Key Parameter Testing Method and Device for Vision-based Forward Collision Warning System

Abstract

The present invention provides a method and device for testing key parameters of a vision-based forward collision warning system, including: simulating the forward collision warning system using a preset simulation algorithm and building a simulation scenario; selecting key parameters to be tested and setting different preset values for the key parameters to be tested, and based on the simulated forward collision warning system and the simulation scenario, performing forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtaining data related to the key parameters of the host vehicle test vehicle in the current driving state; analyzing the data related to the key parameters to obtain the safety threshold boundary of the key parameters. The present invention tests the key parameters for the safe operation of the forward collision warning system through a simulation algorithm, and calculates the safety threshold boundary of the key parameters through scenario simulation, thereby contributing to a forward collision warning system with higher interpretability.

Description

Technical Field

[0001] The present invention relates to the technical field of assisted driving, and particularly to a method and device for testing key parameters of a vision-based forward collision warning system. Background Art

[0002] According to statistics, unconscious forward collisions caused by factors such as driver fatigue, drowsiness or negligence are the main causes of traffic accidents on highways. The Forward Collision Warning (FCW) system is a system that can perceive objects in front of the vehicle in real time through sensors such as cameras and radars, detect the distance between the vehicle and the target, and alert the driver, and is an important part of the advanced driver assistance system.

[0003] FCW can issue an alarm when there is a potential collision risk to help the driver avoid or mitigate a collision accident. Specifically, FCW constantly monitors the vehicle in front through a radar system, etc., judges the distance, azimuth and relative speed between the host vehicle and the vehicle in front, so as to estimate the time-to-collision (TTC). If the collision time is less than a threshold value, an alarm will be issued to remind the driver that the driver of the host vehicle should take emergency braking measures to avoid a collision with the vehicle in front. This can not only assist the driver to effectively avoid a vehicle forward collision and cause an accident, improve the driving safety of the vehicle, but also reduce the driver's operation burden and improve the driving comfort. When the driver ignores the front working conditions due to reasons such as operation errors, inattention or physical fatigue, traffic accidents can be effectively avoided.

[0004] Currently, when researching and developing existing forward collision warning systems, general values are usually used for the design of key parameters. The vehicle driving safety boundaries of the key parameters are unknown, and the specific impacts of the specific key parameters on the performance of the forward collision warning system are not understood, with low interpretability. Therefore, it is necessary to test the key parameters for the safe operation of the forward collision warning system, so as to obtain the vehicle driving safety boundaries related to the key parameters, clarify the impacts caused by the key parameters, and improve the interpretability of the developed warning system. Summary of the Invention

[0005] The present invention provides a method and device for testing key parameters of a vision-based forward collision warning system, to solve the defects in the prior art, realize the calculation of the vehicle driving safety boundaries related to the key parameters, and thus obtain a forward collision warning system with higher interpretability.

[0006] The present invention provides a method for testing key parameters of a vision-based forward collision warning system, including:

[0007] Simulate the forward collision warning system using a preset simulation algorithm and build a simulation scenario; wherein, the simulation scenario includes at least a camera, a front obstacle, a self-vehicle test vehicle, and a test scenario; wherein, the camera is installed on the self-vehicle test vehicle, and the test scenario includes the approach of the front obstacle and the self-vehicle test vehicle.

[0008] Select key parameters to be tested and set different preset values for the key parameters to be tested. Based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtain data related to the key parameters of the self-vehicle test vehicle in the current driving state; wherein, the key parameters to be tested include at least one of delay rate, packet loss rate over time, and resolution.

[0009] Analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0010] According to a method for testing key parameters of a vision-based forward collision warning system provided by the present invention, the step of selecting key parameters to be tested and performing tests specifically includes:

[0011] Select corresponding key parameters to be tested according to the test category-key parameter to be tested mapping relationship through a pre-selected test category.

[0012] According to a method for testing key parameters of a vision-based forward collision warning system provided by the present invention, when the test category is a quantitative test and the key parameter to be tested is one of delay, packet loss rate over time, and resolution, the step of performing forward collision warning simulation according to the preset values based on the simulated forward collision warning system and the simulation scenario until the preset end condition of the simulation scenario is reached, and obtaining data related to the key parameters of the self-vehicle test vehicle in the current driving state specifically includes:

[0013] Based on the simulated forward collision warning system and the simulation scenario, set the non-key parameters to be tested as fixed values, and perform forward collision warning simulation under different preset values of the key parameter to be tested, and obtain the collision time under each preset value as the data related to the key parameter.

[0014] Wherein, the collision time is the time interval between the self-vehicle test vehicle and the front obstacle when the warning is issued.

[0015] A method for testing key parameters of a vision-based forward collision warning system provided by the present invention. When the test category is perception accuracy test and the key parameter to be measured is resolution, based on the simulated forward collision warning system and the simulation scenario, forward collision warning simulation is carried out according to the preset value until the preset end condition of the simulation scenario is reached, and data related to the key parameters of the ego test vehicle in the current driving state is obtained, specifically including:

[0016] Add a preset sensor to the simulation scenario; wherein, the preset sensor is used to detect the actual relative distance between the ego test vehicle and the obstacle ahead.

[0017] Based on the simulated forward collision warning system and the simulation scenario, set the sampling frame rate to a fixed value, and carry out forward collision warning simulation under different preset resolution values, and obtain the detection distance and the actual relative distance at each resolution as data related to the key parameters; wherein, the detection distance is the real-time test distance output by the simulated forward collision warning system.

[0018] A method for testing key parameters of a vision-based forward collision warning system provided by the present invention, the test scenario specifically includes:

[0019] Longitudinal static obstacle ahead recognition test scenario, longitudinal slow-moving obstacle ahead recognition test scenario, and longitudinal decelerating obstacle ahead recognition test scenario.

[0020] A method for testing key parameters of a vision-based forward collision warning system provided by the present invention, the forward collision warning system includes a road video module, an object detection module, a ranging module, a warning module, and a warning information post-processing module;

[0021] The road video module is used to transmit the real-time image obtained by the camera to the object detection module;

[0022] The object detection module is used to obtain the RGB value of the real-time image according to the real-time image, perform object detection according to the RGB value of the real-time image, and obtain the upper and lower limits of the horizontal and vertical coordinates of the obstacle ahead in the real-time image;

[0023] The ranging module is used to obtain the real-time test distance from the obstacle ahead to the camera according to the height, downward deviation angle and focal length of the camera, and the upper and lower limits of the horizontal and vertical coordinates;

[0024] The warning module is used to calculate the time to collision according to the real-time test distance and the relative speed between the obstacle ahead and the ego test vehicle in the current driving state; compare the time to collision with a preset safety threshold to obtain collision information, and perform collision warning according to the collision information;

[0025] The warning information post - processing module is used to perform visual processing on the real - time test distance and the upper and lower limits of the horizontal and vertical coordinates, and output them to an oscilloscope or a video port for real - time observation.

[0026] The present invention also provides a key parameter test device for a vision - based forward collision warning system, including:

[0027] A simulation unit, configured to simulate a forward collision warning system using a preset simulation algorithm and build a simulation scenario; wherein, the simulation scenario at least includes a camera, a front obstacle, a self - vehicle test vehicle, and a test scenario; wherein, the camera is installed on the self - vehicle test vehicle, and the test scenario includes the approaching manner of the front obstacle and the self - vehicle test vehicle;

[0028] A test unit, configured to select key parameters to be tested, set different preset values for the key parameters to be tested, and perform forward collision warning simulation based on the simulated forward collision warning system and the simulation scenario according to the preset values until a preset end condition of the simulation scenario is reached, and obtain data related to the key parameters of the self - vehicle test vehicle in the current driving state; wherein, the key parameters to be tested include at least one of a time delay rate, a time packet loss rate, and a resolution;

[0029] An analysis unit, configured to analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0030] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, it implements the key parameter test method for the vision - based forward collision warning system as described in any one of the above.

[0031] The present invention also provides a non - transitory computer - readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the key parameter test method for the vision - based forward collision warning system as described in any one of the above.

[0032] The present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the key parameter test method for the vision - based forward collision warning system as described in any one of the above.

[0033] The key parameter testing method and device for the vision-based forward collision warning system provided by the present invention simulate the forward collision warning system by using a preset simulation algorithm and build a simulation scenario; wherein, the simulation scenario at least includes a camera, an obstacle ahead, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approaching manner of the obstacle ahead and the self-test vehicle; select the key parameter to be tested and set different preset values for the key parameter to be tested, and based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtain the data related to the key parameters of the self-test vehicle in the current driving state; wherein, the key parameter to be tested includes at least one of delay rate, time packet loss rate, and resolution; analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters. The present invention tests the key parameters for the safe operation of the forward collision warning system through a simulation algorithm, and calculates the safety threshold boundary of the key parameters through scenario simulation, so as to contribute to a forward collision warning system with higher interpretability. Description of the Drawings

[0034] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0035] Figure 1 It is one of the flow diagrams of the key parameter testing method for the vision-based forward collision warning system provided by the present invention;

[0036] Figure 2 It is another flow diagram of the key parameter testing method for the vision-based forward collision warning system provided by the present invention;

[0037] Figure 3 It is the camera ranging model of an embodiment of the key parameter testing method for the vision-based forward collision warning system provided by the present invention;

[0038] Figure 4 It is the model schematic diagram of the ranging module of the key parameter testing method for the vision-based forward collision warning system provided by the present invention;

[0039] Figure 5 The combined simulation result of the target detection module and the ranging module of an embodiment of the key parameter testing method for the vision-based forward collision warning system provided by the present invention;

[0040] Figure 6 It is a three - dimensional analysis diagram of an embodiment of the key parameter testing method for the vision - based forward collision warning system provided by the present invention;

[0041] Figure 7 It is a three - dimensional analysis diagram of another embodiment of the key parameter testing method for the vision - based forward collision warning system provided by the present invention;

[0042] Figure 8 It is a three - dimensional analysis diagram of yet another embodiment of the key parameter testing method for the vision - based forward collision warning system provided by the present invention;

[0043] Figure 9 It is a schematic structural diagram of the key parameter testing device for the vision - based forward collision warning system provided by the present invention;

[0044] Figure 10 It is a schematic structural diagram of the electronic device provided by the present invention. Detailed implementation manners

[0045] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts belong to the scope of protection of the present invention.

[0046] Below in conjunction with Figures 1 - 8 Describe the key parameter testing method for the vision - based forward collision warning system of the present invention. Figure 1 It is one of the flow schematic diagrams of the key parameter testing method for the vision - based forward collision warning system provided by the present invention. As Figure 1 shown, the method includes:

[0047] Step 110: Use a preset simulation algorithm to simulate the forward collision warning system and build a simulation scenario; wherein, the simulation scenario at least includes a camera, a front obstacle, a self - vehicle test vehicle and a test scenario; wherein, the camera is installed on the self - vehicle test vehicle, and the test scenario includes the approaching manner of the front obstacle and the self - vehicle test vehicle.

[0048] It should be noted that the present invention takes the forward collision warning system (FCW) as an example for illustration, but it does not represent a limitation to the present invention. The key parameter testing method for the vision - based forward collision warning system provided by the present invention can be extended to the testing of the key safety operation parameters of any vision - based forward collision warning system.

[0049] The forward collision warning system (FCW) mentioned in the embodiments of the present invention can obtain real-time images of a vehicle driving on the road through a camera, detect vehicle targets on the road through a target detection algorithm, and calculate the distance between the two vehicles, so as to determine whether the time to collision (TTC) between the host vehicle and the preceding vehicle is lower than a threshold. When the system determines that there is a warning situation, the system should immediately issue a collision warning message and remind the driver. The FCW system itself will not take any braking measures to avoid collisions or control the vehicle.

[0050] Specifically, the fundamental purpose of the present invention is to propose the requirements for key indicators such as the delay / packet loss, resolution, and bandwidth-induced quantization, perception output accuracy, and variance of the perception encoding and decoding of a vehicle-based vision forward collision warning system. In some embodiments, the key parameters include delay, packet loss, resolution, and bandwidth, etc.

[0051] In the specific implementation process, based on the "Performance Requirements and Test Methods for Forward Collision Warning Systems in Intelligent Transport Systems" or the "Test Procedures for Autonomous Driving Functions of Intelligent Connected Vehicles" (trial) formulated by the International Organization for Standardization (ISO) or the National Standards of the People's Republic of China (GB / T), in a simulation environment, test and analyze the safety operating boundaries of the vision-based forward collision warning system under the influence of key indicators such as delay / packet loss, resolution, and bandwidth-induced quantization, perception output accuracy, and variance, establish a target perception model based on the detection method, and obtain the quantitative analysis results of the key indicators.

[0052] Furthermore, first, the forward collision warning system is described as follows: The vision-based forward collision warning system can obtain vehicle road driving images through a camera, detect the preceding vehicle through a target detection algorithm, and calculate the distance between the host vehicle and the preceding vehicle through a ranging module, so as to determine whether the safety distance between the host vehicle and the preceding vehicle is too low. The system needs to issue a collision warning within a specified time range, and finally provide a warning signal to the driver through a display device; when the warning situation is not met, there should be no false alarm situation in the system; the vision-based forward collision warning system should be able to detect the main vehicles driving on the road under different lighting conditions.

[0053] To simulate the forward collision warning system, a preset simulation algorithm is used for simulation. The forward collision warning algorithm in the forward collision warning system is designed on a simulation platform, and then the forward collision warning algorithm is solidified on a hardware processing platform to process the data acquired by the camera to simulate the warning process. That is to say, the overall simulation device includes an image acquisition sensor (camera) and a hardware processing platform (the solidified forward collision warning algorithm for simulating forward collision warning). In some embodiments, the simulation device also includes a warning signal display device.

[0054] In a specific embodiment, the hardware configuration of the test system is shown in Table 1 below, and the software configuration of the test system is shown in Table 2 below.

[0055] Table 1 Test System Configuration (Hardware)

[0056] Name Specification CPU Gen Intel(R)Core(TM)i7 - 12700 GPU RTX 3070Ti RAM 64.0GB Hard Disk 2.0T

[0057] Table 2 Test System Configuration (Software)

[0058] Name Version Operating System Windows10 Matlab R2016b Prescan 8.5.0 CUDA 11.1 Pytorch 1.12

[0059] In addition, a preset simulation algorithm needs to be used to simulate and build the forward collision scenario, including the simulation of the camera, the front obstacle, the self-test vehicle, and the test scenario. The test scenario includes the approaching methods of the front obstacle and the self-test vehicle.

[0060] In some embodiments, the simulation experiment uses the joint simulation of Matlab / Simulink and Prescan. The forward collision warning algorithm design is implemented through Matlab / Simulink, and the scenario and sensor design are implemented through Prescan. It should be emphasized that the front obstacle can be the vehicle in front or other obstacles, and the present invention does not limit this.

[0061] In a specific embodiment, the test environment of the vision-based forward collision warning system (FCW) is built in PreScanGUI8.5.0, where the self-vehicle uses Actors / Cars&Motors / Audi A8 Sedan. The actual Bounding Box size of this vehicle is: length 5.21m, width 2.04m, height 1.44m. The vehicle in front selects Actors / Targets / GuideSoftTarget. The actual Bounding Box size of this vehicle is: length 3.82m, width 1.67m, height 1.40m.

[0062] In addition, the camera in Prescan is used to collect vehicle road driving images, and the TIS sensor (Technology Independent Sensor, TIS) displays the actual distance. It should be noted that for the simulation of the forward collision warning system, the camera and the TIS sensor are installed on the self-test vehicle to obtain the forward working conditions.

[0063] In addition, it should be emphasized that the parameters of the vehicle and the camera can be designed according to the actual situation, and the embodiments of the present invention do not limit this.

[0064] Furthermore, for the sensors required in subsequent tests, the design is implemented through Prescan. In a specific embodiment, the sensor uses the AIR sensor integrated in Prescan. It can be understood that the AIR sensor, as an ideal sensor, is used to calculate the actual relative distance between the preceding vehicle and the host vehicle, so as to compare the relative distance measured by the subsequent embodiment based on the test method provided by the present invention with the actual relative distance.

[0065] In some embodiments, the test scenarios specifically include: a longitudinal static front obstacle recognition test scenario, a longitudinal slow-moving front obstacle recognition test scenario, and a longitudinal decelerating front obstacle recognition test scenario.

[0066] Specifically, the test scenarios include: 1) Longitudinal static front obstacle recognition ability test, that is, the host vehicle approaches the stationary preceding vehicle (front obstacle) at a constant speed. In a specific embodiment, the host vehicle starts from a position 150 m behind the preceding vehicle and drives towards the preceding vehicle at a constant speed of 72 km / h.

[0067] 2) Longitudinal slow-moving front obstacle recognition ability test, that is, the host vehicle approaches the preceding vehicle (front obstacle) driving at a constant speed, and the speed of the preceding vehicle (front obstacle) is less than the speed of the host vehicle. In a specific embodiment, the preceding vehicle drives at a constant speed of 32 km / h in the middle of the lane, and the host vehicle starts from a position 150 m behind the preceding vehicle and drives towards the preceding vehicle at a constant speed of 72 km / h.

[0068] 3) Longitudinal decelerating front obstacle recognition ability test, that is, the host vehicle follows the preceding vehicle (front obstacle) driving at a constant speed, and the preceding vehicle (front obstacle) suddenly decelerates continuously. In a specific embodiment, the host vehicle follows the preceding vehicle and drives at a constant speed of 72 km / h in the middle of the lane. The host vehicle is 30 m behind the preceding vehicle. After following for 7 s, the preceding vehicle starts to decelerate, and the deceleration is maintained at 0.3g within 1.5 s after braking.

[0069] In addition, it should be emphasized that at the end of the simulation experiment, the driver of the host vehicle should take braking measures to avoid colliding with the preceding vehicle.

[0070] In addition, regarding the test roads involved in the test scenarios, it should be emphasized that in order to increase the understanding of key parameters, the present invention does not specifically limit the conditions of the test roads. However, in order to be in line with the actual situation and increase the reference value of the test results, the visible lane markings on its road surface comply with the provisions of GB5768, and it includes both straight road scenarios and curved road scenarios.

[0071] Furthermore, in order to increase the universality of the test results, in some embodiments, the conditions of the roads in most real situations are used as a reference, and the test road is set to an asphalt or concrete pavement on a dry platform by adjusting the parameters of the simulation platform; the horizontal visibility is greater than 1 km; the temperature is in the range of -20-40 degrees Celsius, and the test is carried out under daylight conditions.

[0072] Step 120: Select key parameters to be tested and set different preset values ​​for the key parameters to be tested; based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values ​​until the preset end conditions of the simulation scenario are reached, and obtain key parameter-related data of the self-vehicle test vehicle in the current driving state; wherein the key parameters to be tested include at least one of delay rate, time packet loss rate and resolution.

[0073] First, based on the constructed simulation scenario and the forward collision avoidance warning system simulated by the algorithm, different preset values ​​of the key parameters to be tested are set to simulate the entire process of the forward collision avoidance warning.

[0074] Specifically, the simulation of the entire process of the forward collision warning system includes: the FCW system obtains the road image of the vehicle in real time through the image sensor (camera), and then the algorithm solidified on the hardware processing platform (simulating the forward collision warning system) processes the recorded real-time image. Subsequently, the target detection algorithm calculates the real-time distance between the vehicle and the vehicle in front, and finally the forward collision warning system evaluates the possibility of a collision based on the current driving state of the vehicle. Depending on the test scenario, when the distance between the vehicle and the vehicle in front is less than the safe distance or the collision time is less than the preset threshold, the system will actively send out a collision warning signal and provide the driver with a warning signal through the display device.

[0075] In some embodiments, the forward collision avoidance warning system includes a road video module, a target detection module, a range finding module, a warning module, and a warning information post-processing module;

[0076] The road video module is used to transmit the real-time image acquired by the camera to the target detection module;

[0077] The target detection module is used to obtain a real-time image RGB value according to the real-time image, perform target detection according to the real-time image RGB value, and obtain the upper and lower limits of the horizontal and vertical coordinates of the front obstacle in the real-time image;

[0078] The distance measurement module is used to obtain the real-time test distance from the front obstacle to the camera according to the height, downward deflection angle and focal length of the camera, as well as the upper and lower limits of the horizontal and vertical coordinates;

[0079] The warning module is used to calculate the collision time according to the real-time test distance and the relative speed of the front obstacle and the self-vehicle test vehicle in the current driving state; compare the collision time with a preset safety threshold to obtain collision information, and issue a collision warning according to the collision information;

[0080] The warning information post-processing module is used to visualize the real-time test distance and the upper and lower limits of the horizontal and vertical coordinates, and output them to an oscilloscope or a video port for real-time observation.

[0081] Specifically, Figure 2 FIG. 1 is a schematic diagram of a process flow of a forward collision avoidance warning system simulated by an embodiment of the present invention. Figure 2 As shown, first, the road video module obtains a real-time image of the front working conditions through the on-board monocular camera. It can be understood that the video image input by the on-board monocular camera is an M×N×3 color image array, also known as an RGB image. Naturally, it should be pointed out that, for the convenience of description, the embodiment of the present invention adopts the setting of a monocular camera for illustration, which does not mean a limitation of the present invention. The key parameter test method for safe operation of the forward collision warning system provided by the present invention can be simulated with any camera. Since the target detection algorithm used in the embodiment of the present invention is an end-to-end one-stage target detection algorithm, it does not require image preprocessing, so it is only necessary to transfer the real-time image of the vehicle during driving to the target detection module. In some embodiments, the camera provided by Prescan is used to acquire road images. It should be pointed out that the camera can adjust the resolution, installation position, sampling frame rate, etc. to meet the test requirements.

[0082] In the target detection module, target detection is performed based on the acquired real-time image. The embodiments of the present invention do not limit the specific algorithm of target detection. In some embodiments, YOLOv5 is used to implement target detection, and the target in the image and its location are detected. The target in the image can be shown with a label box, and then the coordinates of the pixel matrix of the target in the image are obtained, including the xmin, xmax, ymin and ymax of the target in the image. At the same time, in some embodiments, with the cooperation of the ranging module, the distance of the target from the camera can also be marked in the figure, such as Figure 5 shown.

[0083] The advantages of YOLOv5 are its high flexibility, small model size, fast detection speed, easy deployment to embedded devices, and its accuracy can meet the general application requirements. Specifically, YOLOv5 belongs to a one-stage detection algorithm, also known as the One-Stage object detection algorithm. Its principle is to directly extract features from the original image and predict the category and location of objects through regression analysis. This end-to-end object detection method significantly shortens the time consumed by the algorithm to run. It can be understood that YOLOv5 is a high-precision object detection algorithm that can be used to detect vehicles driving on the road. Compared with YOLOv4 and previous versions, YOLOv5 has been further improved in terms of detection accuracy and practicality. According to the official test results, when using the COCO dataset for training and the IoU (Intersection over Union) is 0.5, the mAP50 (mean average precision) value of YOLOv5 on the test set reaches 89.0%. Compared with the mAP50 value of YOLOv4, YOLOv5 has higher accuracy in vehicle detection tasks. In addition, YOLOv5 also has a faster detection speed and a smaller model size, and can achieve a faster inference speed while maintaining a high level of accuracy.

[0084] In the ranging module, the monocular vision ranging algorithm is used in the data verification stage in the embodiments of the present invention. It can be understood that this algorithm is based on the principle of pinhole imaging. The object to be measured passes through the pinhole of the camera and is incident on the imaging element of the camera. The imaging element of the camera converts the received optical signal into a digital signal through the digital signal switching functional module and transmits the digital signal to the computer chip, and the computer chip restores the scene where the object to be measured is located. The converted digital signal consists of the horizontal x and the vertical y to form a pixel point matrix, which is stored on the display. Therefore, the distance measurement based on monocular vision is based on the coordinate transformation of pixel points. The conversion between the pixel coordinate system, the camera optical center coordinate system and the world coordinate system is realized by using matrix multiplication to achieve distance measurement, and the required calculation steps and parameters are numerous. The calibration of parameters and the conversion between coordinate systems seriously affect the speed and accuracy of visual measurement. Therefore, in order to improve the accuracy, real-time performance and convenience of measurement in practical applications, it is necessary to obtain the downward deflection angle of the camera in real time through the attitude sensor according to the actual scenario of the application. The ranging model diagram used in the experiment is as Figure 4 shown.

[0085] Figure 4The angle between the central optical axis and the horizontal line is α, which is the depression angle of the camera being α. It is set that α > 0 when the optical axis deflects downward, and α < 0 when the optical axis deflects upward. M is the intersection point of P′P′x and the horizontal plane, and N is the intersection point of P′P′y and the horizontal plane, that is, OMN forms a horizontal plane perpendicular to OO′, and OO′ is denoted as the focal length f. Since the horizontal plane is parallel to the ground, so ∠OPO′ = ∠P′OM = ∠P′OP′x + ∠P′xOM. In ΔNO′′O, NO″⊥OO″, therefore:

[0086] O″N = f·tanα

[0087] In O″P′xO, OP′x = x, OO″⊥OO″P′x, so there is:

[0088]

[0089] P′xM = O″N, in ΔP′xOM, OP′x⊥P′xM, so there is:

[0090]

[0091] In ΔP′OP′x, OP′x⊥PP′x, so there is:

[0092]

[0093] In ΔOPO′, PO′⊥OO′, so there is

[0094]

[0095] According to the monocular ranging model, the height H of the camera, the downward deflection angle α and the camera focal length f are obtained, and then the ymin, xmax, ymin and ymax of the target in the image are obtained by the target detection algorithm. The distance d from the measured object to the camera can be calculated by the above formula, and the visual ranging formula used is as follows:

[0096]

[0097] The value of d can be obtained as:

[0098]

[0099] Furthermore, the specific processing method of the images captured by the camera in Prescan using Simulink includes: such as Figure 3As shown in the figure, first, the camera obtains the RGB values of the image through the monocular camera module, and then the RGB values are passed into the distance measurement module. This module integrates the YOLOv5 object detection model and develops a distance measurement algorithm based on it. Finally, the distance measurement result is output. The function of the DISPLAY module is to display the image detected by the camera to the user. Among them, 1 / Z represents the delay module

[0100] In the warning module, warning information is calculated according to the result output by the distance measurement module. Specifically, TTC is a commonly used indicator in the forward collision warning system, representing the time interval until the vehicle in front or an obstacle collides. The calculation method of TTC is as follows:

[0101] First, the distance (D) and relative speed (V rel ) between the host vehicle and the vehicle in front or an obstacle need to be calculated. Then, TTC is calculated using the following formula:

[0102]

[0103] where D is the distance between the host vehicle and the vehicle in front or an obstacle, and V rel is the relative speed between the host vehicle and the vehicle in front or an obstacle.

[0104] If V rel is negative, indicating that the host vehicle is approaching the vehicle in front or an obstacle, then TTC is negative, indicating that a collision has occurred or is about to occur. If V rel is positive, indicating that the host vehicle is moving away from the vehicle in front or an obstacle, then TTC is positive, indicating that a collision is less likely to occur.

[0105] Furthermore, the threshold value of TTC (Time To Collision) is usually set according to the specific application scenario and system requirements. Generally, the smaller the threshold value of TTC, the shorter the time to detect danger, and accordingly it will be more conservative, but it may also lead to an increase in the false alarm rate; on the contrary, the larger the threshold value of TTC, the more confident it will be, but at the same time it may also reduce the number of detected dangers.

[0106] It should be noted that TTC is only an indicator and cannot be used as the sole criterion for judging whether a collision has occurred. Other information, such as the vehicle's acceleration and steering angle, also needs to be combined to comprehensively judge whether there is a collision risk. At the same time, the calculation of TTC is also affected by measurement errors and model assumptions, and needs to be calibrated and adjusted in actual applications.

[0107] In the warning information post - processing module, information such as the position of the vehicle ahead and the distance between the host vehicle and the vehicle ahead can be obtained according to the aforementioned module. After obtaining information such as the position of the vehicle ahead and the distance between the host vehicle and the vehicle ahead, these information need to be processed for real - time observation output to an oscilloscope or a video port. In some embodiments, a target detection algorithm is used to frame the target object and mark the distance between the host vehicle and the vehicle ahead.

[0108] The present invention uses simulation software to simulate the forward collision warning system provided in the above - mentioned embodiments, and tests key parameters in combination with the constructed simulation scenario. Then, based on the data obtained from the simulation (referred to as key - parameter - related data), the failure boundaries of FCW in the simulation system are tested under the conditions of quantization, perception output accuracy, and variance caused by different key parameters. Among them, the significance of the failure boundary is to find the safety threshold boundary with error as the guide.

[0109] In the actual operation process, key parameters include resolution, time delay, packet loss rate, etc. One or more of them are extracted as the key parameters to be tested, and the rest are non - tested key parameters. The non - tested key parameters are set to fixed values, and different preset values are set for the key parameters to be tested and simulations are carried out, so that a plurality of data obtained during the simulation process are used as key - parameter - related data for safety operation analysis.

[0110] During the analysis process, it is necessary to quantitatively test the impact of key parameters on safe operation. For example, the impact of key parameters on safe operation is studied through TTC. At the same time, since the output accuracy may also affect safe operation, it is also necessary to conduct perception tests on the output accuracy. In addition, the mean absolute error and / or variance can be used as indicators for analysis.

[0111] In some embodiments, the selection of key parameters to be tested for testing specifically includes:

[0112] Select the corresponding key parameters to be tested according to the pre - selected test category through the test - category - key - parameter - to - be - tested mapping relationship.

[0113] Specifically, there is a pre - set mapping relationship between the test category and the key parameters to be tested, that is, the test - category - key - parameter - to - be - tested mapping relationship. Specifically, when the test category is perception accuracy test, the key parameter to be tested is resolution. When the test category is quantization test, the key parameter to be tested is one of time delay, packet loss rate, and resolution.

[0114] When conducting the simulation, according to the pre - selected test category, the corresponding key parameters to be tested are determined through the test - category - key - parameter - to - be - tested mapping relationship, and then the forward collision warning system is simulated based on the algorithm and the simulation scenario is constructed for simulation.

[0115] It should be noted that the simulation of forward collision warning has an end condition. Specifically, the end condition is related to the simulation scenario, that is, each simulation scenario has a corresponding preset end condition.

[0116] Based on the above embodiments, in the simulation scenario corresponding to the longitudinal static forward obstacle recognition ability test, in a specific embodiment, the host vehicle starts from a position 150 m behind the leading vehicle and drives uniformly towards the leading vehicle at a speed of 72 km / h. The corresponding preset end conditions include: if the system issues a collision warning when the time to collision (TTC) is not less than 2.70 s, then this experiment passes and ends; if the system issues a collision warning within the range where the time to collision (TTC) is less than 2.70 s, then this experiment fails and ends; if the system still does not issue a collision warning when the time to collision (TTC) drops to 2.43 s, this experiment should be terminated immediately.

[0117] In the simulation scenario corresponding to the longitudinal slow forward obstacle recognition ability test, in a specific embodiment, the leading vehicle drives uniformly at 32 km / h in the middle of the lane, and the host vehicle starts from a position 150 m behind the leading vehicle and drives uniformly towards the leading vehicle at a speed of 72 km / h. The corresponding preset end conditions include: if the system issues a collision warning when the time to collision (TTC) is not less than 2.10 s, then this experiment passes and ends; if the system issues a collision warning within the range where the time to collision (TTC) is less than 2.10 s, then this experiment fails and ends; if the system still does not issue a collision warning when the time to collision (TTC) drops to 1.89 s, this experiment should be terminated immediately.

[0118] In the simulation scenario corresponding to the longitudinal deceleration forward obstacle recognition ability test, in a specific embodiment, the host vehicle follows the leading vehicle and drives uniformly at 72 km / h in the middle of the lane. The host vehicle is 30 m behind the leading vehicle. After following for 7 s, the leading vehicle starts to decelerate, and the deceleration is maintained at 0.3g within 1.5 s after braking. The corresponding preset end conditions include: if the system issues a collision warning when the time to collision (TTC) is not less than 2.40 s, then this experiment passes and ends; if the system issues a collision warning within the range where the time to collision (TTC) is less than 2.40 s, then this experiment fails and ends; if the system still does not issue a collision warning when the time to collision (TTC) drops to 2.16 s, this experiment should be terminated immediately.

[0119] In some embodiments, when the test category is a quantitative test and the key parameter to be measured is one of delay, time packet loss rate, and resolution, based on the simulated forward collision warning system and the simulation scenario, the forward collision warning simulation is performed according to the preset value until the preset end condition of the simulation scenario is reached, and the data related to the key parameters of the host vehicle test vehicle in the current driving state is obtained. Specifically, it includes:

[0120] Based on the simulated forward collision warning system and the simulation scenario, set the non-critical parameters to be measured as fixed values, and conduct forward collision warning simulations under different preset values of the critical parameters to be measured, and obtain the collision time under each preset value as the data related to the critical parameters;

[0121] Wherein, the collision time is the time interval between the test vehicle of the host vehicle and the obstacle in front when the warning is issued.

[0122] Specifically, when the test category is a quantization test, the critical parameter to be measured is one of the delay, packet loss rate over time, and resolution. At this time, a quantization test can be carried out. For example, the impact of each critical parameter to be measured on safe operation can be quantified as a TTC index for testing and analysis.

[0123] It can be understood that, in order to further explain the quantization test, based on the above embodiments, the present invention further gives examples from Test Scenario 1 (Longitudinal Static Obstacle in Front Recognition Ability Test, the speed of the vehicle in front is 0, the speed of the host vehicle is 72 km / h, the real-time distance between the host vehicle and the vehicle in front is set to 60 m, and the threshold of Test Scenario 1 is 2.70 s), Test Scenario 2 (Longitudinal Slow Moving Obstacle in Front Recognition Ability Test, (the speed of the vehicle in front is 32 km / h, the speed of the host vehicle is 72 km / h, the distance between the host vehicle and the vehicle in front is set to 30 m, and the threshold of Test Scenario 2 is 2.10 s), Test Scenario 3 (the speed of the vehicle in front is 69 km / h, the speed of the host vehicle is 72 km / h, the real-time distance between the host vehicle and the vehicle in front is 25 m, and the threshold of Test Scenario 3 is 2.40 s).

[0124] In the case of the delay test, it can be understood that in an unreliable communication scenario, relying on the camera to detect the distance between the host vehicle and the vehicle in front during the driving process, there is a delay in the position, and the delay will reduce the performance of the system. For the delay in unreliable communication, the delay module in the Simulink library is used in this test. Set the camera resolution (i.e., the non-critical parameter to be measured in this test) to 1280×960, and set different time delays (i.e., the critical parameters to be measured) for the test. In order to conduct comparative analysis, different sampling frame rates are also set for the simulation experiment: the camera sampling frame rates are uniformly set to 25 fps, 30 fps, 40 fps, and 60 fps, and the data related to the critical parameters obtained are shown in Table 3-8.

[0125] Table 3 Data records of collision warnings issued at different delays in Test Scenario 1

[0126]

[0127] Table 4 Data records of collision warnings issued at different delays in Test Scenario 2

[0128]

[0129] Data record of collision warning issued at different time delays in Test Scenario 3 in Table 5

[0130]

[0131] Data record of collision warning issued at a time delay of 250 - 300 ms in Test Scenario 1 in Table 6

[0132]

[0133] Data record of collision warning issued at a time delay of 150 - 200 ms in Test Scenario 2 in Table 7

[0134]

[0135] Data record of collision warning issued at a time delay of 50 - 100 ms in Test Scenario 3 in Table 8

[0136]

[0137] In the case of packet loss testing, it can be understood that the packet loss model is a mathematical model used to describe the probability of packet loss during data transmission. In actual network transmission, there can be many reasons for packet loss, such as link failures, network congestion, etc. Therefore, the packet loss rate will also change due to the influence of these factors.

[0138] The embodiments of the present invention use the Markov Monte Carlo method to simulate the data transmission process and obtain the data transmission success rate under different packet loss rates. The specific steps are as follows: Set the total number of data packets n and the packet loss probability p; Generate n random numbers between 0 and 1; For the i-th random number, if it is less than p, it is considered that the data packet is discarded, otherwise it is considered that the data packet is successfully transmitted. Count the number of successfully transmitted data packets k, and calculate the probability P(k) of successfully transmitting k data packets.

[0139] By continuously repeating the above steps, a series of data transmission results with different packet loss rates can be obtained, and the probability of data transmission success at various packet loss rates can be obtained through analysis. The advantage of the Monte Carlo method is that it can simulate extremely complex systems without the need to know the exact equations or models of the system in advance. In actual situations, there is a certain relationship between the results of the Monte Carlo method and the actual packet loss rate. By continuously adjusting the packet loss rate p, the data transmission success rate under different packet loss rates can be obtained. If in an actual network, the packet loss rate is close to the simulated packet loss rate, then the data transmission success rate obtained by the Monte Carlo method will be relatively close.

[0140] For the packet loss test, a packet loss module was built in Simulink, and different packet loss rates (i.e., the key parameters to be measured) were set, which were set to 0, 5%, 15%, 25%, 35%, 45%, and 55% respectively for the detection experiment. The individual resolution of the camera (i.e., the non-key parameter to be measured in this test) was set to 1280×960. For comparative analysis, different sampling frame rates were also set for the simulation experiment. The simulation frequencies of the system and the camera were set to 25 FPS, 30 FPS, 40 FPS, and 60 FPS respectively. The response of the warning signal under different packet losses is shown in Table 9-14:

[0141] Table 9 Data records of collision warnings issued at different packet loss rates in Test Scenario 1

[0142]

[0143] Table 10 Data records of collision warnings issued at different packet loss rates in Test Scenario 2

[0144]

[0145] Table 11 Data records of collision warnings issued at different packet loss rates in Test Scenario 3

[0146]

[0147] Table 12 Data records of collision warnings issued at a packet loss rate of 15-25% in Test Scenario 1

[0148]

[0149] Table 13 Data records of collision warnings issued at a packet loss rate of 15-25% in Test Scenario 2

[0150]

[0151] Table 14 Data records of collision warnings issued at a packet loss rate of 5-15% in Test Scenario 3

[0152]

[0153] In the case of the resolution test, it can be understood that the FCW system detects the position of the vehicle in front through the camera and calculates the vehicle distance. Therefore, the camera resolution will affect the warning performance of the system. With the same control delay (i.e., the non-key parameter to be measured in this test), different resolutions (i.e., the key parameters to be measured) were set for the simulation experiment. For comparative analysis, different sampling frame rates were also set for the simulation experiment. The frame rate of the camera was fixed at 25 FPS, 30 FPS, 40 FPS, and 60 FPS. The time when the vehicle's collision warning was issued at different resolutions was recorded in Tables 15-17.

[0154] Data records of collision warnings issued at different resolutions and sampling frame rates in Test Scenario 1, Table 15

[0155]

[0156] Data records of collision warnings issued at different resolutions and sampling frame rates in Test Scenario 2, Table 16

[0157]

[0158]

[0159] Data records of collision warnings issued at different resolutions and sampling frame rates in Test Scenario 3, Table 17

[0160]

[0161] In some embodiments, when the test category is the perception accuracy test and the key parameter to be measured is the resolution, based on the simulated forward collision warning system and the simulation scenario, forward collision warning simulation is performed according to the preset value until the preset end condition of the simulation scenario is reached, and the data related to the key parameters of the host test vehicle in the current driving state is obtained. Specifically, it includes:

[0162] Add a preset sensor to the simulation scenario; wherein, the preset sensor is used to detect the actual relative distance between the host test vehicle and the obstacle ahead.

[0163] Based on the simulated forward collision warning system and the simulation scenario, set the sampling frame rate to a fixed value, and perform forward collision warning simulation at different preset resolution values, and obtain the detection distance and the actual relative distance at each resolution as the data related to the key parameters; wherein, the detection distance is the real-time test distance output by the simulated forward collision warning system.

[0164] Specifically, when the test category is the perception accuracy test, the key parameter to be measured is the resolution. At this time, the test of the perception output accuracy can be carried out, that is, the test and analysis are carried out with the output accuracy as the index. It can be understood that the accuracy of detecting the distance between the host vehicle and the preceding vehicle based on the camera in the system will directly affect the output accuracy of the system, and this part can be called the perception output accuracy test. During the implementation process, a sensor is added to the simulation scenario, and the sensor can use an ideal sensor to calculate the actual relative distance between the preceding vehicle and the host vehicle. It can be understood that there is a certain deviation between the real-time test distance (measured relative distance) obtained by the simulated forward collision warning system and the actual relative distance, and this deviation will directly affect the braking effect of the automatic emergency braking system.

[0165] Set the sampling frame rate of the camera to 30 FPS and perform simulations at different resolutions ((1280×960, 1600×1200, 2048×1536, 2560×1920, 3840×2160)) to obtain real-time data of the actual relative distance and the measured relative distance at different resolutions.

[0166] Step 130: Analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0167] Further, when the test category is a quantization test, use TTC as the measurement standard, draw a three-dimensional analysis diagram of the key parameter to be measured and TTC, and obtain the safety threshold boundary according to the three-dimensional analysis diagram.

[0168] In the latency test, based on the above embodiment, the three-dimensional analysis diagram drawn in combination with the data in Table 3-8 is as Figure 6 shown. From the data analysis in Table 3-5, it can be seen that for the forward collision avoidance system, the latency and frame rate of the camera are very important factors, which directly affect the warning and response capabilities of the system. The following conclusions can be drawn:

[0169] The camera latency cannot be higher than 150 ms, which means that the image captured by the camera needs to be transmitted to the processor for analysis within 150 ms, otherwise it may cause a problem of excessive latency, resulting in a decline in the warning and response capabilities of the forward collision avoidance system.

[0170] The frequency of the camera cannot be lower than 30 FPS, which means that the camera needs to capture images in the scene at a rate of 40 frames per second, otherwise it may cause problems such as image stuttering or frame loss, resulting in a decline in the warning and response capabilities of the forward collision avoidance system.

[0171] Table 6-8 also gives an analysis of the specific boundary information. Among them, when the vehicle in front is in a stationary state, the most serious situation of latency on TTC mainly occurs between 270 ms - 280 ms. For test scenario 2, when the vehicle in front maintains a certain form speed, the interference of latency on FCW mainly occurs between 160 - 170 ms. For scenario 3, when the two vehicles are relatively close and have similar speeds, the latency should be controlled below 60 ms.

[0172] In the packet loss test, based on the above embodiment, the three-dimensional analysis diagram drawn in combination with the data in Table 9-14 is as Figure 7 shown. From the data in Table 9-11, it can be analyzed that by adjusting the packet loss rate and the camera frequency, the response speed of the vision-based forward collision warning system can be improved. Specifically, the following conclusions are obtained:

[0173] In the vision-based forward collision warning system, the TTC response speed is greatly affected by the packet loss rate. The larger the packet loss rate, the slower the response speed of the warning system.

[0174] The higher the frame rate of the camera, the more the impact of the packet loss rate can be slightly alleviated, that is, the response speed of the warning system can be increased.

[0175] It is recommended to control the packet loss rate below 10%-15%, and at the same time, the frequency of the camera should be above 40 FPS.

[0176] Table 12-14 also makes a detailed analysis of the warning response of FCW with respect to the packet loss rate in specific scenarios. Specifically, when the distance between two vehicles is far and the speed difference is large, the concentrated impact range of the packet loss rate on FCW is 19-20%. For the situation of being closer to the vehicle in front, the packet loss rate should be controlled below 10%, and at this time, a high frame rate of the camera should be maintained. For complex scenarios, it is recommended to increase the camera frame rate and keep the network unobstructed.

[0177] In the resolution test, based on the above embodiments, the three-dimensional analysis diagram drawn in combination with the data in Table 15-18 is as Figure 8 shown. According to the data in Table 15-17, and Figure 8 analysis shows that using a high-definition camera and increasing the frame rate play a very important role in improving the performance of the forward collision warning system, but there are also certain side effects. Specifically: Higher resolution and sampling frame rate can improve the real-time performance and response speed of the system. However, this will also increase the computational load and storage / transmission cost of the system. Therefore, when selecting a high-resolution camera, the sampling frame rate should be strictly controlled. In addition, during the process of increasing the resolution from 1280×960 to 3840×2160, the overall system performance has increased by 15%-20%. In actual working conditions, the sampling frame rate of the camera needs to balance the relationship between system performance and hardware cost. Higher sampling frame rate and resolution can provide more accurate images, which helps to improve the performance and safety of autonomous driving, but it will also increase the computational burden and hardware cost of the system. Here are specific suggestions: Under relatively stable road conditions, it is recommended to use a resolution of 720p or 1080p and appropriately reduce the sampling frame rate to 30fps to reduce the hardware cost and computational burden.

[0178] Under relatively complex working conditions, it is recommended to use a resolution of 1080p or higher and keep the sampling frame rate at a relatively high level, such as 60 FPS, to ensure that the system can accurately identify and respond to traffic conditions in a timely manner.

[0179] Furthermore, when the test category is the perception accuracy test, calculate the absolute error value of the actual relative distance and the measured relative distance and the minimum relative distance data. Based on the above embodiments, the obtained absolute error value and minimum relative distance data are shown in Table 18.

[0180] Table 18 Analysis Table of Perception Output Accuracy Test

[0181] Resolution Mean of Absolute Error Variance of Absolute Error Minimum Relative Distance 1280×960 1.1281 1.3565 3.18 1600×1200 1.1598 1.3270 2.72 2048×1536 1.0754 1.1242 2.63 2560×1920 0.5781 1.0632 2.41 3840×2160 0.4739 0.9655 1.86

[0182] As can be seen from Table 18, at different resolutions, the greater the mean and variance of the absolute error, the shorter the minimum relative distance, and the easier it is to have a collision. The FCW system can achieve early warning well at these five resolutions. Since the minimum resolution of the commonly used camera in the actual operation process is 1280×960, the cameras used can basically meet the requirements for safe operation.

[0183] In addition, compared with the ideal sensor AIR, there is a certain deviation in the visual ranging algorithm designed in the embodiment of the present invention, but its influence can be ignored, and the system can meet our test objectives.

[0184] Further, after step 130, it further includes: outputting an analysis report.

[0185] Based on the above embodiments, the analysis report includes the influence of data delay, packet loss rate, and camera resolution on FCW, as well as the specific situation and suggestions of the data in each scenario. For example: from the perspective of delay, it is recommended to control the frame rate of the target detection algorithm on the computing platform at 30 - 40 FPS in the actual project, and the delay should be less than 150 ms to maintain the stability of the system. At the same time, this also puts certain requirements on the computing power platform. In addition, from the perspective of packet loss rate, considering the actual situation of road driving, the situation where the distance from the vehicle in front is too low and the speeds are quite the same occurs from time to time. From the perspective of safety, the packet loss rate of the system should be controlled below 10 - 20%. Finally, the improvement effect of the camera resolution on the warning function is analyzed. Generally speaking, the improvement of the camera resolution can improve the system performance by about 15% - 20%, but the resources need to be reasonably configured under specific working conditions. Although high resolution and high sampling rate can improve the system performance, they also have higher requirements for hardware computing power.

[0186] The key parameter testing method for a vision-based forward collision warning system provided by the present invention simulates the forward collision warning system by using a preset simulation algorithm and constructs a simulation scenario; wherein, the simulation scenario at least includes a camera, an obstacle ahead, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approach of the obstacle ahead and the self-test vehicle; select the key parameter to be tested and set different preset values for the key parameter to be tested, and based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtain the data related to the key parameters of the self-test vehicle in the current driving state; wherein, the key parameter to be tested includes at least one of the delay rate, the packet loss rate over time, and the resolution; analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters. The present invention tests the key parameters for the safe operation of the forward collision warning system through a simulation algorithm, and calculates the safety threshold boundary of the key parameters through scenario simulation, so as to contribute to a forward collision warning system with higher interpretability.

[0187] The key parameter testing device for a vision-based forward collision warning system provided by the present invention will be described below. The key parameter testing device for a vision-based forward collision warning system described below can be correspondingly referred to the key parameter testing method for a vision-based forward collision warning system described above. As Figure 9 shown, the device includes:

[0188] A simulation unit 910, configured to simulate a forward collision warning system by using a preset simulation algorithm and construct a simulation scenario; wherein, the simulation scenario at least includes a camera, an obstacle ahead, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approach of the obstacle ahead and the self-test vehicle;

[0189] A testing unit 920, configured to select the key parameter to be tested and set different preset values for the key parameter to be tested, and based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtain the data related to the key parameters of the self-test vehicle in the current driving state; wherein, the key parameter to be tested includes at least one of the delay rate, the packet loss rate over time, and the resolution;

[0190] An analysis unit 930, configured to analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0191] A key parameter testing device for a vision-based forward collision warning system provided by the present invention, for testing the selected key parameters to be tested, specifically including:

[0192] Select the corresponding key parameters to be tested according to the test category - key parameter mapping relationship to be tested through the pre-selected test category.

[0193] A key parameter testing device for a vision-based forward collision warning system provided by the present invention, when the test category is a quantization test and the key parameter to be tested is one of delay, packet loss rate over time, and resolution, based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset value until the preset end condition of the simulation scenario is reached, and obtain the key parameter-related data of the self-test vehicle in the current driving state, specifically including:

[0194] Based on the simulated forward collision warning system and the simulation scenario, set the non-key parameters to be tested to fixed values, and perform forward collision warning simulation under different preset values of the key parameter to be tested, and obtain the collision time under each preset value as the key parameter-related data;

[0195] Wherein, the collision time is the time interval between the self-test vehicle and the obstacle in front when the warning is issued.

[0196] A key parameter testing device for a vision-based forward collision warning system provided by the present invention, when the test category is a perception accuracy test and the key parameter to be tested is resolution, based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset value until the preset end condition of the simulation scenario is reached, and obtain the key parameter-related data of the self-test vehicle in the current driving state, specifically including:

[0197] Add a preset sensor to the simulation scenario; wherein, the preset sensor is used to detect the actual relative distance between the self-test vehicle and the obstacle in front.

[0198] Based on the simulated forward collision warning system and the simulation scenario, set the sampling frame rate to a fixed value, and perform forward collision warning simulation under different resolution preset values, and obtain the detection distance and the actual relative distance under each resolution as the key parameter-related data; wherein, the detection distance is the real-time test distance output by the simulated forward collision warning system.

[0199] A key parameter testing device for a vision-based forward collision warning system provided by the present invention, the test scenario specifically includes:

[0200] Longitudinal static front obstacle recognition test scenario, longitudinal slow front obstacle recognition test scenario, and longitudinal deceleration front obstacle recognition test scenario.

[0201] According to a key parameter test device for a vision-based forward collision warning system provided by the present invention, the forward collision warning system includes a road video module, an object detection module, a ranging module, a warning module, and a warning information post-processing module;

[0202] The road video module is used to transmit the real-time image obtained by the camera to the object detection module;

[0203] The object detection module is used to obtain the RGB value of the real-time image according to the real-time image, perform object detection according to the RGB value of the real-time image, and obtain the upper and lower limits of the horizontal and vertical coordinates of the front obstacle in the real-time image;

[0204] The ranging module is used to obtain the real-time test distance from the front obstacle to the camera according to the height, downward deviation angle, and focal length of the camera, and the upper and lower limits of the horizontal and vertical coordinates;

[0205] The warning module is used to calculate the time to collision according to the real-time test distance and the relative speed of the front obstacle and the self-vehicle test vehicle in the current driving state; compare the time to collision with a preset safety threshold to obtain collision information, and perform collision warning according to the collision information;

[0206] The warning information post-processing module is used to perform visualization processing on the real-time test distance and the upper and lower limits of the horizontal and vertical coordinates, and output them to an oscilloscope or a video port for real-time observation.

[0207] The key parameter testing device for the vision-based forward collision warning system provided by the present invention simulates the forward collision warning system by using a preset simulation algorithm and constructs a simulation scenario; wherein, the simulation scenario at least includes a camera, an obstacle ahead, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approach of the obstacle ahead and the self-test vehicle; select the key parameters to be tested and set different preset values for the key parameters to be tested. Based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtain the data related to the key parameters of the self-test vehicle in the current driving state; wherein, the key parameters to be tested include at least one of the delay rate, the packet loss rate over time, and the resolution; analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters. The present invention tests the key parameters for the safe operation of the forward collision warning system through a simulation algorithm, and calculates the safety threshold boundary of the key parameters through scenario simulation, thereby contributing to a forward collision warning system with higher interpretability.

[0208] Figure 10 An example of the physical structure diagram of an electronic device is as Figure 10 shown. The electronic device may include: a processor 1010, a communication interface 1020, a memory 1030, and a communication bus 1040. Among them, the processor 1010, the communication interface 1020, and the memory 1030 communicate with each other through the communication bus 1040. The processor 1010 can call the logical instructions in the memory 1030 to execute the method for testing the key parameters of the vision-based forward collision warning system. The method includes: using a preset simulation algorithm to simulate the forward collision warning system and construct a simulation scenario; wherein, the simulation scenario at least includes a camera, an obstacle ahead, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approach of the obstacle ahead and the self-test vehicle; select the key parameters to be tested and set different preset values for the key parameters to be tested. Based on the simulated forward collision warning system and the simulation scenario, perform forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtain the data related to the key parameters of the self-test vehicle in the current driving state; wherein, the key parameters to be tested include at least one of the delay rate, the packet loss rate over time, and the resolution; analyze the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0209] In addition, when the logical instructions in the above-mentioned memory 1030 are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs.

[0210] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the key parameter testing method for the vision-based forward collision warning system provided by the above-mentioned various methods. The method includes: using a preset simulation algorithm to simulate the forward collision warning system and building a simulation scenario; wherein, the simulation scenario at least includes a camera, a forward obstacle, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approaching manner of the forward obstacle and the self-test vehicle; selecting key parameters to be tested and setting different preset values for the key parameters to be tested. Based on the simulated forward collision warning system and the simulation scenario, forward collision warning simulation is performed according to the preset values until the preset end condition of the simulation scenario is reached, and the data related to the key parameters of the self-test vehicle in the current driving state is obtained; wherein, the key parameters to be tested include at least one of a delay rate, a time packet loss rate, and a resolution; analyzing the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0211] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the key parameter testing method of the vision-based forward collision warning system provided by the above-mentioned various methods. The method includes: simulating the forward collision warning system by using a preset simulation algorithm and building a simulation scenario; wherein, the simulation scenario at least includes a camera, a front obstacle, a self-test vehicle, and a test scenario; wherein, the camera is installed on the self-test vehicle, and the test scenario includes the approaching manner of the front obstacle and the self-test vehicle; selecting key parameters to be tested and setting different preset values for the key parameters to be tested, and based on the simulated forward collision warning system and the simulation scenario, performing forward collision warning simulation according to the preset values until the preset end condition of the simulation scenario is reached, and obtaining data related to the key parameters of the self-test vehicle in the current driving state; wherein, the key parameters to be tested include at least one of a delay rate, a time packet loss rate, and a resolution; analyzing the data related to the key parameters to obtain the safety threshold boundary of the key parameters.

[0212] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.

[0213] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0214] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A vision-based forward collision warning system key parameter testing method, characterized in that: include: A forward collision avoidance warning system is simulated using a preset simulation algorithm and a simulation scene is constructed; wherein the simulation scene at least includes a camera, a front obstacle, an ego test vehicle, and a test scene; wherein the camera is installed on the ego test vehicle, and the test scene includes the approaching manner of the front obstacle and the ego test vehicle; Select key parameters to be tested and set different preset values ​​for the key parameters to be tested, perform forward collision avoidance warning simulation according to the preset values ​​based on the simulated forward collision avoidance warning system and the simulation scenario, until the preset end condition of the simulation scenario is reached, and obtain key parameter-related data of the self-vehicle test vehicle in the current driving state; wherein the key parameters to be tested include at least one of delay rate, time packet loss rate and resolution; Analyze the key parameter related data to obtain the safety threshold boundary of the key parameter; The selecting the key parameters to be tested for testing specifically includes: selecting the corresponding key parameters to be tested for testing according to the pre-selected test category through the test category-key parameter to be tested mapping relationship; In the case where the test category is a quantitative test and the key parameter to be tested is one of latency, time packet loss rate and resolution, the forward collision avoidance warning simulation is performed based on the simulated forward collision avoidance warning system and the simulation scenario according to the preset value until the preset end condition of the simulation scenario is reached, and the key parameter-related data of the self-vehicle test vehicle in the current driving state are obtained, specifically including: Based on the simulated forward collision avoidance warning system and the simulation scenario, non-key parameters to be measured are set to fixed values, and forward collision avoidance warning simulation is performed under different preset values ​​of the key parameters to be measured, and collision time under each preset value is obtained as the key parameter related data; Wherein, the collision time is the time interval between the collision of the self-vehicle test vehicle and the front obstacle when the warning is issued; In the case where the test category is a perception accuracy test and the key parameter to be tested is resolution, the forward collision avoidance warning simulation is performed based on the simulated forward collision avoidance warning system and the simulation scenario according to the preset value until the preset end condition of the simulation scenario is reached, and the key parameter-related data of the self-vehicle test vehicle in the current driving state are obtained, specifically including: Adding a preset sensor in the simulation scene; wherein the preset sensor is used to detect the actual relative distance between the self-vehicle test vehicle and the front obstacle; Based on the simulated forward collision avoidance warning system and the simulation scenario, the sampling frame rate is set to a fixed value, and forward collision avoidance warning simulation is performed at different resolution preset values, and the detection distance and actual relative distance at each resolution are obtained as key parameter related data; wherein the detection distance is the real-time test distance output by the simulated forward collision avoidance warning system.

2. The method for testing key parameters of a forward collision avoidance warning system based on vision according to claim 1, characterized in that: The test scenarios specifically include: Longitudinal stationary front obstacle recognition test scenario, longitudinal slow-moving front obstacle recognition test scenario and longitudinal decelerating front obstacle recognition test scenario.

3. The method for testing key parameters of a forward collision avoidance warning system based on vision according to claim 1, characterized in that: The forward collision avoidance warning system includes a road video module, a target detection module, a distance measurement module, a warning module and a warning information post-processing module; The road video module is used to transmit the real-time image acquired by the camera to the target detection module; The target detection module is used to obtain a real-time image RGB value according to the real-time image, perform target detection according to the real-time image RGB value, and obtain the upper and lower limits of the horizontal and vertical coordinates of the front obstacle in the real-time image; The distance measurement module is used to obtain the real-time test distance from the front obstacle to the camera according to the height, downward deflection angle and focal length of the camera, as well as the upper and lower limits of the horizontal and vertical coordinates; The early warning module is used to calculate the collision time according to the real-time test distance and the relative speed of the front obstacle and the self-vehicle test vehicle in the current driving state; Comparing the collision time with a preset safety threshold, obtaining collision information, and performing a collision warning according to the collision information; The warning information post-processing module is used to visualize the real-time test distance and the upper and lower limits of the horizontal and vertical coordinates, and output them to an oscilloscope or a video port for real-time observation.

4. A vision-based forward collision warning system key parameter test device, characterized in that: include: A simulation unit, used to simulate a forward collision warning system using a preset simulation algorithm and build a simulation scene; wherein the simulation scene at least includes a camera, a front obstacle, an ego test vehicle, and a test scene; wherein the camera is installed on the ego test vehicle, and the test scene includes the approaching manner of the front obstacle and the ego test vehicle; A test unit, used for selecting key parameters to be tested for testing and setting different preset values ​​for the key parameters to be tested, performing forward collision avoidance warning simulation according to the preset values ​​based on the simulated forward collision avoidance warning system and the simulation scenario, until a preset end condition of the simulation scenario is reached, and obtaining key parameter-related data of the self-vehicle test vehicle in the current driving state; wherein the key parameters to be tested include at least one of a delay rate, a time packet loss rate, and a resolution; An analysis unit, used to analyze the key parameter related data to obtain the safety threshold boundary of the key parameter; The selecting the key parameters to be tested for testing specifically includes: selecting the corresponding key parameters to be tested for testing according to the pre-selected test category through the test category-key parameter to be tested mapping relationship; In the case where the test category is a quantitative test and the key parameter to be tested is one of latency, time packet loss rate and resolution, the forward collision avoidance warning simulation is performed based on the simulated forward collision avoidance warning system and the simulation scenario according to the preset value until the preset end condition of the simulation scenario is reached, and the key parameter-related data of the self-vehicle test vehicle in the current driving state are obtained, specifically including: Based on the simulated forward collision avoidance warning system and the simulation scenario, non-key parameters to be measured are set to fixed values, and forward collision avoidance warning simulation is performed under different preset values ​​of the key parameters to be measured, and collision time under each preset value is obtained as the key parameter related data; Wherein, the collision time is the time interval between the collision of the self-vehicle test vehicle and the front obstacle when the warning is issued; In the case where the test category is a perception accuracy test and the key parameter to be tested is resolution, the forward collision avoidance warning simulation is performed based on the simulated forward collision avoidance warning system and the simulation scenario according to the preset value until the preset end condition of the simulation scenario is reached, and the key parameter-related data of the self-vehicle test vehicle in the current driving state are obtained, specifically including: Adding a preset sensor in the simulation scene; wherein the preset sensor is used to detect the actual relative distance between the self-vehicle test vehicle and the front obstacle; Based on the simulated forward collision avoidance warning system and the simulation scenario, the sampling frame rate is set to a fixed value, and forward collision avoidance warning simulation is performed at different resolution preset values, and the detection distance and actual relative distance at each resolution are obtained as key parameter related data; wherein the detection distance is the real-time test distance output by the simulated forward collision avoidance warning system.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the key parameter testing method of the forward collision avoidance warning system based on vision as described in any one of claims 1 to 3 is implemented.

6. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for testing key parameters of a forward collision avoidance warning system based on vision as claimed in any one of claims 1 to 3 is implemented.

7. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for testing key parameters of a forward collision avoidance warning system based on vision as claimed in any one of claims 1 to 3 is implemented.

Images