A tunnel target detection and longitudinal control method

Abstract

This invention provides a method for tunnel target detection and longitudinal control, belonging to the field of intelligent connected vehicles. This method utilizes illumination information, vehicle information, and traffic sign attribute information acquired by a camera to detect tunnel targets and perform longitudinal control for both tunnel entry and exit scenarios with and without a target. This method utilizes a purely visual approach, replacing the conventional camera-and-radar solution, to achieve driver assistance while reducing vehicle costs. It also enhances driver comfort when using adaptive cruise control during tunnel entry and exit, and strengthens the driver's sense of safety during braking.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent networking, and specifically relates to a tunnel target detection and longitudinal control method. Background Art

[0002] At present, models equipped with driving assistance radar + camera can realize head ACC adaptive cruise control, can maintain a steady speed and follow the vehicle stably, follow the speed of the vehicle in front adaptively, and realize full-speed follow-up, stop, start and go functions.

[0003] Vehicles equipped with pure vision solutions (including monocular, binocular or multi-cameras) can also achieve ACC adaptive cruise control in the full cruising speed range. However, in rainy days and scenes with large changes in lighting, pure vision solutions have inherent deficiencies in target detection and control compared to radar. Specifically, when entering and exiting a tunnel with large changes in light intensity, the camera will be temporarily blinded due to strong exposure at the moment of exiting the tunnel, or at the moment of entering a tunnel, it may mistakenly think that there is a target in the tunnel ahead and the vehicle will decelerate inadvertently.

[0004] Compared with the ACC function that relies on driving assistance radar, the ACC function of the pure vision solution is not as accurate as the radar in judging the longitudinal distance, longitudinal speed and acceleration of the vehicle in front when entering and exiting tunnel scenes. In addition, the data fluctuates, resulting in inaccurate steady-state control of following the vehicle and frequent switching between acceleration and deceleration, which makes the driver feel uncomfortable and unsafe when following the vehicle. Summary of the Invention

[0005] The purpose of the present invention is to solve the difficulties existing in the above-mentioned prior art and provide a tunnel target detection and longitudinal control method to solve the problem that when a vehicle equipped with a pure vision solution enters and exits a tunnel, the ACC function does not accurately judge the longitudinal distance, longitudinal speed and acceleration of the vehicle in front as compared to the radar, and the data fluctuates, resulting in inaccurate steady-state control of the following vehicle and frequent switching of acceleration and deceleration; and solve the problem that when a vehicle equipped with a pure vision solution enters and exits a tunnel, the camera's perception has an inherent disadvantage. At the moment of entering and exiting the tunnel, the camera will be temporarily blinded due to strong exposure, or the vehicle will mistakenly decelerate because it thinks there is a target in the tunnel ahead.

[0006] The present invention is achieved through the following technical solutions:

[0007] The present invention provides a tunnel target detection and longitudinal control method, which uses identification illumination information, vehicle information, and traffic sign attribute information obtained by a camera to detect tunnel targets, and performs longitudinal control in scenarios where there is no target being followed and where there is a target being followed.

[0008] A further improvement of the present invention is:

[0009] The method comprises:

[0010] Step 1: Process the image captured by the camera to obtain identification illumination information, vehicle information, and traffic sign attribute information;

[0011] Step 2: Determine whether there is vehicle information. If not, determine that there is no target to follow, and perform longitudinal control for entering and exiting the tunnel without a target to follow. If yes, determine that there is a target to follow, and perform longitudinal control for entering and exiting the tunnel with a target to follow.

[0012] A further improvement of the present invention is:

[0013] The operation of step 1 includes:

[0014] (11) Input the image captured by the camera into the neural network processing model to obtain simplified intermediate image information;

[0015] (12), the intermediate image information is input into the segmentation model to obtain illumination change information, vehicle information and traffic sign attribute information.

[0016] A further improvement of the present invention is:

[0017] The operation of performing longitudinal control in the tunnel entry and exit scenario without a follow-up target in step 2 includes:

[0018] If the traffic sign attribute information contains tunnel and speed limit signs, the system determines whether the speed limit sign value is lower than the vehicle's cruising speed. If so, the system changes the cruising speed to the speed limit sign value and slowly reduces the speed. At the same time, a voice announcement reminds the user of the reason for the vehicle's deceleration. If not, the vehicle maintains its current cruising speed and automatically filters out the vehicle information generated by the exposure, without accelerating or decelerating before entering a tunnel.

[0019] If the traffic sign attribute information only contains tunnel, the vehicle maintains its current cruising speed, automatically filters out the vehicle information generated by the exposure, and does not accelerate or decelerate before entering the tunnel.

[0020] If the traffic sign attribute information only contains a speed limit sign, the camera will determine the situation ahead based on the illumination change information. If the speed limit sign is lower than the vehicle's cruising speed, the camera will determine whether the speed limit sign is lower than the vehicle's cruising speed. If so, the vehicle will change its cruising speed to the speed limit sign and slowly reduce its speed. Meanwhile, a voice announcement will be made on the vehicle to remind the user of the reason for the reduction. If not, the vehicle will maintain its current cruising speed and automatically filter out the vehicle information generated by the exposure. The vehicle will not accelerate or decelerate before entering the tunnel.

[0021] If there is no traffic sign attribute information, the camera will judge the situation ahead based on the illumination change information. If there is a tunnel entrance ahead, the vehicle will maintain the current cruising speed, automatically filter the vehicle information generated by the exposure, and will not accelerate or decelerate before entering the tunnel. If there is a tunnel exit ahead, the vehicle will maintain the current cruising speed, automatically filter the vehicle information generated by the exposure, and will not accelerate or decelerate before exiting the tunnel.

[0022] A further improvement of the present invention is:

[0023] The longitudinal control operation in step 2 for following the target into and out of the tunnel includes:

[0024] If the traffic sign attribute information contains tunnel and speed limit signs, it is determined whether the speed limit sign value is lower than the cruise speed of the ego vehicle. If so, the cruise speed is changed to the speed limit sign value, and the vehicle is smoothly decelerated to a steady-state following distance. At the same time, a voice broadcast is made on the vehicle to remind the user of the reason for the deceleration of the ego vehicle. Before the ego vehicle enters the tunnel, the vehicle information of the preceding vehicle before entering the tunnel is maintained, and the vehicle information generated by exposure is automatically filtered. The ego vehicle does not accelerate or decelerate. If not, the ego vehicle maintains the current cruise speed, and smoothly decelerates to a steady-state following distance. The vehicle information generated by exposure is automatically filtered, and the ego vehicle does not accelerate or decelerate before entering the tunnel.

[0025] If the traffic sign attribute information only contains tunnels, the vehicle maintains its current cruising speed and smoothly decelerates to a steady-state following distance. The vehicle information generated by the exposure is automatically filtered, and no acceleration or deceleration is performed before entering the tunnel.

[0026] If there is only a speed limit sign in the traffic sign attribute information, the camera will judge the situation ahead based on the illumination change information. If it is a tunnel entrance, it will determine whether the speed limit sign value is lower than the vehicle's cruising speed. If so, the cruising speed will be changed to the speed limit sign value, and the vehicle will be smoothly decelerated to a steady-state following distance. At the same time, a voice broadcast will be made on the vehicle to remind the user of the reason for the vehicle's deceleration. Before the vehicle enters the tunnel, the vehicle information of the preceding vehicle before entering the tunnel will be maintained, and the vehicle information generated by exposure will be automatically filtered. The vehicle will not accelerate or decelerate. If not, the vehicle will maintain the current cruising speed, and smoothly decelerate to a steady-state following distance, and the vehicle information generated by exposure will be automatically filtered. The vehicle will not accelerate or decelerate before entering the tunnel.

[0027] If there is no traffic sign attribute information, the camera will judge the situation ahead based on the illumination change information. If there is a tunnel entrance ahead, the vehicle will maintain its current cruising speed, smoothly decelerate to a steady-state following distance, and automatically filter the vehicle information generated by exposure. It will not accelerate or decelerate before entering the tunnel. If there is a tunnel exit ahead, the vehicle will maintain its current cruising speed, smoothly decelerate to a steady-state following distance, and automatically filter the vehicle information generated by exposure. It will not accelerate or decelerate before exiting the tunnel.

[0028] Preferably, the steady-state following distance refers to the distance obtained by adding a 0.4s interval to the current interval.

[0029] Preferably, the operation of the camera judging the situation ahead according to the illumination change information includes:

[0030] If the illumination changes from 60,000-80,000 lux to 1,000-2,000 lux, it is determined that the front is a tunnel entrance;

[0031] If the illumination change information remains between 60,000 and 80,000 lux, it is determined that the front is not a tunnel entrance;

[0032] If the illumination change information changes from 30-1500 lux to 40,000-50,000 lux, or from 1-3 lux to 40,000-50,000 lux, it is determined that the front is a tunnel exit.

[0033] A further improvement of the present invention is:

[0034] The operation of automatically filtering the vehicle information generated by exposure includes: determining whether the vehicle information is vehicle information generated by exposure, and if so, eliminating the vehicle information generated by exposure.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention adopts a pure vision solution, replacing the original camera plus radar solution, realizing the driving assistance function, while reducing the cost of the entire vehicle, improving the driver's comfort when using adaptive cruise control when entering and exiting tunnels, and enhancing the sense of braking safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 There is no flowchart for following the target into the tunnel;

[0038] Figure 2 There is no flowchart for following the target out of the tunnel;

[0039] Figure 3 A diagram of the steps involved in following a target into a tunnel;

[0040] Figure 4 A diagram of the steps to follow the target out of the tunnel;

[0041] Figure 5 Real road pictures;

[0042] Figure 6 Road information;

[0043] Figure 7 Image segmentation of the rear of the target vehicle in front;

[0044] Figure 8 A block diagram of the steps of the method of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below with reference to the accompanying drawings:

[0046] In order to solve the problem that "when a vehicle equipped with a pure vision solution enters and exits a tunnel, the ACC function's judgment of the longitudinal distance, longitudinal speed and acceleration of the vehicle in front is not as accurate as that of the radar, and the data fluctuates, resulting in inaccurate steady-state control of the vehicle following and frequent switching of acceleration and deceleration", and the problem that "when a vehicle equipped with a pure vision solution enters and exits a tunnel, the camera's perception has an inherent disadvantage. At the moment of entering and exiting the tunnel, the camera may cause temporary blindness due to strong exposure, or may mistakenly think that there is a target in the tunnel ahead and cause erroneous deceleration", the present invention analyzes and solves two scenarios respectively. The first scenario is the scenario of entering and exiting the tunnel without following a target, and there is a problem of false braking; the second scenario is the scenario of following the vehicle in front in and out of the tunnel, and there is switching of acceleration and deceleration.

[0047] Currently, cameras installed on vehicles periodically output information about the target vehicle ahead. This information, which refers to the perceived attributes of the target vehicle, includes its acceleration, speed, and distance from the vehicle itself. Existing cameras can identify vehicles (including cars, trucks, and specialized vehicles), two-wheeled vehicles, and three-wheeled vehicles, and output this information to the instrument cluster and the ACC adaptive cruise control system, allowing the vehicle to adaptively follow these vehicles.

[0048] When entering a tunnel, if there is no vehicle information, it can be determined that the vehicle is entering the tunnel without a target. Under normal circumstances, the camera will not recognize the image of the tunnel entrance as the target vehicle of ACC. However, due to the drastic changes in light at the tunnel entrance, the camera may be exposed, which in turn causes image quality problems. At this time, the camera will mistakenly think that the shadow position at the tunnel entrance is a target vehicle and assign vehicle information. At this time, the ACC adaptive cruise system of the vehicle will slow down according to the vehicle information. However, since there is no target vehicle at this time, the vehicle information is inaccurate and should not be decelerated. That is, the vehicle information generated by the exposure should be eliminated and the current driving state should be maintained.

[0049] When entering a tunnel, if there is vehicle information, it can be determined that the vehicle is following a target into the tunnel. At the moment of entering and exiting the tunnel, the vehicle information of the target vehicle will also change due to camera exposure. At this time, the ACC adaptive cruise control system of the vehicle will slow down according to the vehicle information. However, since there is no new target vehicle at this time, the camera should not consider the target and its vehicle information generated by the exposure as credible, that is, it should not accelerate or decelerate for the target identified after exposure, but should eliminate the vehicle information generated by the exposure and maintain the current driving state.

[0050] The present invention provides a tunnel target detection and longitudinal control method, which uses identification illumination information, vehicle information, and traffic sign attribute information obtained by a camera to detect tunnel targets, and performs longitudinal control in scenarios where there is no target being followed and where there is a target being followed.

[0051] Specifically, such as Figure 8 As shown, the method of the present invention comprises:

[0052] Step 1: Obtain information based on the image captured by the camera: Process the image captured by the camera to obtain identification illumination information, vehicle information, and traffic sign attribute information.

[0053] The operations in step one include:

[0054] (11) Input the image captured by the camera into the neural network processing model, remove invalid image information, and obtain the semantic definition and correlation of the pixels in each category in the image to obtain the simplified intermediate image information. This step is processed using the existing neural network processing model, which will not be repeated here. The neural network processing model compares the image with the defined feature library and retains the features with high confidence. The simplified intermediate image information thus obtained retains the highly relevant morpheme information in the image.

[0055] (12) Input the intermediate image information into the segmentation model to obtain illumination change information, vehicle information, and traffic sign attribute information:

[0056] The intermediate image information is segmented using a segmentation model. The segmented illumination change information, vehicle information, and traffic sign attribute information are obtained according to the definition, completing the image segmentation and recognition process. The segmentation model can be an existing segmentation model. It segments the image information based on color, pixel density, and other factors, continuously cutting the image into small blocks. These blocks are then compared with predefined feature values in a feature library. The information with the highest overlap or confidence level is selected as the illumination change information, vehicle information, and traffic sign attribute information. This is all implemented using existing image processing algorithms and will not be further described here. Figure 7A is the illumination change information of the vehicle, B1, B2, and B3 are vehicle information, and C1 and C2 are traffic sign attribute information.

[0057] Figure 5 It is an image obtained by performing target cutting on the picture captured by the actual camera. Figure 5 The valid information is framed by solid lines, such as tunnels, speed limit signs, and light intensity. Figure 6 yes Figure 5 A simplified schematic diagram of Figure 7 The segmentation map of the rear image of the target vehicle in front is shown.

[0058] In addition, based on a large number of road tests, statistics show that the illumination variation information outside the tunnel is 60,000-80,000 lux, the illumination variation information at the tunnel entrance is 1,000-2,000 lux, the illumination variation information inside the tunnel when there are lights in the tunnel is 30-1,500 lux, the illumination variation information inside the tunnel when there are no lights in the tunnel is 1-3 lux, and the illumination variation information at the tunnel exit is 40,000-50,000 lux. Using these statistical data, it is possible to determine whether a vehicle is entering or exiting a tunnel.

[0059] Step 2: Determine whether there is vehicle information. If not, determine that there is no target to follow, and perform longitudinal control for entering and exiting the tunnel without a target to follow; if yes, determine that there is a target to follow, and perform longitudinal control for entering and exiting the tunnel with a target to follow.

[0060] Specifically, such as Figure 1 and Figure 2 As shown, the specific operations of the longitudinal control for entering and exiting the tunnel without a follow-up target in step 2 include:

[0061] If the traffic sign attribute information includes a tunnel and a speed limit sign, it is determined whether the speed limit sign value is lower than the cruise speed of the ego vehicle. If so, the ego vehicle's ACC system changes the cruise speed to the speed limit sign value and slowly reduces the speed at a more comfortable acceleration (for example, if the ego vehicle's current speed is the set cruise speed of 80 kph and there is a 60 kph speed limit sign at the tunnel entrance, the ACC adaptive cruise system will change the cruise speed setting from 80 to 60. Since the current speed is higher than the expected value, the ACC adaptive cruise system will send an acceleration, such as -0.2 m / s^2, to the actuator that controls deceleration, and then decelerate to the desired speed). At the same time, a voice announcement is made on the vehicle to remind the user of the reason for the deceleration. If not, the ego vehicle maintains the current cruise speed (for example, if the current cruise speed of the ego vehicle is 80 kph and the tunnel speed limit sign is 90 kph, the cruise speed of the ego vehicle will remain at 80 kph and will not be increased to 90 kph). The vehicle information generated by the exposure is automatically filtered, and no acceleration or deceleration is performed before entering the tunnel.

[0062] If the traffic sign attribute information only contains a tunnel (i.e., there is a tunnel but no speed limit sign), the vehicle maintains its current cruising speed, automatically filters the vehicle information generated by the exposure, and does not accelerate or decelerate before entering the tunnel.

[0063] If the traffic sign attribute information only contains a speed limit sign (i.e., there is a speed limit sign but no tunnel), the camera determines the situation ahead based on the illumination change information. The specific operations include:

[0064] If the illuminance changes from 60,000-80,000 lux to 1,000-2,000 lux, the vehicle determines that a tunnel entrance is ahead and determines whether the speed limit sign is lower than the vehicle's cruising speed. If so, the vehicle's ACC system changes the cruising speed to the speed limit sign and slowly decelerates at a comfortable rate. A voice announcement alerts the user to the reason for the deceleration. If not, the vehicle maintains its current cruising speed, automatically filtering out vehicle information generated by exposure, and neither accelerating nor decelerating until entering the tunnel. If the illuminance change remains between 60,000-80,000 lux, the current cruising state is maintained.

[0065] If there is no traffic sign attribute information (i.e., no speed limit sign or tunnel), the camera determines the situation ahead based on the illumination change information, as follows:

[0066] If the illumination changes from 60,000-80,000 lux to 1,000-2,000 lux, the system determines that there is a tunnel entrance in front of the vehicle, maintains the current cruising speed, automatically filters the vehicle information generated by the exposure, and does not accelerate or decelerate before entering the tunnel.

[0067] If the illumination change information remains between 60,000 and 80,000 lux, the current cruise state is maintained.

[0068] If the illumination change information changes from 30-1500 lux to 40,000-50,000 lux, or from 1-3 lux to 40,000-50,000 lux, it is determined that there is a tunnel exit in front of the vehicle, and the current cruising speed is maintained. The vehicle information generated by the exposure is automatically filtered, and no acceleration or deceleration is performed before exiting the tunnel.

[0069] Specifically, such as Figure 3 and Figure 4 As shown, the specific operations of performing longitudinal control in the scenario of following the target into and out of the tunnel in step 2 include:

[0070] If there are tunnel and speed limit signs in the traffic sign attribute information, the system determines whether the speed limit sign value is lower than the cruise speed of the vehicle. If so, the ACC system of the vehicle will change the cruise speed to the speed limit sign value and slowly reduce the speed at a more comfortable acceleration, smoothly decelerating to the current time distance ("current time distance" refers to the time distance T set by the driver to follow the vehicle in front stably). gap , T gap = the distance between the vehicle and the target vehicle / (the current speed of the vehicle - the current speed of the target vehicle) plus the steady-state following distance (T gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to the steady-state following distance in this case; when S t Reach (T gap +0.4), the acceleration a value will slowly become 0, and the ego vehicle will maintain the current distance and follow the vehicle. At the same time, the vehicle will broadcast a voice to remind the user why the ego vehicle is slowing down. Before the ego vehicle enters the tunnel, the perception information of the preceding vehicle before entering the tunnel is maintained (meaning that the vehicle information of the preceding vehicle remains unchanged), and the vehicle information generated by exposure is automatically filtered. The ego vehicle will not accelerate or decelerate. If not, the ego vehicle will maintain the current cruising speed and slowly and smoothly decelerate at a more comfortable acceleration to transition to the steady-state following distance (T gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to the steady-state following distance in this case; when S t Reach T gap At a steady-state following distance of +0.4, the acceleration value a slowly decreases to 0, and the ego vehicle maintains the current following distance. Vehicle information generated by the exposure is automatically filtered, and no acceleration or deceleration occurs before entering a tunnel.

[0071] The specific formula for adjusting the acceleration is as follows:

[0072]

[0073] a=0, when S t Equal to the current time interval plus the steady-state following distance after 0.4s (2)

[0074] a is the requested value of the longitudinal acceleration of the vehicle; S t is the real-time distance between the current vehicle and the preceding vehicle; S gap T is the steady-state following distance under the set time interval of the self-vehicle ((self-vehicle cruising speed - current speed of the preceding vehicle) * set time interval); gap is the time interval value currently set by the ACC system, in seconds; C is the empirical coefficient, based on S t -S gapThe acceleration a is obtained from a table lookup and is used to gain or filter the acceleration value. V1 is the cruising speed of the ego vehicle, and V2 is the current speed of the preceding vehicle (target vehicle). Formulas (1) and (2) are related, meaning that during a gradual deceleration process, a is not a fixed value but a variable value, such as a gradual change from -0.2 m / s^2 to 0. When the distance between the ego vehicle and the preceding vehicle reaches the steady-state following distance, acceleration a becomes 0, and deceleration stops.

[0075] If the traffic sign attribute information only contains tunnels (i.e., there is a tunnel but no speed limit sign), the vehicle maintains the current cruising speed and decelerates slowly and smoothly at a comfortable acceleration to a steady-state following distance (T) equal to the current time distance plus 0.4s. gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to a time distance of T in this case. gap +0.4 steady-state following distance; when S t Reach T gap At a steady-state following distance of +0.4, the acceleration value a slowly decreases to 0, and the ego vehicle maintains the current following distance. Vehicle information generated by exposure is automatically filtered, and no acceleration or deceleration occurs before entering a tunnel.

[0076] If the traffic sign attribute information only contains a speed limit sign (i.e., there is a speed limit sign but no tunnel), the camera determines whether the front is a tunnel entrance based on the illumination change information. The specific operations include:

[0077] If the illuminance changes from 60,000-80,000 lux to 1,000-2,000 lux, it is determined that there is a tunnel entrance ahead of the vehicle, and the speed limit sign value is determined to be lower than the vehicle's cruising speed. If so, the vehicle's ACC system will change the cruising speed to the speed limit sign value and slowly reduce the speed at a more comfortable acceleration, smoothly decelerating to the steady-state following distance (T) of the current time distance plus 0.4s. gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to a time distance of T in this case. gap +0.4 steady-state following distance; when S t Reach T gap +0.4 steady-state following distance, the acceleration a value will slowly become 0, the ego vehicle maintains the current distance and follows the vehicle, and a voice broadcast is played on the vehicle to remind the user of the reason for the ego vehicle to slow down; and before the ego vehicle enters the tunnel, the perception information of the preceding vehicle before entering the tunnel is maintained, and the vehicle information generated by exposure is automatically filtered, and the ego vehicle does not accelerate or decelerate. If not, the ego vehicle maintains the current cruising speed and slowly and smoothly decelerates at a more comfortable acceleration to transition to the steady-state following distance (T gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to a time distance of T in this case. gap+0.4 steady-state following distance; when S t Reach T gap At a steady-state following distance of +0.4, the acceleration value a will slowly decrease to 0, and the ego vehicle will maintain the current distance and follow the vehicle. Vehicle information generated by automatic filtering exposure will not be accelerated or decelerated before entering a tunnel. The specific formulas for adjusting the acceleration are shown in formulas (1) and (2) above. If the illumination change information remains small between 60,000 and 80,000 lux, maintain the current cruising state.

[0078] If there is no traffic sign attribute information (i.e., no speed limit sign or tunnel), the camera determines whether the front is a tunnel entrance based on the illumination change information, as follows:

[0079] If the illuminance changes from 60,000-80,000 lux to 1,000-2,000 lux, it is determined that the tunnel entrance is in front of the vehicle. The vehicle maintains the current cruising speed and decelerates slowly and smoothly at a comfortable acceleration to a steady-state following distance (T) equal to the current time distance plus 0.4 seconds. gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to a time distance of T in this case. gap +0.4 steady-state following distance; when S t Reach T gap At a steady-state following distance of +0.4, the acceleration value a will slowly decrease to 0, and the ego vehicle will maintain the current distance and follow the vehicle. Vehicle information generated by automatic filtering exposure will not be accelerated or decelerated before entering the tunnel. The specific formulas for adjusting the acceleration are shown in formulas (1) and (2) above. If the illumination change information remains unchanged between 60,000 and 80,000 lux, the ego vehicle will maintain its current cruising state.

[0080] If there is light in the tunnel, and the illuminance changes from 30-1500 lux or from 1-3 lux to 40,000-50,000 lux, it is determined that the tunnel exit is in front of the vehicle. The vehicle maintains the current cruising speed and decelerates slowly and smoothly at a comfortable acceleration to a steady-state following distance (T) equal to the current time distance plus 0.4s. gap +0.4), that is, the distance between the vehicle and the preceding vehicle will be extended to a time distance of T in this case. gap +0.4 steady-state following distance; when S t Reach T gap At a steady-state following distance of +0.4, the acceleration value a slowly decreases to 0, and the ego vehicle maintains the current following distance. Vehicle information generated by the exposure is automatically filtered, and no acceleration or deceleration occurs before exiting the tunnel.

[0081] The operations of automatically filtering the vehicle information generated by the exposure in each of the above steps include:

[0082] Determine whether the vehicle information is vehicle information generated by exposure. If so, discard the vehicle information generated by exposure, that is, do not save the vehicle information generated by exposure.

[0083] There are many ways to determine whether the vehicle information is generated by exposure. For example, if the vehicle information meets the following three conditions at the same time, the vehicle information is determined to be generated by exposure:

[0084] 1. Use the relative distance output by the camera to calculate the relative speed. The relative speed is greater than the set threshold. The formula for calculating the relative speed is as follows:

[0085] Relative speed calculated by distance = c × |distance N cycles ago - distance in current cycle| / t,

[0086] Among them, c is the filter coefficient, which can make the relative speed smoother, and the distances are the relative distances output by the camera.

[0087] 2. The target ID changes, and the changed target ID does not appear within the next Q consecutive cycles. Q is set according to actual needs. For example, if the previous target ID was 0 and suddenly changes to 440, if 440 appears within the next Q consecutive cycles (for example, 8), the target is selected; otherwise, it is not selected.

[0088] 3. The distance jump within a unit cycle exceeds the set threshold. For example, based on the relative speed of the ego vehicle and the preceding vehicle, the relative distance change between the ego vehicle and the preceding vehicle within a cycle should be fixed or within a certain range. For example, if the relative distance change within a cycle was previously 1m / s, but now suddenly changes to 10m / s, this change is obviously too large and cannot be trusted.

[0089] In summary, the present invention does not perform acceleration or deceleration operations based on the vehicle information generated by exposure when entering or exiting a tunnel. This solves the problem of false braking caused by inaccurate tunnel scene perception information in pure visual solutions, or unstable ACC control caused by inaccurate detection of the distance and acceleration of the vehicle in front during the following process. This improves the driver's comfort when using adaptive cruise control when entering or exiting a tunnel, and enhances the sense of braking safety.

[0090] The above technical solution is only one embodiment of the present invention. For those skilled in the art, it is easy to make various types of improvements or modifications based on the principles disclosed in the present invention, and it is not limited to the technical solution described in the above specific embodiments of the present invention. Therefore, the above description is only preferred and does not have a restrictive meaning.

Claims

1. A tunnel target detection and longitudinal control method, characterized by: The method uses the identification illumination information, vehicle information, and traffic sign attribute information obtained by the camera to detect tunnel targets and perform longitudinal control in scenarios where there is no target to follow and scenarios where there is a target to follow. The method comprises: Step 1: Process the image captured by the camera to obtain identification illumination information, vehicle information, and traffic sign attribute information; Step 2: Determine whether there is vehicle information. If not, determine that there is no target to follow, and perform longitudinal control for the scenario where there is no target to follow. If yes, determine that there is a target to follow, and perform longitudinal control for the scenario where there is a target to follow. The operation of performing longitudinal control in the tunnel entry and exit scenario without a follow-up target in step 2 includes: If the traffic sign attribute information contains tunnel and speed limit signs, the system determines whether the speed limit sign value is lower than the vehicle's cruising speed. If so, the system changes the cruising speed to the speed limit sign value and slowly reduces the speed. At the same time, a voice announcement reminds the user of the reason for the vehicle's deceleration. If not, the vehicle maintains its current cruising speed and automatically filters out the vehicle information generated by the exposure, without accelerating or decelerating before entering a tunnel. If the traffic sign attribute information only contains tunnel, the vehicle maintains its current cruising speed, automatically filters out the vehicle information generated by the exposure, and does not accelerate or decelerate before entering the tunnel. If the traffic sign attribute information only contains a speed limit sign, the camera will determine the situation ahead based on the illumination change information. If the speed limit sign is lower than the vehicle's cruising speed, the camera will determine whether the speed limit sign is lower than the vehicle's cruising speed. If so, the vehicle will change its cruising speed to the speed limit sign and slowly reduce its speed. Meanwhile, a voice announcement will be made on the vehicle to remind the user of the reason for the reduction. If not, the vehicle will maintain its current cruising speed and automatically filter out the vehicle information generated by the exposure. The vehicle will not accelerate or decelerate before entering the tunnel. If there is no traffic sign attribute information, the camera will determine the situation ahead based on the illumination change information. If the vehicle is at the entrance to a tunnel, the vehicle will maintain its current cruising speed and automatically filter the vehicle information generated by the exposure. It will not accelerate or decelerate before entering the tunnel. If the vehicle is at the exit of a tunnel, the vehicle will maintain its current cruising speed and automatically filter the vehicle information generated by the exposure. It will not accelerate or decelerate before exiting the tunnel. The longitudinal control operation in step 2 for following the target into and out of the tunnel includes: If the traffic sign attribute information contains tunnel and speed limit signs, it is determined whether the speed limit sign value is lower than the cruise speed of the ego vehicle. If so, the cruise speed is changed to the speed limit sign value, and the vehicle is smoothly decelerated to a steady-state following distance. At the same time, a voice broadcast is made on the vehicle to remind the user of the reason for the deceleration of the ego vehicle. Before the ego vehicle enters the tunnel, the vehicle information of the preceding vehicle before entering the tunnel is maintained, and the vehicle information generated by exposure is automatically filtered. The ego vehicle does not accelerate or decelerate. If not, the ego vehicle maintains the current cruise speed, and smoothly decelerates to a steady-state following distance. The vehicle information generated by exposure is automatically filtered, and the ego vehicle does not accelerate or decelerate before entering the tunnel. If the traffic sign attribute information only contains tunnels, the vehicle maintains its current cruising speed and smoothly decelerates to a steady-state following distance. The vehicle information generated by the exposure is automatically filtered, and no acceleration or deceleration is performed before entering the tunnel. If there is only a speed limit sign in the traffic sign attribute information, the camera will judge the situation ahead based on the illumination change information. If it is a tunnel entrance, it will determine whether the speed limit sign value is lower than the vehicle's cruising speed. If so, the cruising speed will be changed to the speed limit sign value, and the vehicle will be smoothly decelerated to a steady-state following distance. At the same time, a voice broadcast will be made on the vehicle to remind the user of the reason for the vehicle's deceleration. Before the vehicle enters the tunnel, the vehicle information of the preceding vehicle before entering the tunnel will be maintained, and the vehicle information generated by exposure will be automatically filtered. The vehicle will not accelerate or decelerate. If not, the vehicle will maintain the current cruising speed, and smoothly decelerate to a steady-state following distance, and the vehicle information generated by exposure will be automatically filtered. The vehicle will not accelerate or decelerate before entering the tunnel. If there is no traffic sign attribute information, the camera will judge the situation ahead based on the illumination change information. If there is a tunnel entrance ahead, the vehicle will maintain its current cruising speed, smoothly decelerate to a steady-state following distance, and automatically filter the vehicle information generated by exposure. It will not accelerate or decelerate before entering the tunnel. If there is a tunnel exit ahead, the vehicle will maintain its current cruising speed, smoothly decelerate to a steady-state following distance, and automatically filter the vehicle information generated by exposure. It will not accelerate or decelerate before exiting the tunnel.

2. The tunnel target detection and longitudinal control method according to claim 1, characterized in that: The operation of step 1 includes: (11) Input the image captured by the camera into the neural network processing model to obtain simplified intermediate image information; (12) The intermediate image information is input into the segmentation model to obtain illumination change information, vehicle information, and traffic sign attribute information.

3. The tunnel target detection and longitudinal control method according to claim 1, characterized in that: The steady-state following distance refers to the distance obtained by adding the current time distance to the time distance of 0.4s.

4. The tunnel target detection and longitudinal control method according to claim 1, characterized in that: The operation of the camera judging the situation ahead according to the illumination change information includes: If the illuminance changes from 60,000-80,000 lux to 1,000-2,000 lux, it is determined that the front is a tunnel entrance; If the illumination change information remains between 60,000 and 80,000 lux, it is determined that the front is not a tunnel entrance; If the illumination change information changes from 30-1500 lux to 40,000-50,000 lux, or from 1-3 lux to 40,000-50,000 lux, it is determined that the front is a tunnel exit.

5. The tunnel target detection and longitudinal control method according to claim 1, characterized in that: The operation of automatically filtering the vehicle information generated by exposure includes: determining whether the vehicle information is vehicle information generated by exposure, and if so, eliminating the vehicle information generated by exposure.

Images