Obstacle detection method for a vehicle, electronic device, program product

Abstract

This application provides an obstacle detection method for vehicles, which acquires echo signals detected by radar sensors; determines whether adjacent echo signals originate from the same obstacle by the correlation of adjacent echo signal trajectories; and uses the echo data corresponding to the determined echo signals from the same obstacle as a first detection number that can be used to determine the obstacle. Electronic devices and software products are also provided.

Description

Technical Field

[0001] This application relates to vehicle driver assistance technology, and more specifically, to vehicle obstacle detection technology. Background Technology

[0002] Radar sensors are typically positioned around vehicle bodies to detect obstacles, and are therefore widely used in systems such as automatic parking, blind spot detection, and collision avoidance. Taking an ultrasonic sensor system (USS) as an example, multiple ultrasonic radars mounted on the vehicle scan the surrounding area and receive echo signals from obstacles. These echo signals can be used to determine obstacle-related information, including obstacle distance. Typically, the signal is filtered with a fixed threshold before determining obstacle information; this fixed threshold is usually set based on attributes such as echo amplitude and salience. In practice, it has been found that this filtering method may cause some usable data to be filtered out. Therefore, it is necessary to improve obstacle detection. Summary of the Invention

[0003] According to one aspect of this application, an obstacle detection method for vehicles is provided to effectively improve obstacle detection methods.

[0004] The obstacle detection method for a vehicle according to the example of this application may include: acquiring echo signals from a radar sensor; determining whether adjacent echo signals originate from the same obstacle by the correlation between the trajectories of adjacent echo signals; and using the echo data corresponding to the determined echo signals originating from the same obstacle as first detection data that can be used to determine the obstacle.

[0005] As a supplement or example, the obstacle detection method for vehicles may determine whether adjacent echo signals originate from the same obstacle by correlating the trajectories of adjacent echo signals. Specifically, this may include: determining the difference between the estimated echo distance at the current moment and the measured echo distance at the current moment; comparing the absolute value of the difference with a preset threshold; and determining that the echo signal at the current moment and the echo signal at the previous moment originate from the same obstacle when the absolute value is less than the preset threshold; wherein the estimated echo distance at the current moment is determined based on the measured echo distance at the previous moment.

[0006] As a supplement or example, the obstacle detection method for vehicles may further include: filtering the echo signal according to the properties of the radar echo signal; and using the echo data corresponding to the filtered echo signal as second detection data for determining the obstacle.

[0007] As a supplement or example, the obstacle detection method for vehicles may include one or any combination of echo amplitude, salience, and high salience.

[0008] As a supplement or example, the obstacle detection method for vehicles described herein may involve acquiring the echo signal of an obstacle detected by a radar sensor, which may be filtered out.

[0009] As a supplement or example, the obstacle detection method for vehicles may include determining the obstacle based on the first detection data and the second detection data.

[0010] For example, determining whether adjacent echo signals originate from the same obstacle by the correlation of adjacent echo signal trajectories may include processing the echo signals using a Kalman filter.

[0011] According to another aspect of this application, a parking method is provided, which may include any of the methods described above.

[0012] According to another aspect of this application, an electronic device is provided, which may include a storage unit for storing program instructions; and a processing unit configured to implement any of the methods described above when the program instructions are executed. As an example, the electronic device may be a parking controller or a domain controller.

[0013] According to this application, a computer program product is also provided, the program product including program instructions, which, when executed, can perform any of the methods described above. Attached Figure Description

[0014] This application will be more fully understood by referring to the following detailed description of specific embodiments in conjunction with the accompanying drawings, wherein the same reference numerals in the drawings refer to the same elements, wherein:

[0015] Figure 1 The structure of the vehicle obstacle detection architecture is illustrated in simple terms.

[0016] Figure 2 This is a flowchart of an obstacle detection method for a vehicle according to some embodiments of this application;

[0017] Figure 3 This is a flowchart of an obstacle detection method for a vehicle according to other embodiments of this application;

[0018] Figure 4 This is a flowchart of an obstacle detection method for a vehicle according to some embodiments of this application;

[0019] Figure 5 This is a schematic diagram of the structure of an obstacle detection device for a vehicle according to some embodiments of this application;

[0020] Figure 6 The actual and predicted traces of the echo signal are illustrated.

[0021] Figure 7 The illustration shows the increase in echo signal quantity and detection distance after executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the implementation methods of this application will be clearly and completely described below with reference to the accompanying drawings. It should be noted that the described implementation methods are only a part of the implementation methods of the technical solutions of this application, and not all of them. All other implementation methods obtained by those skilled in the art based on the implementation methods described in this application document without creative effort are covered by the protection scope of this application.

[0023] In this application, the term "obstacle" refers to an object that may affect the movement or parking of a vehicle. An obstacle may be an object such as a stone, or a living organism such as a small animal. In the examples of this application, "obstacle" is also referred to as an object.

[0024] Figure 1 The structure of the vehicle obstacle detection architecture is illustrated in the figure. As shown, a radar sensor 10 mounted on vehicle 1 detects objects around the vehicle. The radar sensor 10 is, for example, an ultrasonic radar or other sensor capable of detecting objects. The radar sensor 10 receives echo signals reflected from objects outside the vehicle and transmits the data carried by the echo signals (hereinafter referred to as echo data) to the controller 12. The controller 12 determines relevant information about the external object, such as its position and distance. The controller 12 can also issue a warning when necessary, such as when an obstacle is too close to the vehicle. The controller 12 can also construct obstacles based on the detected echo data and output them to the vehicle's display device for presentation. In the following text, the controller's determination of obstacle-related information based on echo data is also referred to as obstacle determination. Typically, before determining obstacles based on echo data, a threshold is first set based on the properties of the radar echo signal, such as echo amplitude, salience, and high salience, and the received echo signal is filtered using this threshold. This filtering is also referred to as echo attribute-based filtering in this paper.

[0025] Figure 2 This is a flowchart of an obstacle detection method for a vehicle according to some embodiments of this application. The method can be performed by the vehicle; for example, the vehicle's radar sensors detect obstacles, and the obstacles can be determined by a controller in the vehicle (e.g., a controller for a vehicle parking assistance system). Alternatively, either the radar sensor or the controller can be separately configured without using existing vehicle components.

[0026] like Figure 2As shown, in step S200, the echo signal of the obstacle detected by the radar sensor is acquired. For example, a radar sensor installed on a vehicle scans objects outside the vehicle and receives the echo signal reflected by the scanned object.

[0027] In step S202, the correlation between adjacent echo signal trajectories is used to determine whether adjacent echo signals originate from the same obstacle. For example, for the acquired echo signals, the controller determines the difference between the estimated echo distance at current time t and the measured echo distance at current time t; the absolute value of this difference is compared with a preset threshold; if the absolute value is less than the preset threshold, it is determined that the echo signal at current time t and the echo signal at the previous time t-1 originate from the same obstacle. In this example, the estimated echo distance at current time t is determined based on the measured distance of the echo signal at the previous time t-1. Therefore, the measured distance of the echo signal at the previous time and the measured distance of the echo signal at the current time are correlated through the estimated echo distance at the current time.

[0028] In step S204, the echo data corresponding to the echo signal from the same obstacle is used as the first detection data, which can be used to determine the obstacle. In other words, the echo data corresponding to the echo signal from the same obstacle determined in step S202 can be used to determine the obstacle, and is referred to as the first detection data to distinguish it from other data used to determine the obstacle. The echo data corresponding to the echo signal from the same obstacle can be stored in a predetermined storage area so that it can be used as the first detection data for subsequent use. It should be understood that in subsequent use, the first detection data may be used entirely to determine the obstacle, or it may be used partially to determine the obstacle model. For example, the following describes... Figure 3 In the described example, all the data corresponding to the echo signals from the same obstacle, determined by the trajectory tracking model, can be used to identify the obstacle. The following text, in conjunction with... Figure 4 In the described example, the data corresponding to the echo signal from the same obstacle, determined by the trajectory tracking model, may be partially used to determine the obstacle, as detailed in the following description. In summary, the first detection data determined in step S204, in whole or in part, can be used to determine the obstacle as needed.

[0029] Typically, the origin of an echo signal from the same obstacle is determined based on properties such as the amplitude of the echo signal. However, these properties largely depend on the properties of the object reflecting the echo, making it easy to filter out useful signals when filtering based on echo properties. For example, in parking, if the edge or corner of a parking space is made of a material very similar to the parking space floor, the echo signal reflected from the edge or corner is easily filtered out as an echo from the parking space floor, causing the parking system to misjudge the edge or corner. Again, using parking as an example, if there is an obstacle in the parking space that is very similar to the parking space floor material, the echo signal reflected from that obstacle may also be filtered out as a parking space floor echo, causing the parking system to miss the obstacle. However, using the obstacle detection method for vehicles exemplified in this application, the vehicle can determine whether the echo signal originates from the same obstacle based on the trajectory correlation of the echo signal, avoiding the discarding of useful signals due to the similarity of the material or other properties of the reflecting echo objects.

[0030] According to some examples of this application, the trajectory tracking model can be pre-trained to execute step S202, that is, the echo signal can be input into the trajectory tracking model to determine whether the echo signal comes from the same obstacle by the correlation of adjacent echo signal trajectories. As an example, the trajectory tracking model is a Kalman filter-based model. The Kalman filter includes a prediction period and an update period, and correspondingly, the trajectory tracking model also includes a prediction period and a period update. In the prediction period, the trajectory tracking model predicts the echo distance at the current time (i.e., the estimated echo distance) based on the echo measurement distance detected at the previous time. In the update period, the trajectory tracking model uses the echo measurement distance detected at the current time to correct the estimated echo distance at the current time. Through continuous iteration of the prediction period and the update period, the trajectory tracking model can continuously optimize the estimation of the radar sensor's sensing state, remove noise, and restore the true data. In the example of this application, the trajectory tracking model determines the estimated echo distance at the current time based on the echo distance at the previous time, and also considers parameters such as vehicle speed in the process of estimating the echo distance.

[0031] For example, the estimated distance of the echo at time t can be determined according to formula (1):

[0032] D_pred(t)=D_echo(t-1)+Vehspeed×Δt+ 1 / 2×a×Δt (1)

[0033] Where D_pred(t) is the estimated echo distance at the current time t; D_echo(t-1) is the echo measurement distance at the previous time t-1; Vehspeed is the vehicle speed; Δt is the time difference between the current time and the previous time; and a is the vehicle acceleration.

[0034] In the examples of this application, the echo measurement distance refers to the distance calculated based on the echo signal. For example, the echo measurement distance at the current time t or the echo signal measurement distance at the current time t is the distance of the obstacle reflecting the echo signal determined based on the echo signal at time t; the echo measurement distance at the previous time t-1 is the distance of the obstacle reflecting the echo signal determined based on the echo signal at time t-1.

[0035] For the Kalman filter mentioned here, its state prediction period includes state prediction according to equation (3) and covariance prediction according to equation (4), and its update period includes state update according to equation (5) and covariance update according to equation (6), as follows:

[0036]

[0037] Where x is the system state, corresponding to the trajectory tracking model, x represents the state of the echo point represented by the echo signal input into the trajectory tracking model; This is an estimated value of the system state, which corresponds to the trajectory tracking model and represents the estimated echo distance; is the optimal estimate of the state; corresponding to the trajectory tracking model, it represents the optimal echo prediction distance; A is the time state transition matrix, which describes how the system state evolves from one time step to the next time step, mapping the current state at a given time to the state at the next time step; B is the control input matrix, which describes how external control inputs affect the system, and this matrix is ​​usually associated with the external inputs that affect the system. The covariance prediction process is as shown in equation (4):

[0038] P + =APA T +Q (4)

[0039] Where P is the state prediction covariance matrix; Q is the process noise covariance matrix. The state update equation (5) in the update cycle is:

[0040]

[0041] Wherein, matrix K is the gain matrix with the smallest values ​​in the posterior estimated covariance matrix; H is the measurement matrix, which maps the system state to the measurement space and converts the state into actual sensed values, usually related to the sensing of the radar sensor; R is the measurement noise covariance matrix, which describes the noise level in the radar sensor's sensed signal. Covariance update equation in the update cycle (6):

[0042] P=(I-KH)P - (6)

[0043] Figure 3 This is a flowchart of an obstacle detection method for a vehicle according to other embodiments of this application. The method is performed by the vehicle, for example by the vehicle's parking system.

[0044] In step S300, the echo signal from the radar sensor detecting the obstacle is acquired. For example, the parking system controller acquires the echo signal detected by the ultrasonic radar deployed on the vehicle. In step S302, the echo signal received in step S300 is filtered according to the attributes of the radar echo signal. These attributes include, for example, one or a combination of echo amplitude, salience, and high salience. After filtering based on these attributes, filtered signals from objects with different attributes can be clearly distinguished. For example, the parking lot floor and edges will emit echo signals, but these echo signals are not from obstacles and should be filtered out from the echo data used to determine obstacles. Therefore, the parking system filters out these echo signals from non-obstacle objects based on preset thresholds according to echo attributes. In step S304, the echo data corresponding to the filtered echo signal is used as filtered obstacle data to determine obstacles. To distinguish it from the first detection data determined based on trajectory similarity mentioned above, the filtered obstacle data determined here is called the second detection data. As an example, this second detection data can be saved to a predetermined storage area.

[0045] According to a further example of this application, for the filtered echo signals, it can also be determined whether adjacent echo signals originate from the same obstacle based on the trajectory correlation of adjacent echo signals. More specifically, the filtered echo signals are input into the trajectory tracking model, as shown in step S306. In this example, the echo signals received by the trajectory tracking model are echo signals filtered out based on echo signal attributes. It should be understood that the objects that provide feedback echo signals usually have a certain size, and therefore there are usually more than one echo signal. Thus, the echo signals that are filtered out and input into the trajectory tracking model usually include multiple echo signals and not just one echo signal. In step S308, by using the correlation of the trajectories of adjacent echo signals in the received echo signals, the trajectory tracking model determines whether adjacent echo signals in the received echo signals originate from the same obstacle. In step S310, the echo signals determined to originate from the same obstacle are used as the first detection data. For example, if an echo signal from the same obstacle is identified from these filtered echo signals, these signals are stored as first detection data in the storage area storing second detection data. That is, the trajectory tracking model retrieves data from the same obstacle from the filtered data (step S302) and uses it as data to determine the obstacle. In step S312, the obstacle is determined based on both the first and second detection data, such as determining the obstacle's distance, position, and other relevant information. In some examples, this also includes constructing an obstacle based on this data for display on the vehicle's display screen.

[0046] according to Figure 3 The example method first filters the echo signals detected by the radar sensor based on echo attributes. Then, for example, a trajectory tracking model identifies the filtered echo signals to determine whether there are echo signals from the same obstacle. The echo data of the identified echo signals from the same obstacle is saved as the first detection data to the storage area storing the second detection data, which is used as the data to determine the obstacle.

[0047] Let's continue with the example of the corner parking space mentioned above. Execute... Figure 3 In the method shown, even if the echo signal reflected from the corner of the parking space is filtered out during the radar echo signal attribute filtering process due to the very similarity between the corner material and the ground material, the trajectory tracking model will still identify it from the filtered echo signal based on the correlation of the echo signal trajectory, and retrieve the echo data of these filtered echo signals to use as data for obstacle identification. Thus, in this example, the corner of the parking space can be identified without being mistakenly filtered out as a ground echo signal, as in existing technologies, causing the vehicle's parking system to fail to identify the corner of the parking space in a timely manner.

[0048] Figure 4This is a flowchart of an obstacle detection method for a vehicle according to other embodiments of this application. The method can be performed by a vehicle, for example, by a vehicle parking system. In step S400, an echo signal from a radar sensor detecting an obstacle is acquired. In step S402a, the echo signal is filtered; specifically, the echo signal received in step S400 is filtered according to the properties of the radar echo signal, such as echo amplitude, saliency, and high saliency, or any combination of two or three of these properties. In step S404a, the echo data corresponding to the filtered echo signal is used as second detection data to determine the obstacle, that is, obstacle-based data. As an example, the filtered second detection data can be saved to a predetermined storage area.

[0049] In step S402b, the echo signals acquired in step S400 are processed to determine whether adjacent echo signals originate from the same obstacle by analyzing the correlation between their trajectories. In a specific example, the echo signals acquired in step S400 are fed into a trajectory tracking model. The trajectory tracking model determines whether adjacent echo signals originate from the same obstacle by analyzing the correlation between their trajectories. In step S404b, the adjacent echo signals from the same obstacle determined by the trajectory tracking model are used as first detection data, and this first detection data is stored, for example, in a predetermined storage area for storing second detection data. In other cases, this first detection data may be stored in a storage area different from the predetermined storage area.

[0050] In step S406, obstacles are determined based on the second detection data and the first detection data. This includes determining the obstacle's location, distance, and even constructing the obstacle. It should be understood that in step S406, when determining obstacles based on the second and first detection data, there may be overlap between the two sources of data. Therefore, deduplication can be performed before determining obstacles. For example, when storing the first detection data in a predetermined storage area, it can be compared with already stored data. If the storage area already contains the data, it does not need to be stored again, thus avoiding duplicate storage.

[0051] The storage area in this application may be part of a storage component in a vehicle, such as a non-temporary memory.

[0052] According to this application, a parking method is also provided, which includes any of the obstacle detection methods for vehicles described above.

[0053] The method according to any example of this application can be implemented as a software module and integrated into the vehicle's controller for execution by the vehicle. Alternatively, the method according to any example of this application can be implemented as a software module loaded into a processor, which can be configured in the vehicle so that the vehicle can execute the method corresponding to the software.

[0054] Figure 5 These are schematic diagrams of the structure of an electronic device according to some embodiments of this application. For example... Figure 5 As shown, the electronic device includes a processing unit 50 and a storage unit 52. The storage unit 52 stores program instructions. The processing unit 50 is configured to execute the program instructions stored in the storage unit 52. According to the example of this application, the processing unit 50 can achieve the above-described combination during the execution of the program instructions. Figures 2 to 4 Any of the methods described. In some examples, the storage unit 52 and the processing unit 50 can be configured separately; for example, the storage unit 52 may be a memory, and the processing unit 50 may be a processor. In other examples, the storage unit 52 and the processing unit 50 are integrated to form a separate device, such as a processor including a storage unit. Figure 5 The described electronic device may be a vehicle parking controller or a domain controller.

[0055] The processing unit and processor described in this application can be, but are not limited to, any single-processor or multi-processor system of a wide variety of possible architectures, including field-programmable gate arrays (FPGAs), central processing units (CPUs), application-specific integrated circuits (ASICs), digital signal processors (DSPs), or graphics processing units (GPUs) arranged similarly or dissimilarly. The memory can be, but is not limited to, random access memory (RAM), read-only memory (ROM), or other electronic, optical, magnetic, or any other computer-readable medium. Furthermore, according to examples of this application, the processing unit and processor may also include the execution of program instructions in the process of implementing their functions. In addition, the processing unit and processor described in this application can be implemented as a controller in a vehicle.

[0056] Figure 6 The diagram illustrates the actual and predicted traces of the echo signal. The predicted trace is formed during the execution of any detection method according to the examples of this application, where the horizontal axis represents time in seconds and the vertical axis represents distance in meters. Figure 6 As shown, the sawtooth-like trace 60 is the actual trace, that is, the trajectory line formed by the measured distance of the echo signal, and the relatively smooth trace 62 is the predicted trace, that is, the trajectory line formed by the estimated distance estimated according to formula (1) during the execution of the embodiments of this application.

[0057] Figure 7This diagram illustrates the increase in echo signal quantity and detection distance after executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application. The horizontal axis represents the quantity, and the vertical axis represents the distance. Figure 7 As shown, several comparison groups are illustrated. In each comparison group, the darker-colored bars on the left represent the echo signals after executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application. These are labeled 70 in the figure, and only a portion of the bars in the comparison group are marked. The lighter-colored bars on the right in each comparison group represent the echo signals before executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application. These are labeled 72 in the figure, and only a portion of the bars in the comparison group are marked. Furthermore, the figure also illustrates the average detectable distance after executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application, as shown by line 74. Line 76 illustrates the average detectable distance before executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application. It can be seen that after executing the obstacle detection method according to the example of this application or using the obstacle detection device according to the embodiment of this application, both the number of available echo signals and the detectable distance increase.

[0058] According to examples of this application, a computer program product is also provided, which includes program instructions that, when executed, enable the obstacle detection method described according to any example of this application.

[0059] The obstacle detection method of this application can be implemented in vehicles with existing obstacle detection devices or systems. That is, the method according to this application can be implemented after the vehicle has performed obstacle detection in its original manner. For example, after the vehicle has performed obstacle detection and filtered the echo signal in its original manner, the filtered data can then be processed as follows: Figure 2 The process is illustrated below. For example, after the vehicle performs obstacle detection and acquires the echo signal in the original manner, the echo signal is filtered in the original way, and on the other hand, the echo signal is processed as follows: Figure 2 The process shown will filter the data obtained according to the original method and then... Figure 2 The data obtained after the processing shown is used to determine obstacles. Therefore, the embodiments provided in this application can be implemented without modifying the vehicle hardware structure, making them convenient to use.

[0060] When the obstacle detection method or the obstacle detection device or system according to the example of this application is executed, more echo data is available in the echo signal of the radar sensor than when the method according to the example of this application is not used. This greatly reduces the possibility of obstacles being missed. On the other hand, the increased available echo data also increases the detectable distance.

[0061] The technical features in the various embodiments of this application can be combined with each other to form new implementation methods without departing from the spirit of this application and without conflicting with each other. Although specific embodiments of this application have been shown and described in detail to illustrate the principles of this application, it should be understood that this application can be implemented in other ways without departing from such principles.

Claims

1. An obstacle detection method for a vehicle, characterized by, The method includes: Acquire the detection echo signal from the radar sensor; The correlation between adjacent echo signal trajectories is used to determine whether the adjacent echo signals originate from the same obstacle; The echo data corresponding to the echo signal from the same obstacle is used as the first detection data that can be used to determine the obstacle.

2. The method of claim 1, wherein, Determining whether adjacent echo signals originate from the same obstacle by analyzing the correlation between adjacent echo signal trajectories includes: Determine the difference between the estimated echo distance and the measured echo distance at the current moment; Compare the absolute value of the difference with a preset threshold; and When the absolute value is less than the preset threshold, it is determined that the echo signal at the current moment and the echo signal at the previous moment come from the same obstacle. The estimated distance of the echo at the current moment is determined based on the measured distance of the echo at the previous moment.

3. The method of claim 2, wherein, The method further includes: The echo signal is filtered according to its properties; The echo data corresponding to the filtered echo signal is used as the second detection data for determining the obstacle.

4. The method of claim 3, wherein, The attribute includes one or any combination of echo amplitude, significance, and high significance.

5. The method of claim 3, wherein, Acquiring the echo signal of the radar sensor detecting the obstacle means acquiring the echo signal that has been filtered out.

6. The method according to any one of claims 3 to 5, characterized in that, The method further includes: The obstacle is determined based on the first and second detection data.

7. The method of claim 1, wherein, Determining whether adjacent echo signals originate from the same obstacle by analyzing the correlation between their trajectories includes processing the echo signals using a Kalman filter.

8. A parking method characterized by, The parking method includes the method according to any one of claims 1 to 7.

9. An electronic device, comprising: The electronic device includes: Storage components are used to store program instructions; A processing unit configured to implement the method according to any one of claims 1 to 7 when executing the program instructions.

10. The electronic device of claim 9, wherein, The electronic device is used in the vehicle's parking system.

11. The electronic device of claim 10, wherein, The electronic device is a parking controller or a domain controller.

12. A computer program product, characterised in that, The program product includes program instructions that, when executed, implement the method according to any one of claims 1 to 7.

Images