, and the map information from the information storage unit D009. Subsequently, the own-vehicle surrounding information integration unit D011 integrates the acquired information as own-vehicle surrounding information and outputs the own-vehicle surrounding information to the autonomous driving plan determination device D007.

[0011] The sensor detection integration device D006 is configured by a computer (microcomputer) including an arithmetic operation device, a data memory, and an input / output device.

[0012] The arithmetic operation device includes a processor and executes a program stored in data memory. A portion of the processing performed by the arithmetic operation device executing the program may be performed by another arithmetic operation device (e.g., hardware such as a field-programmable gate array (FPGA) and an application-specific integrated circuit (ASIC)).

[0013] The data memory includes a ROM and a RAM, which are non-volatile storage elements. The ROM stores an unchangeable program (e.g., BIOS) and the like. The RAM includes a high-speed volatile storage element such as dynamic random access memory (DRAM) and a non-volatile storage element such as static random access memory (SRAM). The RAM stores a program executed by the arithmetic operation device and data used when the program is executed. The program executed by the arithmetic operation device is stored in a non-volatile storage element, which is a non-volatile storage medium of the sensor detection integration device D006.

[0014] The input / output device is an interface that transmits processing contents by the sensor detection integration device D006 to the outside or receives data from the outside in accordance with a predetermined protocol.

[0015] Fig. 2 is a flowchart illustrating an entirety of the integration processing in the present embodiment.

[0016] The information storage unit D009 stores sensor data. The sensor data is information about an object (a target object) detected by various sensors (radar, camera, sonar, and the like) of the external area detection sensor group D002, and includes data on a relative position, a relative speed, and a relative position / speed of the detected object, in addition to data on a distance and direction to the object. The relative position / speed may be represented by a range (e.g., a Gaussian distribution-type error ellipse) in which the object exists with a predetermined probability at a predetermined time. The Gaussian distribution-type error ellipse can be represented by a covariance matrix shown in the following expression and can be represented in another format.For example, the object's existence domain can be represented as another form by a general distribution other than Gaussian, which is estimated using the particle filter.

[0017] The covariance matrix shown in the following expression contains an element indicating a correlation between positions, an element indicating a correlation between velocities, and an element indicating a correlation between positions and velocities. P=(PxxPxyPxvxPxvyPyxPyyPyvxPyvyPvxxPvxyPvxvxPvxvyPvyxPvyyPvyvxPvyvy) where PxxPxyPyxPyy Elements that indicate a correlation between positions and PvxvxPvxvyPvyvxPvyvy Elements that indicate a correlation between speeds.

[0018] The data storage of the sensor object information integration unit D010 stores tracking data indicating a trajectory of an object detected by the various sensors of the external area detection sensor group D002.

[0019] In the integration processing, the sensor object information integration unit D010 first estimates an error of the sensor data (S1). This error is determined by the type of sensor, the position of an object detected within a detection area (e.g., the error is large when the distance to the object is large, and the object detected in the center of the detection area has a small error), and an external environment (brightness of the external area, visibility, rain, snowfall, temperature, and the like). Furthermore, when coordinate systems of components of sensor data output from the various sensors of the sensor group D002 for detecting the external area are different from each other, multiple components of sensor data are converted into a common coordinate system, and then an error of the sensor data is estimated. Details of an error estimation process (S1) will be described later.

[0020] The sensor object information integration unit D010 updates prediction data of the tracking data (S2). For example, assuming that the object represented by the tracking data performs a smooth, straight-line movement from the previously detected point without changing its moving direction and speed, the position of the object at the next time point is predicted, and the tracking data is updated. Details of a prediction data update process (S1) will be described later.

[0021] Subsequently, the sensor object information integration unit D010 performs a grouping process of integrating data representing an object from the position predicted using the tracking data and the position observed using the sensor data (S3). For example, an overlap between the error range of the position predicted using the tracking data and the error range of the position observed using the sensor data is determined, and the predicted position and the observed position where the error ranges overlap are grouped as data representing the same object. Details of a grouping process (S3) will be described later.

[0022] Subsequently, the sensor object information integration unit D010 integrates the observation results using the data determined as the group representing the same object (S4). For example, a weighted average of the predicted positions and the observed positions grouped as the data representing the same object is calculated, taking into account errors of the predicted positions and the observed positions, and an integrated position of the object is calculated.

[0023] The integrated position is then output as a fusion result and the tracking data is further updated (S5).

[0024] Fig. 3 is a flowchart of the prediction update process in step S2 of Fig. 2. Fig. 4 is a diagram illustrating a process in each step. In Fig. 4 the speed is shown by an arrow, the position is shown by a position on Fig. 4 and the position / relative velocity is represented by an error ellipse.

[0025] First, the sensor object information integration unit D010 acquires a first relative speed Vr_t1_t1, a first relative position X_t1_t1, and a first relative position / relative speed Pr_t1_t1 of an object around a vehicle at a predetermined time t1 (S21). The relative speed, the relative position, and the relative position / relative speed are generally represented in a following coordinate system (also referred to as a relative coordinate system) based on a position of the vehicle center of the own vehicle, but may be represented in a coordinate system based on the position of the sensor that measured the sensor data.

[0026] Subsequently, the sensor-object information integration unit D010 converts the relative speed data in the following coordinate system into absolute speed data in a fixed coordinate system. For example, the sensor-object information integration unit D010 uses the first relative position X_t1_t1 to convert the detected first relative speed Vr_t1_t1 and the first relative position / relative speed Pr_t1_t1 in the following coordinate system into a first absolute speed Va_t1_t1 and a first relative position / absolute speed Pa_t1_t1 in the fixed coordinate system (S22).

[0027] Subsequently, the sensor object information integration unit D010 obtains the position at time t2 from the position at time t1. For example, the sensor object information integration unit D010 converts the first absolute speed Va_t1_t1, the first relative position X_t1_t1, and the first relative position / absolute speed Pa_t1_t1 at time t1 with the position O_t1_t1 of the vehicle as the origin into the second absolute speed Va_t2_t1, the second relative position X_t2_t1, and the second relative position / absolute speed Pa_t2_t1 at time t2 (S23).

[0028] Subsequently, the sensor object information integration unit D010 updates the origin position of the coordinate system from time t1 to time t2, that is, from the coordinate system at time t1 to the coordinate system at time t2. For example, the sensor object information integration unit D010 updates the second relative position X_t2_t1, the second absolute speed Va_t2_t1, and the second relative position / absolute speed Pa_t2_t1 of the object with the position O_t1_t1 of the vehicle at time t1 as the origin to the second relative position X_t2_t2, the second absolute speed Va_t2_t2, and the second relative position / absolute speed Pa_t2_t2 of the object with the position O_t2_t1 of the vehicle at time t2 as the origin (S24).

[0029] When converting from the original position O_t1_t1 at time t1 to the original position O_t2_t1 at time t2, the measured values (i.e., the turning process) of the vehicle speed and the yaw rate of the own vehicle are used.

[0030] Since the measured values of vehicle speed and yaw rate contain errors, the error range indicated by the second relative position / absolute speed Pa_t2_t2 may be increased taking into account the error of vehicle speed and the error of yaw rate.

[0031] The sensor-object information integration unit D010 then converts the absolute speed data in the fixed coordinate system into relative speed data in the following coordinate system. For example, the sensor-object information integration unit D010 uses the second relative position X_t2_t2 to convert the second absolute speed Va_t2_t2 and the second relative position / absolute speed Pa_t2_t2 in the fixed coordinate system into the second relative speed Vr_t2_t2 and the second relative position / relative speed Pr_t2_t2 in the following coordinate system in the updated coordinate system (S25).

[0032] As described above, according to the prediction update process of the present embodiment, it is possible to calculate the relative position / relative speed (the error range) more accurately.

[0033] In addition, it is possible to improve the grouping performance of the sensor data of the target object and improve the determination performance of an operation plan.

[0034] Next, details of the grouping process (S3) are described.

[0035] For example, consider a case that Fig. 5, that is, a case where the observation values of a sensor A and a sensor B and the prediction update result are obtained, the error range of the observation value of the sensor is set to a constant value, and the error range after the prediction update is also set to a constant value. At an observation point 1, the error range of the observation value of sensor A, the error range of the observation value of sensor B, and the error range of the prediction update result overlap each other. Therefore, three target objects observed at the observation point 1 are integrated into one and recognized as one object. At the observation point 1 shown in Fig. As illustrated in Figure 5, the three error ranges overlap each other. Even in a case where the error range of the observation value of sensor A overlaps the error range of the prediction update result and the error range of the observation value of sensor B overlaps the error range of the prediction update result, that is, a case where the error range of the observation value of sensor A and the error range of the observation value of sensor B overlap each other beyond the error range of the prediction update result, the three target objects are integrated into one and recognized as one object. At an observation point 2, there is no overlap between the error range of the observation value of sensor A, the error range of the observation value of sensor B, and the error range of the prediction update result.

[0036] Therefore, the three target objects observed at observation point 2 are not integrated into one and are recognized as three objects.

[0037] Fig. 6 is a diagram illustrating a grouping process (S3) in the present embodiment. In the present embodiment, the grouping process is performed using the error range calculated in accordance with the type of sensor. For example, sensor A is a radar that measures a distance and direction to a target object. Sensor A has a small error in a distance direction (vertical direction), which is a direction from the sensor to the target object, but has a large error in a rotation direction (lateral direction), which is perpendicular to the distance direction. Sensor B is, for example, a camera that captures an image of the external area. Sensor B has a small error in the rotation direction (horizontal direction), but has a large error in the distance direction (vertical direction).Therefore, an error range is obtained taking into account the error characteristic depending on the type of sensor, as shown in . Fig. 6. When the grouping process is carried out using the error range calculated in this way, the error range of the observation value of sensor A, the error range of the observation value of sensor B, and the error range of the prediction update result overlap each other at the observation point 1, similar to the above description ( Fig.5). Thus, the three target objects observed at observation point 1 are integrated into one and recognized as one object. Furthermore, at observation point 2, the error range of the observation value of sensor A overlaps the error range of the prediction update result, and the error range of the observation value of sensor B overlaps the error range of the prediction update result. Thus, the three target objects observed at observation point 2 are integrated into one and recognized as one object.

[0038] In the error estimation process (S1) in the present embodiment, since the error is calculated in accordance with the type and characteristic of the sensor and the error range is adjusted, it is possible to accurately integrate target objects observed by the multiple sensors and recognize the target objects as one object. That is, since the accuracy of the grouping process is improved and the position of an object outside the vehicle can be accurately observed, it is possible to accurately control the vehicle.

[0039] Furthermore, in the present embodiment, the error can be calculated according to the observation result of the sensor. Therefore, the sensor object information integration unit D010 can determine the error range using a function that uses the observation result (e.g., the distance to the target object) as a parameter. The sensor object information integration unit D010 can determine the error range using an error table that is set in advance instead of the function.

[0040] For example, the sensor generally has a larger error at the end of a detection range than at the center of a detection range. Therefore, the error of the target object detected in the center of the detection range can be set smaller, and the error of the target object detected in a section closer to the end of a detection displacement can be set larger.

[0041] In addition, radar, as a type of sensor, has a small error in the distance (vertical) direction and a large error in the rotation (horizontal) direction, but the error range varies depending on the distance to the target object. That is, the error in the rotation (horizontal) direction increases proportionally with the distance, and the error in the distance (vertical) direction is essentially the same regardless of the distance. Furthermore, for radars with a range switching function, the error in the rotation (horizontal) direction increases on the wide-angle (short-range) side, and the error in the distance (vertical) direction is essentially the same regardless of the range.

[0042] As a type of sensor, a camera has a small error in the distance (vertical) direction and a large error in the rotation (horizontal) direction. However, the error range varies depending on the distance to the target object. That is, the error in the rotation (horizontal) direction increases proportionally with the distance, and the error in the distance (vertical) direction increases proportionally with the square of the distance.

[0043] In the present embodiment, as described above, since the error of the sensor is calculated in accordance with the position of the observed target object and the error range is adjusted, it is possible to accurately obtain the error of the observed value of the sensor. In particular, the error is increased when the distance to the target object is large; when the error is changed in accordance with the detection direction of the target object, the error of the target object near the end of the detection range is increased, and the error is increased on a wide-angle side. Therefore, it is possible to use an appropriate error range for the grouping process. Therefore, it is possible to accurately integrate target objects observed by multiple sensors and recognize the target objects as one object.That is, since the accuracy of the grouping process is improved and the position of an object outside the vehicle can be observed more accurately, it is possible to control the vehicle accurately.

[0044] In the present embodiment, the error can be calculated according to the environment of the external area. For example, the sensor error is small in good weather and large in rainy weather. Furthermore, the camera, as a type of sensor, has a small error during the day when the external area's illuminance is high, and a large error at night when the external area's illuminance is low.

[0045] As described above, in the present embodiment, since the error is calculated in accordance with the environment outside the vehicle, it is possible to calculate a more accurate error, improve the accuracy of the grouping process, and accurately control the vehicle.

[0046] The present invention is not limited to the above-described embodiment, and includes various modifications and equivalent configurations within the spirit of the appended claims. For example, the above examples are described in detail to explain the present invention in an easy-to-understand manner, and the present invention is not necessarily limited to a case including all of the described configurations. In addition, a portion of the configuration of one example may be replaced with the configuration of another example. Furthermore, the configuration of one example may be added to the configuration of another example.

[0047] For some components in the examples, other components may be added, deleted, or replaced.

[0048] In addition, some or all of the above-described configurations, functions, processing units, processing means, and the like may be realized by hardware by being designed with, for example, an integrated circuit, or may be realized by software by having a processor interpret and execute a program for realizing each function.

[0049] Information such as a program, a table, and a file that realizes each function can be stored in a data storage device, a storage device such as a hard disk and a solid state drive (SSD), or a recording medium such as an IC card, an SD card, and a DVD.

[0050] Control lines and information lines deemed necessary for the descriptions are illustrated, and not all control lines and information lines are necessarily shown in assembly. In practice, it can be assumed that almost all components are interconnected. List of reference symbols D001 Sensor for detecting movement of the own vehicle D002 Sensor group for detecting an external area D003 Positioning system D004 Card unit D005 Input communication network D006 Sensor detection integration device D007 Device for determining a plan for autonomous driving D008 Actuator group D009 Information storage unit D010 Sensor object information integration unit D011 Integration unit for environmental information of the own vehicle

Claims

[1] Vehicle control system comprising: an integration unit that estimates, based on information from a sensor that detects information about the external area of a self-vehicle, information about a position and a speed of a target object present in an external area and errors of the position and speed, where the integration unit determines an error of a detection result from the detection result of a sensor detecting an external area of a vehicle in accordance with a characteristic of the sensor, a correlation between detection results of several sensors is determined and correlated detection results are integrated and the errors in the position and velocity of the target object are calculated. [2] A vehicle control system according to claim 1, wherein the error is represented by a probability distribution. [3] A vehicle control system according to claim 1, wherein the integration unit predicts a position and a speed of the target object and errors of the position and the speed at a second time after a first time from a position and a speed of the target object and errors of the position and the speed at the first time without using the detection result of the sensor, a correlation between the detection results of the multiple sensors and a predicted position of the target object is determined and the detection result and the predicted position of the target object, which are correlated with each other, are integrated and the errors of the position and velocity of the target object are calculated. [4] The vehicle control system according to claim 1, wherein the integration unit estimates the error of each of the detection results after the detection results of the plurality of sensors are converted into a coordinate system. [5] The vehicle control system according to claim 1, wherein the integration unit determines the error of the detection result in accordance with the position of the target object. [6] The vehicle control system according to claim 5, wherein the integration unit estimates the error such that the error of the detection result increases as the distance to the target object increases. [7] The vehicle control system according to claim 6, wherein the integration unit estimates the error such that the error of the detection result is proportional to a square of the distance to the target object. [8] The vehicle control system according to claim 1, wherein the integration unit estimates the error such that the error of the detection result increases as an end of a detection range of the sensor approaches. [9] The vehicle control system according to claim 1, wherein the integration unit estimates the error such that the error of the detection result increases as the characteristic of the sensor changes to a wide-angle side. [10] A vehicle control method performed by a vehicle control system including an integration unit that estimates, based on information from a sensor that detects information about the external area of a subject vehicle, information about a position and a speed of a target object present in an external area and errors of the position and speed, the vehicle control method comprising: Estimating, by the integration unit, an error of a detection result from a detection result of a sensor detecting an external area of a vehicle in accordance with a characteristic of the sensor; Determining a correlation between detection results from multiple sensors by the integration unit; and Integrating correlated detection results and calculating the errors of the position and velocity of the target object by the integration unit.

Images