Vision-based road surface condition adaptive compensation brake-by-wire system

Abstract

The invention provides a pavement condition self-adaptive compensation brake-by-wire system based on vision, and belongs to the technical field of brake-by-wire. A pavement condition self-adaptive compensation brake-by-wire system based on vision is characterized by comprising a brake-by-wire module used for executing electronic brake control of a vehicle; the visual detection module is configured to collect road surface image data in real time and output actual road surface condition information; the expected data acquisition module is used for acquiring an expected road surface condition based on historical or model data; the comparison and judgment module is used for comparing the actual road surface condition with the expected road surface condition and generating an inconsistent signal when detecting that the actual road surface condition is inconsistent with the expected road surface condition; and the compensation execution module responds to the inconsistent signal and adjusts the brake parameters of the brake-by-wire module to carry out dynamic compensation, so that the brake performance of the vehicle is optimized.

Description

Technical Field

[0001] This invention belongs to the field of brake-by-wire technology, and in particular to a vision-based adaptive compensation brake-by-wire system for road conditions. Background Technology

[0002] With the increasing level of vehicle electrification, brake-by-wire technology has been widely applied in various vehicle braking scenarios due to its ability to achieve electronic transmission of braking commands, precise pressure adjustment, and multi-functional coordination. Its core is that the electronic control unit generates braking parameters based on the vehicle's status, driving the actuator to complete tasks such as deceleration and stability control.

[0003] Existing brake-by-wire systems mainly rely on vehicle dynamic parameters and preset road and weather information for control. However, actual road surfaces are highly variable in time and regional. Factors such as rain and sunlight can cause rapid changes in road conditions, such as slippery surfaces or debris covering the road. Relying solely on historical information or low-frequency updated data can easily lead to deviations between braking parameters and actual road conditions. Summary of the Invention

[0004] The purpose of this invention is to address the aforementioned problems in existing technologies by proposing a vision-based adaptive compensation line control braking system for road conditions.

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A vision-based adaptive road condition compensation brake system, characterized in that it includes:

[0007] Brake-by-wire module, used to perform electronic braking control of the vehicle;

[0008] The visual inspection module is configured to acquire road surface image data in real time and output actual road surface condition information;

[0009] The expected data acquisition module is used to acquire expected road surface conditions based on historical or model data;

[0010] The comparison and judgment module is used to compare the actual road surface condition with the expected road surface condition, and generate an inconsistency signal when an inconsistency is detected.

[0011] The compensation execution module, in response to the inconsistency signal, adjusts the braking parameters of the brake-by-wire module to perform dynamic compensation, thereby optimizing the vehicle's braking performance.

[0012] Furthermore, the visual detection module is specifically used for:

[0013] Road surface image data of the area in front of the vehicle is collected according to a preset sampling frequency, and the collection time, vehicle position and speed information are recorded synchronously with the road surface image data.

[0014] The road surface image data is subjected to distortion correction, brightness normalization, noise suppression, and field of view cropping to obtain a standardized road surface image.

[0015] The standardized road surface image is subjected to texture features, water reflection features, attachment edge features, and road marking occlusion features to obtain road surface feature parameters;

[0016] The actual road surface condition information is generated based on the road surface feature parameters, and the actual road surface condition information is output to the comparison and judgment module.

[0017] 3. The vision-based adaptive road condition compensation brake system according to claim 1, wherein the expected data acquisition module is specifically used for:

[0018] Obtain historical road condition records, historical weather records, and historical braking response records corresponding to the current vehicle location, as well as the corresponding historical time period identifiers;

[0019] The historical road condition records, historical weather records, and historical braking response records are time-aligned, location-mapped, and anomaly-removed to obtain the basic expected dataset.

[0020] Based on the aforementioned basic expected dataset, extract the associated features corresponding to the current vehicle position, current vehicle speed, current driving direction, and current steering angle;

[0021] The expected road surface condition is generated based on the associated features and output to the comparison and judgment module.

[0022] 4. The vision-based adaptive road condition compensation brake system according to claim 3, wherein the expected data acquisition module is further specifically used for:

[0023] The historical road condition records, historical weather records, and historical braking response records in the basic expected dataset are feature-encoded to obtain the expected input feature vector;

[0024] The expected input feature vector is input into the road surface condition prediction model to obtain candidate expected road surface conditions;

[0025] The candidate expected road surface conditions are matched and corrected with the model data to obtain corrected candidate expected road surface conditions;

[0026] The expected road surface condition is determined based on the correspondence between the corrected candidate expected road surface condition and the current vehicle position;

[0027] The expected road surface condition is sent to the comparison and judgment module, and the corresponding generation time is written into the expected data cache.

[0028] Furthermore, the comparison and judgment module is specifically used for:

[0029] Receive the actual road surface condition information and the expected road surface condition, and establish corresponding comparison items according to road surface type, degree of slipperiness, degree of foreign object coverage and degree of visible marking integrity;

[0030] The difference value is calculated for each comparison item and a difference matrix is ​​formed. Then, the existence of inconsistency is determined according to the preset inconsistency judgment rule and the continuous frame change threshold, and the inconsistency position index is recorded.

[0031] When an inconsistency is determined to exist, an inconsistency signal is generated, and the inconsistency signal, along with the corresponding difference value, difference matrix, and inconsistency location index, is sent to the compensation execution module.

[0032] Furthermore, the compensation execution module is specifically used for:

[0033] The system acquires the inconsistency signal, the actual road surface condition information, the expected road surface condition, and the current braking parameters of the brake-by-wire module.

[0034] A compensation input feature vector is constructed based on the inconsistency signal, the actual road surface condition information, the expected road surface condition, and the current braking parameters;

[0035] The compensation input feature vector is input into the braking parameter compensation model to obtain the braking parameter adjustment amount;

[0036] Based on the braking parameter adjustment amount, the brake force distribution parameter, brake pressure increase parameter, and anti-lock braking trigger parameter are constrained and calculated to obtain the target braking parameters;

[0037] The target braking parameters are sent to the brake-by-wire module to perform the dynamic compensation.

[0038] Furthermore, the compensation execution module is also specifically used for:

[0039] After the brake-by-wire module performs the dynamic compensation according to the target braking parameters, it collects the vehicle deceleration, wheel slip ratio, vehicle attitude change and current vehicle speed.

[0040] The consistency of the vehicle deceleration, the wheel slip ratio, the change in vehicle posture, the current vehicle speed, the braking parameter adjustment, and the inconsistency signal is verified to obtain the compensation execution result and the compensation deviation identifier.

[0041] The model parameters of the braking parameter compensation model are updated based on the compensation execution result and the compensation deviation identifier, and the updated model parameters are written into the compensation model storage area.

[0042] Furthermore, the brake-by-wire module is specifically used for:

[0043] Receive the target braking parameters sent by the compensation execution module, and perform parameter integrity verification and boundary constraint verification on the target braking parameters;

[0044] Based on the verified target braking parameters, electronic braking control commands are generated for the corresponding wheels.

[0045] The electronic braking control commands are sent to the braking execution unit to control the establishment, maintenance, or release of braking pressure on each wheel;

[0046] During the electronic braking control process, the execution status, pressure establishment status, and command execution status are fed back in real time, and the execution status is sent to the compensation execution module and the comparison judgment module.

[0047] Furthermore, it also includes a collaborative control module, which is specifically used for:

[0048] The outputs of the visual detection module, the expected data acquisition module, the comparison and judgment module, and the compensation execution module are synchronized in time to obtain a synchronization control sequence.

[0049] The synchronization control sequence is subjected to integrity verification to determine whether there is data loss, timing limit violation or signal conflict;

[0050] A degradation control signal is generated when there is missing data, timing limit exceedance, or signal conflict.

[0051] The braking parameter adjustment amount is frozen according to the degradation control signal, and the brake-by-wire module executes electronic braking control according to the most recent valid target braking parameters.

[0052] After the synchronization control sequence is restored to meet the integrity check, the degradation control signal is released and the dynamic compensation is restored.

[0053] Furthermore, the comparison and judgment module is also used to receive the actual road surface information and the expected road surface information again when no inconsistency is detected between the actual road surface condition and the expected road surface condition, and to establish corresponding comparison items according to the road surface type, degree of slipperiness, degree of foreign object coverage and degree of visible marking integrity.

[0054] Compared with the prior art, this application has the following advantages:

[0055] The core advantage of this application lies in constructing a complete closed-loop control link, achieving adaptive compensation for road conditions through multi-module collaboration. The visual detection module extracts key features through image preprocessing, transforming them into structured road information to ensure data consistency. The expected data acquisition module combines historical data with the current vehicle state, generating accurate expected road conditions through multi-level constraints to reduce deviations. The system generates a difference matrix and inconsistency location index by comparing actual and expected key road indicators item by item, providing fine-grained basis for compensation. The compensation execution module constructs targeted feature vectors and adjusts braking parameters based on model output constraints, balancing real-time performance and continuity. Furthermore, the system has execution feedback and model update mechanisms, verifying compensation effects and optimizing the model through real-time feedback. The collaborative control module achieves time synchronization among multiple modules, triggering degraded control in case of anomalies to ensure braking continuity and reliability. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the overall structure of the vision-based adaptive road condition compensation line control braking system in an embodiment of the present invention.

[0057] Figure 2 This is a schematic diagram of the processing flow of the visual detection module in an embodiment of the present invention.

[0058] Figure 3 This is a schematic diagram of the processing flow of the expected data acquisition module in an embodiment of the present invention.

[0059] Figure 4 This is a schematic diagram of the comparison and analysis process of the comparison and judgment module in an embodiment of the present invention.

[0060] Figure 5 This is a schematic diagram of the dynamic compensation process of the compensation execution module in an embodiment of the present invention.

[0061] Figure 6 This is a schematic diagram illustrating the feedback update relationship between the linear braking module and the compensation execution module in an embodiment of the present invention.

[0062] Figure 7 This is a schematic diagram of the time synchronization, integrity verification, and degradation control process of the collaborative control module in this embodiment of the invention. Detailed Implementation

[0063] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings. It should be understood that the following embodiments are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Where there is no conflict, the technical features in the following embodiments can be combined with each other.

[0064] like Figures 1 to 7As shown: In some embodiments, the vision-based adaptive compensation brake-by-wire system for road conditions provided by the present invention has the following overall technical thread: First, a vision detection module collects and processes road image data in real time to obtain actual road condition information; then, an expected road condition is generated based on historical or model data by an expected data acquisition module; subsequently, a comparison and judgment module compares the actual road condition information with the expected road condition to determine whether there is an inconsistency, and generates an inconsistency signal if an inconsistency exists; then, a compensation execution module constructs a compensation input feature vector based on the inconsistency signal, the actual road condition information, the expected road condition, and the current braking parameters of the brake-by-wire module to obtain the braking parameter adjustment amount and generate the target braking parameters; finally, the brake-by-wire module executes electronic braking control according to the target braking parameters and feeds back the execution status to the compensation execution module and the comparison and judgment module. This forms a complete chain control process from road perception input to braking execution output and then to feedback update.

[0065] In some embodiments, the technical objective of the above-described overall link is to enable the vehicle to perform basic braking control not only according to the current braking request during online braking, but also to dynamically compensate for the braking force distribution parameters, braking pressure increase parameters, and anti-lock braking trigger parameters according to the difference between the actual road conditions and the expected road conditions, so as to improve the matching degree between braking control and the current road conditions.

[0066] Reference Figure 1 A vision-based adaptive compensation brake system for road conditions includes:

[0067] The brake-by-wire module is used to perform electronic braking control of the vehicle.

[0068] The brake-by-wire module generates electronic braking control logic for the corresponding wheels based on the received target braking parameters, and then drives the braking actuator to establish, maintain or release wheel braking pressure.

[0069] Specifically, the brake-by-wire module, as the execution end of the system, does not directly identify road conditions. Instead, it receives the target braking parameters from the compensation execution module and converts the parameter-level control results into actual braking actions. In brake-by-wire scenarios, electronic braking control can include front-to-rear wheel braking force distribution control, brake pressure increase rate control, and anti-lock braking system (ABS) related trigger control. Since the brake-by-wire module directly affects the vehicle's braking performance, it needs to maintain a consistent parameter interface with the compensation execution module and a feedback channel for execution status information with the comparison and judgment module.

[0070] In some embodiments, the braking force distribution parameter in the brake-by-wire module can characterize the braking force distribution ratio between the front axle and the rear axle or between the left wheel and the right wheel; the braking pressure increase parameter can characterize the magnitude of the increase in braking pressure with the control cycle; and the anti-lock braking trigger parameter can characterize the threshold boundary for triggering anti-lock braking control when the wheel slip ratio is detected to reach a certain condition.

[0071] The visual inspection module is configured to acquire road surface image data in real time and output information on the actual road surface conditions.

[0072] The visual inspection module uses the image acquisition unit to acquire road surface image data and extracts and expresses the road surface condition-related features in the image.

[0073] Specifically, the visual detection module corresponds to the front-end perception link of the entire system, and its output of actual road surface information describes the real road surface condition of the area where the vehicle is currently traveling. Unlike simply outputting raw images, the visual detection module in this invention outputs structured results that can participate in comparison and compensation calculations. These structured results may include road surface type, slipperiness, foreign object coverage, and the integrity of visible road markings. Since the output of the visual detection module is directly input into the comparison and judgment module, this module needs to ensure that the output data is synchronized with the vehicle's position and speed information in time, and is comparable in format to the expected road surface conditions output by the expected data acquisition module.

[0074] The visual inspection module acquires continuous road surface image data of the area in front of the vehicle at a sampling frequency of 20Hz. For the current control cycle, the visual inspection module identifies water accumulation and reflective strips, edges of light-colored attachments, and partial obstruction of road markings in the images, and converts this information into actual road surface condition information. This actual road surface condition information includes: road surface type asphalt pavement, slipperiness level of 0.72, foreign object coverage level of 0.18, and visible road marking integrity level of 0.41.

[0075] The expected data acquisition module is used to acquire expected road surface conditions based on historical or model data.

[0076] The expected data acquisition module obtains data related to the current vehicle position from historical road condition records, historical weather records, historical braking response records, or model data, and forms the expected road condition corresponding to the current control cycle.

[0077] Specifically, the expected data acquisition module generates a priori judgment result for the current road segment based on existing data; essentially, it is a reference road surface condition. This reference road surface condition can be formed directly based on historical statistical results, or it can be inferred from multi-source historical data through a model. The expected road surface condition does not replace the actual road surface condition information, but is used to compare with the actual road surface condition information to identify whether the current road surface condition has undergone abnormal changes relative to existing knowledge.

[0078] The expected data acquisition module generates the expected road conditions for the cost control cycle based on historical road conditions corresponding to the current vehicle location, historical weather records for the past three hours, and historical braking response records for the same road segment. These expected road conditions include: road type asphalt pavement, slipperiness level of 0.48, foreign object coverage of 0.05, and visible road marking integrity of 0.78.

[0079] The comparison and judgment module is used to compare the actual road surface condition with the expected road surface condition, and generate an inconsistency signal when an inconsistency is detected.

[0080] The comparison and judgment module compares the two types of input data in the same dimension to determine whether they are consistent. If there is an inconsistency that exceeds the preset judgment rules, an inconsistency signal is generated.

[0081] Specifically, this module does not simply compare a single value, but compares multiple items one by one, and can record the location and degree of differences. Since actual changes in road surface may manifest in various forms, such as abrupt changes in slipperiness, increased coverage of foreign objects, or decreased integrity of visible road markings, the comparison and judgment module can quantitatively compare the actual road surface conditions with the expected road surface conditions within the same comparison framework, thereby providing the compensation execution module with directly usable evidence of differences.

[0082] In some embodiments, the inconsistency signal can be in the form of a binary signal or a structured signal containing a timestamp, difference level, or location index. When the comparison and judgment module outputs an inconsistency signal, it can also simultaneously output the main comparison items that caused the inconsistency and their corresponding difference values.

[0083] After comparing the actual road surface condition information output by the visual detection module and the expected road surface condition information output by the expected data acquisition module, the comparison and judgment module obtains the following differences: slipperiness difference (0.24), foreign object coverage difference (0.13), and visible road marking integrity difference (0.37). Since both the slipperiness difference and visible road marking integrity difference exceed their respective thresholds and show a continuously increasing trend over three consecutive frames, the comparison and judgment module generates an inconsistency signal and records that the inconsistency is mainly concentrated in the section of road 8m to 15m ahead of the vehicle.

[0084] The compensation execution module, in response to the inconsistency signal, adjusts the braking parameters of the brake-by-wire module to perform dynamic compensation, thereby optimizing the vehicle's braking performance.

[0085] The compensation execution module determines the direction and magnitude of compensation required based on the input information, and adjusts the braking parameters of the brake-by-wire module to form the target braking parameters.

[0086] Specifically, the compensation execution module occupies a central decision-making position between discrepancy identification and braking execution. This module does not simply perform fixed increases or decreases based on inconsistent signals; instead, it comprehensively generates braking parameter adjustments by combining actual road conditions, anticipated road conditions, and current braking parameters. Subsequently, it performs constraint calculations on the brake force distribution parameters, brake pressure increase parameters, and anti-lock braking system (ABS) trigger parameters within the execution boundary range to prevent the compensation process from being overly aggressive or exceeding control boundaries.

[0087] In some embodiments, if the actual road surface information indicates that the slipperiness is higher than the expected road surface condition, the compensation execution module can reduce the braking pressure increase parameter and advance the trigger boundary of the anti-lock braking system (ABS) trigger parameter. If the actual road surface information indicates that the coverage of foreign objects is increased, the compensation execution module can increase the constraint requirements for establishing wheel end pressure balance. If the actual road surface information indicates that the integrity of visible road markings is reduced, the compensation execution module can enhance the monitoring priority of the current vehicle body attitude change. After receiving the above-mentioned inconsistent signals, the compensation execution module reads the current braking parameters, where the brake force distribution parameter is 0.50 / 0.50, the brake pressure increase parameter is 0.30 MPa / cycle, and the ABS trigger parameter is 0.18. Based on the combination difference between the high slipperiness and the low integrity of visible road markings, the compensation execution module generates the braking parameter adjustment amount, and then outputs the target braking parameters as follows: brake force distribution parameter 0.46 / 0.54, brake pressure increase parameter 0.22 MPa / cycle, and ABS trigger parameter 0.14.

[0088] Reference Figure 2 The visual detection module is specifically used for:

[0089] Road surface image data of the area in front of the vehicle is collected according to a preset sampling frequency, and the collection time, vehicle position and speed information are recorded synchronously with the road surface image data.

[0090] Road surface image data are collected according to a preset sampling frequency, and the collection time, vehicle position and speed information are recorded at the same sampling time.

[0091] Specifically, the preset sampling frequency is used to limit the temporal resolution of the road surface image data acquired by the visual detection module. A higher sampling frequency is more conducive to capturing rapidly changing road surface conditions within a short period of time. At the same time, the acquisition time, vehicle position, and vehicle speed information are recorded synchronously with the road surface image data, which is beneficial for establishing a one-to-one correspondence between the image and the vehicle state later. The vehicle position can be output by the on-board positioning unit, the vehicle speed information can be output by the vehicle speed sensing unit, and the acquisition time can be provided by the on-board controller clock.

[0092] In some embodiments, if the preset sampling frequency is denoted as The sampling period is denoted as Then it can be satisfied:

[0093]

[0094] in, Indicates the preset sampling frequency. This represents the sampling time interval between two adjacent frames of road surface image data. (Controlled by...) This ensures that the visual inspection module obtains a sufficient amount of road surface image data per unit time.

[0095] In a post-rain urban road scene, the visual detection module acquires road surface image data of the area in front of the vehicle at a preset sampling frequency of 20Hz. The current sampling time is 10:15:23.150, the vehicle position is at road coordinates P (126.3m, 48.7m), and the vehicle speed is 60km / h. The input for this step is the real-time scene of the road surface in front, the intermediate processing results are synchronous sampling and status recording, and the output is road surface image data containing acquisition time, vehicle position, and vehicle speed information.

[0096] The road surface image data is subjected to distortion correction, brightness normalization, noise suppression, and field of view cropping to obtain a standardized road surface image.

[0097] Distortion correction, brightness normalization, noise suppression, and field of view cropping are performed sequentially on the road image data.

[0098] Specifically, distortion correction is used to eliminate geometric distortion caused by the image acquisition lens, making the spatial relationship of the road surface area closer to the real state; brightness normalization is used to reduce the impact of different lighting conditions on the image brightness distribution; noise suppression is used to reduce errors caused by raindrops, sensor noise, and high-frequency disturbances; and field-of-view cropping is used to retain the forward driving area most relevant to the vehicle's current braking decision and remove irrelevant areas such as the sky and vehicle body structure. Through the above preprocessing, road surface image data from different sources and under different shooting conditions can be unified onto a relatively consistent processing basis.

[0099] In some embodiments, the image pixel values ​​after brightness normalization It can be represented as:

[0100]

[0101] in, This indicates the original road surface image data in coordinates. Pixel value at that location, This represents the minimum pixel value in the current road surface image data. This represents the maximum pixel value in the current road surface image data. This represents the pixel value after brightness normalization.

[0102] For the acquired road surface image data, the visual inspection module first corrects the barrel distortion caused by the wide-angle lens, then normalizes the brightness of the bright areas caused by local sunlight reflection, subsequently suppresses raindrop noise in the image through filtering, and finally crops the road surface area within a range of 3m to 20m in front of the vehicle to obtain a standardized road surface image. The input to this step is the original road surface image data, the intermediate processing results are the distortion-corrected image, the brightness-normalized image, and the noise-suppressed image, and the output is the standardized road surface image.

[0103] The standardized road surface image is subjected to texture features, water reflection features, attachment edge features, and road marking occlusion features to obtain road surface feature parameters;

[0104] Texture features, water reflection features, attachment edge features, and road marking occlusion features are extracted from standardized road images, and these features are combined to form road feature parameters.

[0105] Specifically, texture features are used to reflect the surface roughness, particle structure, and local material variations of the road surface; water reflection features are used to reflect specular reflection and highlight distribution caused by water accumulation in the image; attachment edge features are used to reflect the boundary distribution of attachments such as mud, fallen leaves, or debris on the road surface; and road marking occlusion features are used to reflect the visibility and coverage of road markings in the current image. These four types of features together constitute the feature basis for judging road surface condition.

[0106] In some embodiments, the road surface feature parameter vector It can be represented as:

[0107]

[0108] in, Represents texture features, Indicates the reflective characteristics of accumulated water. Indicates the edge features of the attachment. Indicates the occlusion feature of the road markings. This represents the vector of road surface feature parameters.

[0109] After the visual inspection module extracts features from the standardized road surface image, it obtains texture features. =0.37, water reflection characteristics =0.81, edge features of the attachment =0.28, marking occlusion feature =0.59. Among them, the higher water reflection characteristic indicates that there is a significant water reflection area ahead, and the higher marking obscuration characteristic indicates that the markings in some lanes are obscured by water film and attached objects.

[0110] The actual road surface condition information is generated based on the road surface feature parameters, and the actual road surface condition information is output to the comparison and judgment module.

[0111] The current road surface condition is mapped, classified, or quantified based on road surface feature parameters to generate actual road surface condition information, which is then sent to the comparison and judgment module.

[0112] Specifically, the actual road surface condition information can be a structured result containing multiple dimensions, used to characterize the road surface type, slipperiness, foreign object coverage, and visible road marking integrity in the current vehicle's forward driving area. This step transforms the image feature layer results into control decision layer results and is a key interface step between the visual detection module and the comparison and judgment module.

[0113] In some embodiments, slipperiness Foreign object coverage and the integrity of visible markings It can be obtained by mapping road surface feature parameters, for example:

[0114]

[0115] in, Indicates the degree of slipperiness. Indicates the degree of foreign object coverage. Indicates the completeness of visible markings; Indicates the reflective characteristics of accumulated water. Represents texture features, Indicates the edge features of the attachment. Indicates the occlusion feature of the road markings; , , and Represents the feature mapping coefficients.

[0116] Based on the obtained road surface feature parameters, the visual detection module generates the following actual road surface condition information: road surface type is asphalt pavement, slipperiness level is 0.72, foreign object coverage is 0.18, visible road marking integrity is 0.41, and this actual road surface condition information is output to the comparison and judgment module. The input of this step is the road surface feature parameters, the intermediate processing result is the mapping calculation of each state variable, and the output result is the actual road surface condition information.

[0117] Reference Figure 3 The expected data acquisition module is specifically used for:

[0118] Obtain historical road condition records, historical weather records, and historical braking response records corresponding to the current vehicle location, as well as the corresponding historical time period identifiers;

[0119] Based on the current vehicle location, retrieve historical road condition records, historical weather records, and historical braking response records from the historical data storage unit, and simultaneously obtain the corresponding historical time period identifiers for the above records.

[0120] Specifically, the key to this step lies in "corresponding to the current vehicle location." That is, the retrieved historical data is not arbitrary, but rather data content associated with the current spatial location of the vehicle. Historical time period identifiers reflect the time interval in which the aforementioned historical data was generated, facilitating subsequent time alignment and scene similarity analysis. Historical road condition records characterize the road surface condition at that location during a historical time period; historical weather records characterize environmental conditions such as rainfall and temperature during the corresponding time period; and historical braking response records characterize the vehicle's braking response results in similar locations and environments.

[0121] The current vehicle location is at road coordinates P (126.3m, 48.7m). The expected data acquisition module extracts 120 sets of historical road surface condition records, 120 sets of historical weather records, and 95 sets of historical braking response records from the historical database within a 15m radius of this location over the past 30 days, recording their corresponding historical time period identifiers, such as "30 minutes after rain," "Morning rush hour on a cloudy day," and "Nighttime after continuous rainfall." The input for this step is the current vehicle location, the intermediate processing result is a location-related retrieval process, and the output is a set of historical data with historical time period identifiers.

[0122] The historical road condition records, historical weather records, and historical braking response records are time-aligned, location-mapped, and anomaly-removed to obtain the basic expected dataset.

[0123] First, time alignment is performed, then location mapping is performed, and then anomaly removal is performed to form the basic expected dataset.

[0124] Specifically, time alignment is used to unify the time base of historical data from different sources, enabling data from the same or similar time periods to be combined accordingly; location mapping is used to map each historical data point to a spatial index frame consistent with the current vehicle location; anomaly removal is used to remove missing items, obviously distorted items, or data items that exceed the normal range. After this process, the subsequently extracted correlation features are based on a relatively consistent data foundation.

[0125] In some embodiments, time deviations may be utilized and positional deviation The data is filtered, including:

[0126]

[0127] in, Indicates the historical time. Indicates the current time reference point. Indicates time deviation; This indicates the location coordinates corresponding to the historical records. Indicates the current vehicle position coordinates. This indicates positional deviation.

[0128] The expected data acquisition module aligns the aforementioned 120 sets of historical road surface condition records, 120 sets of historical weather records, and 95 sets of historical braking response records along a unified timeline and maps them to the location grid corresponding to the current road coordinates P (126.3m, 48.7m). Subsequently, 17 sets of data with missing time, 9 sets with excessive positional deviations, and 6 sets with abnormal braking responses are removed, ultimately yielding the basic expected dataset, which contains 183 valid associated records.

[0129] Based on the aforementioned basic expected dataset, extract the associated features corresponding to the current vehicle position, current vehicle speed, current driving direction, and current steering angle;

[0130] Extract data patterns that match the current state from the basic expected dataset to form associated features.

[0131] Specifically, the correlation features reflect the similarity between the current location, current speed, current direction of travel, and current steering angle and historical samples. Since the braking response and road experience may differ on the same road segment under different speeds and steering states, this step incorporates the current speed, current direction of travel, and current steering angle as filtering or weighting conditions to make the expected road conditions closer to the current operating conditions.

[0132] In some embodiments, associated feature weights It can be represented as:

[0133]

[0134] in, Indicates the first The correlation weights corresponding to the basic expected data. Indicates the similarity at the current position. Indicates the similarity of current vehicle speeds. Indicates the similarity of the current driving direction. Indicates the similarity of the current steering angle. , , and This represents the weight coefficient corresponding to each similarity level.

[0135] For example, the current vehicle speed is 60 km / h, the current direction of travel is east, and the current steering angle is 2.5°. The expected data acquisition module filters out 37 highly correlated samples from the basic expected dataset that are close to the current location, have a vehicle speed between 55 km / h and 65 km / h, a direction of travel near east, and a steering angle between 0° and 4°, and extracts correlation features based on these samples.

[0136] The expected road surface condition is generated based on the associated features and output to the comparison and judgment module.

[0137] Based on the associated features, the system generates information such as road surface type, slipperiness, foreign object coverage, and the integrity of visible markings to form the expected road surface condition, which is then output to the comparison and judgment module.

[0138] Specifically, this step transforms the associated features from sample-level information to result-level information. The expected road surface condition can be formed through statistical summarization, weighted averaging, or model inference. The key is that it can serve as input for the subsequent comparison and judgment module, and be compared with the actual road surface condition information generated by the visual detection module at the same semantic level.

[0139] For example, the expected data acquisition module generates the expected road surface condition based on 37 highly correlated samples: the road surface type is asphalt road surface, the slipperiness is 0.48, the foreign object coverage is 0.05, the visible marking integrity is 0.78, and outputs it to the comparison and judgment module.

[0140] The expected data acquisition module is also specifically used for:

[0141] The historical road condition records, historical weather records, and historical braking response records in the basic expected dataset are feature-encoded to obtain the expected input feature vector;

[0142] Feature encoding is performed on historical road condition records, historical weather records, and historical braking response records in the basic expected dataset.

[0143] Specifically, feature encoding transforms different types of data into a unified vector representation for subsequent input into the road condition prediction model. Historical road condition records can be encoded as road condition-related vectors, historical weather records as weather condition-related vectors, and historical braking response records as braking dynamic-related vectors. The encoded expected input feature vectors retain the correspondence between these various data types.

[0144] In some embodiments, the expected input feature vector It can be represented as:

[0145]

[0146] in, This represents the expected input feature vector. This represents the encoding vector corresponding to the historical road surface condition record. This represents the encoding vector corresponding to historical weather records. This represents the encoding vector corresponding to the historical braking response record.

[0147] For example, the expected data acquisition module encodes 37 highly correlated samples from the basic expected dataset to form an expected input feature vector of length 48 dimensions. The first 16 dimensions correspond to historical road surface condition records, the middle 12 dimensions correspond to historical weather records, and the last 20 dimensions correspond to historical braking response records.

[0148] The expected input feature vector is input into the road surface condition prediction model to obtain candidate expected road surface conditions;

[0149] The expected input feature vector is input into the road condition prediction model, and the model inference is performed.

[0150] Specifically, the road condition prediction model is used to extract corresponding patterns from the encoded features of historical samples to predict the possible road conditions at the current location and under the current operating conditions. The candidate expected road conditions are the initial results output by the model, and they still need to be further matched and corrected with the model data.

[0151] For example, the expected data acquisition module inputs the 48-dimensional expected input feature vector into the road condition prediction model, and the model outputs candidate expected road conditions: the road type is asphalt road, the slipperiness is 0.52, the foreign object coverage is 0.07, and the visible marking integrity is 0.73.

[0152] The candidate expected road surface conditions are matched and corrected with the model data to obtain corrected candidate expected road surface conditions;

[0153] The candidate expected road surface conditions are matched and corrected with the model data.

[0154] Specifically, model data can be understood as reference parameters, calibration samples, or state mapping rules stored in conjunction with the road condition prediction model. Through matching calibration, the candidate expected road conditions can be kept consistent with the distribution constraints formed during long-term model training, preventing the model from deviating under individual input conditions. The key to this step is "matching calibration," which means not simply accepting the model output, but further constraining the model output.

[0155] In some embodiments, the candidate expected road surface condition is corrected. Can be determined by the expected road surface conditions of the candidates and model data reference values To be determined together, for example:

[0156]

[0157] in, This represents a state variable in the candidate expected road surface condition for correction. This represents the corresponding state quantity in the candidate expected road surface conditions. This represents the corresponding reference state quantity in the model data. This represents the matching correction coefficient.

[0158] For example, regarding the degree of slipperiness, the candidate expected road surface condition output value is 0.52, and the corresponding reference state variable in the model data is 0.46, with a correction coefficient... Under the condition of 0.6, the corrected slipperiness is 0.496. Similarly, the foreign object coverage and visible marking integrity were also matched and corrected, and the corrected candidate expected pavement conditions were finally obtained: pavement type is asphalt pavement, slipperiness is 0.50, foreign object coverage is 0.06, and visible marking integrity is 0.76.

[0159] The expected road surface condition is determined based on the correspondence between the corrected candidate expected road surface condition and the current vehicle position;

[0160] Based on the correspondence between the candidate expected road conditions and the current vehicle position, the expected road conditions to be used in the current control cycle are determined from the candidate results.

[0161] Specifically, different candidate road surface conditions may correspond to different location units, different road segment indices, or different spatial mapping ranges. This step selects the result that best matches the current vehicle position as the final expected road surface condition based on the current location correspondence.

[0162] The current vehicle position is located in segment 18, at road coordinates P (126.3m, 48.7m). Based on the location mapping relationship, the expected data acquisition module determines the current expected road surface condition for segment 18 from the candidate expected road surface conditions for correction. Specifically, the road surface type is asphalt pavement, the slipperiness level is 0.48, the foreign object coverage level is 0.05, and the visible road marking integrity level is 0.78.

[0163] The expected road surface condition is sent to the comparison and judgment module, and the corresponding generation time is written into the expected data cache.

[0164] The expected road surface conditions are output to the comparison and judgment module, and the generation time is written to the expected data cache.

[0165] Specifically, the expected data cache is used to store the expected road surface conditions generated in each control cycle and their generation time, so that the most recent valid results can be called when the comparison and judgment module or the cooperative control module performs timing checks, backtracking comparisons, or degraded control. This step makes the generation process of expected road surface conditions traceable.

[0166] For example, the expected data acquisition module sends the expected road surface conditions generated in the current period to the comparison and judgment module, and writes the generation time "10:15:23.162" and the corresponding road segment identifier into the expected data cache area.

[0167] Reference Figure 4 The comparison and judgment module is specifically used for:

[0168] Receive the actual road surface condition information and the expected road surface condition, and establish corresponding comparison items according to road surface type, degree of slipperiness, degree of foreign object coverage and degree of visible marking integrity;

[0169] The two types of inputs are matched one-to-one according to road surface type, degree of wetness, degree of foreign object coverage, and degree of visible marking integrity to establish corresponding comparison items.

[0170] Specifically, the focus of this step is to map the two types of input results into the same comparison framework. Road surface category is a discrete comparison item, while slipperiness, foreign object coverage, and visible road marking integrity are continuous comparison items. After establishing the corresponding comparison items, the difference values ​​can then be calculated for each item.

[0171] For example, the comparison and judgment module receives the actual road surface condition information "asphalt road surface, slipperiness 0.72, foreign object coverage 0.18, visible marking integrity 0.41" and the expected road surface condition "asphalt road surface, slipperiness 0.48, foreign object coverage 0.05, visible marking integrity 0.78", forming four comparison items.

[0172] The difference value is calculated for each comparison item and a difference matrix is ​​formed. Then, the existence of inconsistency is determined according to the preset inconsistency judgment rule and the continuous frame change threshold, and the inconsistency position index is recorded.

[0173] The difference values ​​are calculated separately to form a difference matrix. Then, the inconsistency is determined by combining the preset inconsistency judgment rules and the continuous frame change threshold, and the inconsistency position index is recorded.

[0174] Specifically, the difference value is used to quantify the degree of deviation of each comparison item, the difference matrix is ​​used to carry the change relationship of multiple comparison items in consecutive frames in a structured manner, the preset inconsistency judgment rule is used to determine what combination of differences constitutes a valid inconsistency, and the consecutive frame change threshold is used to filter out misjudgments caused by accidental fluctuations in a single frame. The inconsistency location index is used to identify the spatial location or image region where the inconsistency occurs.

[0175] In some embodiments, for continuous comparison terms, the difference value It can be represented as:

[0176]

[0177] in, Indicates the first The difference values ​​of the comparison items The first in the actual road surface condition information A state quantity, The first in the expected road surface condition A state variable.

[0178] If the differences in slipperiness, foreign object coverage, and visible road marking integrity are combined into a difference vector... Then it can be expressed as:

[0179]

[0180] in, Indicates the difference in slipperiness. Indicates the difference in the degree of foreign object coverage. This indicates the difference in the completeness of the visible markings.

[0181] For example, the comparison and judgment module calculates the difference in slipperiness. Difference in foreign object coverage The difference in the completeness of the markings can be seen. The difference matrix is ​​further formed by combining the differences from three consecutive frames, and a pre-defined inconsistency judgment rule is applied: when the difference in slipperiness is greater than 0.15 and the difference in the integrity of visible road markings is greater than 0.25, and this condition persists for more than two frames, an inconsistency is identified. Since the current three consecutive frames meet the conditions, the comparison and judgment module determines that an inconsistency exists and records the inconsistency location index as "forward image region ROI-3, corresponding to 8m to 15m ahead of the road".

[0182] When an inconsistency is determined to exist, an inconsistency signal is generated, and the inconsistency signal, along with the corresponding difference value, difference matrix, and inconsistency location index, is sent to the compensation execution module.

[0183] An inconsistency signal is generated and sent to the compensation execution module along with the difference value, difference matrix, and inconsistency location index.

[0184] Specifically, this step enables the compensation execution module to obtain not only the signal of whether compensation is triggered, but also the basis and location range for triggering compensation, thereby facilitating the compensation execution module to construct compensation input feature vectors in a more targeted manner.

[0185] The comparison and judgment module generates an inconsistency signal, the content of which includes "Inconsistency type: wet slipperiness too high + road marking integrity too low", "Inconsistency level: Level 2" and "Time stamp: 10:15:23.168", and simultaneously sends the inconsistency signal, difference value, difference matrix and inconsistency location index to the compensation execution module.

[0186] Furthermore, if no inconsistency is determined, the actual road surface condition information and the expected road surface condition information are received again until an inconsistency occurs.

[0187] Reference Figure 5 The compensation execution module is specifically used for:

[0188] The system acquires the inconsistency signal, the actual road surface condition information, the expected road surface condition, and the current braking parameters of the brake-by-wire module.

[0189] Specifically, this step is equivalent to the input preparation stage for compensation decision-making. Among them, inconsistency signals indicate whether dynamic compensation needs to be triggered, actual road surface conditions and expected road surface conditions indicate the direction and degree of compensation, and current braking parameters reflect the basic state of the system before compensation.

[0190] For example, at 10:15:23.170, the compensation execution module obtains the following information: actual road surface condition information "slippery degree 0.72, foreign object coverage degree 0.18, visible marking integrity degree 0.41", expected road surface condition "slippery degree 0.48, foreign object coverage degree 0.05, visible marking integrity degree 0.78", and current braking parameters "brake force distribution parameter 0.50 / 0.50, brake pressure increase parameter 0.30MPa / cycle, anti-lock braking trigger parameter 0.18".

[0191] A compensation input feature vector is constructed based on the inconsistency signal, the actual road surface condition information, the expected road surface condition, and the current braking parameters;

[0192] The input for this step is the original input set collected by S601. The processing involves encoding and combining inconsistency signals, actual road surface condition information, expected road surface conditions, and current braking parameters in a unified format to form a compensation input feature vector. The output is the compensation input feature vector.

[0193] Specifically, the compensation input feature vector is used to simultaneously express the environmental differences that need to be compensated and the current braking baseline state, thereby facilitating the calculation of the output braking parameter adjustment by the braking parameter compensation model. This vector contains both difference information and the current parameter baseline value, thus the output result is targeted.

[0194] In some embodiments, the compensation input feature vector It can be represented as:

[0195]

[0196] in, This represents the compensation input feature vector. This indicates the encoding result of the inconsistent signal. This represents a vector containing information about the actual road surface conditions. This represents the vector indicating the expected road surface conditions. This represents the current braking parameter vector.

[0197] For example, the compensation execution module encodes the inconsistency signals of "inconsistency level 2, high slipperiness, and low pavement integrity" as follows: The actual road surface condition information is encoded as The expected road surface conditions are coded as Encode the current braking parameters as This forms a compensated input feature vector.

[0198] The compensation input feature vector is input into the braking parameter compensation model to obtain the braking parameter adjustment amount;

[0199] The compensation input feature vector is input into the braking parameter compensation model, and model inference is performed.

[0200] Specifically, the adjustment of braking parameters can include increments or decrements to brake force distribution parameters, brake pressure increase parameters, and anti-lock braking system (ABS) trigger parameters. The braking parameter compensation model determines the direction and magnitude of adjustment for each parameter based on the type and degree of inconsistency in the input features and the current braking state.

[0201] The braking parameter compensation model outputs braking parameter adjustments based on the compensation input feature vector: the adjustment for braking force distribution parameter is [-0.04, +0.04], the adjustment for braking pressure increase parameter is -0.08 MPa / cycle, and the adjustment for anti-lock braking trigger parameter is -0.04.

[0202] Based on the braking parameter adjustment amount, the brake force distribution parameter, brake pressure increase parameter, and anti-lock braking trigger parameter are constrained and calculated to obtain the target braking parameters;

[0203] Specifically, the purpose of constraint calculation is to prevent the direct summation of parameter adjustments from exceeding the allowable range. Braking force distribution parameters need to satisfy total constraints and balance constraints, braking pressure growth parameters need to satisfy the rate of change boundary, and anti-lock braking system (ABS) triggering parameters need to satisfy the safe triggering range constraint.

[0204] In some embodiments, the target braking parameters It can be represented as:

[0205]

[0206] in, Represents any parameter in the target braking parameters. This indicates the corresponding parameter in the current braking parameters. This indicates the corresponding adjustment value in the braking parameter adjustment amount. This represents the minimum constraint boundary of the parameter. This indicates the maximum constraint boundary of the parameter. Represents the boundary constraint function.

[0207] For example, the current braking parameters are "0.50 / 0.50, 0.30MPa / cycle, 0.18", and the braking parameter adjustment is "[-0.04, +0.04], -0.08MPa / cycle, -0.04". When the target parameter boundaries are "braking force distribution parameter [0.35, 0.65], braking pressure increase parameter [0.10, 0.40]MPa / cycle, and anti-lock braking system trigger parameter [0.10, 0.22]", the compensation execution module calculates the target braking parameters as "0.46 / 0.54, 0.22MPa / cycle, 0.14" after constraint calculation.

[0208] The target braking parameters are sent to the brake-by-wire module to perform the dynamic compensation.

[0209] Specifically, this step completes the control transfer between the compensation execution module and the brake-by-wire module, enabling the aforementioned difference judgment and parameter calculation to truly take effect on the vehicle braking control process.

[0210] The compensation execution module sends the target braking parameters "0.46 / 0.54, 0.22MPa / cycle, 0.14" to the brake-by-wire module, which then begins to generate electronic braking control commands for the corresponding wheels.

[0211] Reference Figure 6 The compensation execution module is also specifically used for:

[0212] After the brake-by-wire module performs the dynamic compensation according to the target braking parameters, it collects the vehicle deceleration, wheel slip ratio, vehicle attitude change and current vehicle speed.

[0213] Specifically, the aforementioned feedback data is used to characterize the actual vehicle response after the target braking parameters are applied. Vehicle deceleration characterizes the deceleration effect, wheel slip ratio characterizes the adhesion utilization between the tires and the road surface, vehicle attitude change characterizes the longitudinal and lateral attitude stability of the vehicle, and the current vehicle speed can be used to characterize the operating stage corresponding to the feedback data.

[0214] For example, after the brake-by-wire module executes the target braking parameters for 200ms, the compensation execution module collects the vehicle deceleration as 4.1m / s², the wheel slip ratio as 0.13, the vehicle attitude change as 1.8°, and the current vehicle speed as 53km / h.

[0215] The consistency of the vehicle deceleration, the wheel slip ratio, the change in vehicle posture, the current vehicle speed, the braking parameter adjustment, and the inconsistency signal is verified to obtain the compensation execution result and the compensation deviation identifier.

[0216] Specifically, the consistency check is used to determine whether the actual vehicle response after the current compensation is consistent with the road risk level described by the inconsistency signal and the expected control effect of the braking parameter adjustment. If they are consistent, it means that the dynamic compensation is effective; if they are inconsistent, it means that the current braking parameter compensation model still needs to be corrected.

[0217] In some embodiments, a consistency verification index can be constructed. :

[0218]

[0219] in, Indicates consistency verification index This represents the normalized value of the vehicle's deceleration. Indicates wheel slip ratio, This represents the normalized value indicating the change in vehicle body posture. This represents the normalized value of the current vehicle speed. , , and This represents the verification weight coefficient.

[0220] For example, the compensation execution module performs consistency checks on the vehicle deceleration of 4.1 m / s², wheel slip ratio of 0.13, body posture change of 1.8°, and current vehicle speed of 53 km / h, with the braking parameter adjustment amount "[-0.04,+0.04], -0.08 MPa / cycle, -0.04" and the inconsistency signal "high wet slip + low road marking integrity". The module determines that the compensation execution result is "effective compensation" and the compensation deviation is marked as "minor deviation".

[0221] The model parameters of the braking parameter compensation model are updated based on the compensation execution result and the compensation deviation identifier, and the updated model parameters are written into the compensation model storage area.

[0222] The model parameters of the braking parameter compensation model are updated based on the two, and the updated model parameters are written into the compensation model storage area.

[0223] Specifically, if the compensation execution result is valid and the compensation deviation is small, the current model parameters can be slightly enhanced; if the compensation execution result is invalid or the deviation is large, the model parameters can be adjusted to output more appropriate braking parameter adjustment amounts in subsequent similar but inconsistent scenarios. The compensation model storage area is used to save the latest model parameters of the braking parameter compensation model.

[0224] For example, based on the results of "effective compensation" and "slight deviation", the compensation execution module slightly updates the parameter weights related to the "high slippery" scenario in the braking parameter compensation model, and writes the updated model parameters into the compensation model storage area at 10:15:23.410.

[0225] The brake-by-wire module is specifically used for:

[0226] Receive the target braking parameters sent by the compensation execution module, and perform parameter integrity verification and boundary constraint verification on the target braking parameters;

[0227] Receive the target braking parameters and perform parameter integrity verification and boundary constraint verification respectively.

[0228] Specifically, parameter integrity verification checks for missing items, format errors, or inconsistent timestamps in the target braking parameters; boundary constraint verification checks whether the target braking parameters fall within the allowable control range. Only after passing both verifications can the brake-by-wire module proceed to the subsequent instruction generation process.

[0229] For example, after receiving the target braking parameters "0.46 / 0.54, 0.22MPa / cycle, 0.14", the brake-by-wire module first confirms that the brake force distribution parameter, brake pressure increase parameter, and anti-lock braking system (ABS) trigger parameter all exist and that their timestamps are consistent with the current control cycle; then it confirms that all of the above parameters are within the allowable boundaries. The input to this step is the target braking parameters, the intermediate processing results are integrity verification and boundary constraint verification, and the output result is the verified target braking parameters.

[0230] Based on the verified target braking parameters, electronic braking control commands are generated for the corresponding wheels.

[0231] Specifically, the electronic braking control commands for the corresponding wheels can include pressure build-up commands, pressure holding commands, or pressure release commands for the front left wheel, front right wheel, rear left wheel, and rear right wheel. The generation process needs to comprehensively consider brake force distribution parameters, brake pressure increase parameters, and anti-lock braking system (ABS) trigger parameters.

[0232] For example, the brake-by-wire module generates electronic braking control commands for the four wheels based on the verified target braking parameters. The front wheels use a gentler pressure increase curve, while the rear wheels use a smoother pressure build-up curve. The anti-lock braking trigger threshold is also lowered in sync.

[0233] The electronic braking control commands are sent to the braking execution unit to control the establishment, maintenance, or release of braking pressure on each wheel;

[0234] Specifically, the braking actuator can be an electro-hydraulic actuator or an electronic pressure regulating unit corresponding to each wheel. This step is the direct control step in which the target braking parameters actually act on the vehicle.

[0235] For example, the brake-by-wire module sends electronic braking control commands for the front left wheel, front right wheel, rear left wheel, and rear right wheel to the corresponding braking actuators, causing the four wheels to establish braking pressure according to a predetermined pressure curve over the next few control cycles. The input for this step is the electronic braking control command, the intermediate result of the processing is the wheel-end command issuance, and the output is the establishment, maintenance, or release of braking pressure for each wheel.

[0236] During the electronic braking control process, the execution status, pressure establishment status, and command execution status are fed back in real time, and the execution status is sent to the compensation execution module and the comparison judgment module.

[0237] The system provides real-time feedback on execution status, pressure establishment status, and instruction execution status, and sends the execution status to the compensation execution module and the comparison and judgment module.

[0238] Specifically, the execution status indicates whether the brake-by-wire module is executing the current target braking parameters correctly; the pressure establishment status indicates the actual establishment of braking pressure at each wheel; and the command execution status indicates whether the actuators at each wheel end are acting according to the control commands. The compensation execution module uses this feedback to update the model, and the comparison and judgment module uses this feedback to evaluate the persistence of the comparison results during the execution process.

[0239] For example, during the online braking process, the brake-by-wire module provides real-time feedback: the front wheel pressure build-up status is normal, the rear wheel pressure build-up status is normal, and the command execution delay is less than 2ms. It then sends the above execution status to the compensation execution module and the comparison judgment module.

[0240] Reference Figure 7 This application also includes a collaborative control module, which is specifically used for:

[0241] The outputs of the visual detection module, the expected data acquisition module, the comparison and judgment module, and the compensation execution module are synchronized in time to obtain a synchronization control sequence.

[0242] Specifically, since the computation delay and sampling frequency of different modules may differ, it is necessary to align the output results of each module according to a unified time base to form a synchronization control sequence that can be verified for integrity in subsequent operations.

[0243] For example, the collaborative control module synchronizes the "road surface image data collected at 10:15:23.150", the "expected road surface condition generated at 10:15:23.162", the "inconsistent signal generated at 10:15:23.168", and the "target braking parameters generated at 10:15:23.175" in time to form a synchronous control sequence under the same control cycle.

[0244] The synchronization control sequence is subjected to integrity verification to determine whether there is data loss, timing limit violation or signal conflict;

[0245] Specifically, data missing refers to the missing output of a key module in the synchronization control sequence; timing exceedance refers to the output time of a module exceeding the allowable control time limit; signal conflict refers to logical contradictions between the outputs of different modules, such as the mismatch between the target braking parameters output by the compensation execution module and the inconsistency level given by the comparison and judgment module.

[0246] For example, the collaborative control module verifies the synchronization control sequence and finds that the outputs of all modules in this cycle are present, with a maximum timing deviation of 25ms, which does not exceed the preset timing limit of 30ms. Furthermore, there is no conflict between the inconsistent signals and the target braking parameters. Therefore, it is determined that there is no missing data, timing exceedance, or signal conflict. The input to this step is the synchronization control sequence, the intermediate result is the integrity verification, and the output is the integrity verification conclusion.

[0247] A degradation control signal is generated when there is missing data, timing limit exceedance, or signal conflict.

[0248] Specifically, the degradation control signal is used to block the continued transmission of new compensation results when the multi-module link of the system is abnormal, thereby preventing abnormal data from affecting the current braking control.

[0249] For example, if the expected data acquisition module experiences timing exceedances due to cache read delays in the next control cycle, the collaborative control module generates a degradation control signal. The signal content includes "Degradation Reason: Timing Exceeds Limits," "Degradation Level: Level 1," and "Freeze Period: Current Cycle to Recovery Cycle." The input for this step is the anomaly verification result, the intermediate processing result is the degradation signal determination, and the output is the degradation control signal.

[0250] The braking parameter adjustment amount is frozen according to the degradation control signal, and the brake-by-wire module executes electronic braking control according to the most recent valid target braking parameters.

[0251] Freeze the braking parameter adjustment amount and enable the brake-by-wire module to perform electronic braking control according to the most recent valid target braking parameters.

[0252] Specifically, freezing the braking parameter adjustment means that no new compensation results will be accepted until the anomaly is resolved, and the most recent valid target braking parameter will be used as a temporary stable control benchmark. This approach avoids frequent fluctuations in braking control parameters caused by multi-module link anomalies.

[0253] For example, after a timing overrun occurs, the cooperative control module freezes the newly generated braking parameter adjustments for the current cycle and instructs the brake-by-wire module to continue executing electronic braking control according to the target braking parameters of the previous valid cycle: "0.46 / 0.54, 0.22 MPa / cycle, 0.14". The input to this step is a degraded control signal, the intermediate result of which is the freeze control, and the output is electronic braking control based on the most recent valid target braking parameters.

[0254] After the synchronization control sequence is restored to meet the integrity check, the degradation control signal is released and the dynamic compensation is restored.

[0255] Specifically, after the visual detection module, expected data acquisition module, comparison and judgment module, and compensation execution module re-form a synchronous control sequence that meets the integrity verification requirements, the cooperative control module allows the system to reuse the new braking parameter adjustment amount and the new target braking parameter to perform dynamic compensation.

[0256] For example, in a subsequent control cycle, the expected data acquisition module resumes normal response, the synchronization control sequence again meets the integrity verification requirements, the cooperative control module removes the degraded control signal, and the compensation execution module is allowed to generate new target braking parameters based on the latest inconsistency signal. The inputs to this step are the restored synchronization control sequence and the integrity verification result; the intermediate result is the removal of degraded control; and the output is the restoration of dynamic compensation.

[0257] This application also includes: the comparison and judgment module is further configured to receive the actual road surface information and the expected road surface information again when no inconsistency is detected between the actual road surface condition and the expected road surface condition, and establish corresponding comparison items according to the road surface type, degree of slipperiness, degree of foreign object coverage and degree of visible marking integrity.

[0258] During this process, the brake-by-wire module does not adjust the braking parameters until an inconsistent signal is subsequently detected.

[0259] As can be seen from the above embodiments, the present invention addresses the problem of dynamic changes in road conditions under brake-by-wire scenarios by establishing a systematic technical solution consisting of a visual detection module, an expected data acquisition module, a comparison and judgment module, a compensation execution module, a brake-by-wire module, and a cooperative control module. This solution forms a complete and well-connected chain from road image data acquisition to target braking parameter execution, from execution feedback to model update, and from normal control to degraded control recovery.

Claims

1. A vision-based adaptive compensation brake system for road conditions, characterized in that, include: Brake-by-wire module, used to perform electronic braking control of the vehicle; The visual inspection module is configured to acquire road surface image data in real time and output actual road surface condition information; The expected data acquisition module is used to acquire expected road surface conditions based on historical or model data; The comparison and judgment module is used to compare the actual road surface condition with the expected road surface condition, and generate an inconsistency signal when an inconsistency is detected. The compensation execution module, in response to the inconsistency signal, adjusts the braking parameters of the brake-by-wire module to perform dynamic compensation, thereby optimizing the vehicle's braking performance.

2. The vision-based adaptive road condition compensation brake system according to claim 1, characterized in that, The visual inspection module is specifically used for: Road surface image data of the area in front of the vehicle is collected according to a preset sampling frequency, and the collection time, vehicle position and speed information are recorded synchronously with the road surface image data. The road surface image data is subjected to distortion correction, brightness normalization, noise suppression, and field of view cropping to obtain a standardized road surface image. The standardized road surface image is subjected to texture features, water reflection features, attachment edge features, and road marking occlusion features to obtain road surface feature parameters; The actual road surface condition information is generated based on the road surface feature parameters, and the actual road surface condition information is output to the comparison and judgment module.

3. The vision-based adaptive road condition compensation brake system according to claim 1, characterized in that, The expected data acquisition module is specifically used for: Obtain historical road condition records, historical weather records, and historical braking response records corresponding to the current vehicle location, as well as the corresponding historical time period identifiers; The historical road condition records, historical weather records, and historical braking response records are time-aligned, location-mapped, and anomaly-removed to obtain the basic expected dataset. Based on the aforementioned basic expected dataset, extract the associated features corresponding to the current vehicle position, current vehicle speed, current driving direction, and current steering angle; The expected road surface condition is generated based on the associated features and output to the comparison and judgment module.

4. The vision-based adaptive road condition compensation brake system according to claim 3, characterized in that, The expected data acquisition module is also specifically used for: The historical road condition records, historical weather records, and historical braking response records in the basic expected dataset are feature-encoded to obtain the expected input feature vector; The expected input feature vector is input into the road surface condition prediction model to obtain candidate expected road surface conditions; The candidate expected road surface conditions are matched and corrected with the model data to obtain corrected candidate expected road surface conditions; The expected road surface condition is determined based on the correspondence between the corrected candidate expected road surface condition and the current vehicle position; The expected road surface condition is sent to the comparison and judgment module, and the corresponding generation time is written into the expected data cache.

5. The vision-based adaptive road condition compensation line control braking system according to claim 1, characterized in that, The comparison and judgment module is specifically used for: Receive the actual road surface condition information and the expected road surface condition, and establish corresponding comparison items according to road surface type, degree of slipperiness, degree of foreign object coverage and degree of visible marking integrity; The difference value is calculated for each comparison item and a difference matrix is ​​formed. Then, the existence of inconsistency is determined according to the preset inconsistency judgment rule and the continuous frame change threshold, and the inconsistency position index is recorded. When an inconsistency is determined to exist, an inconsistency signal is generated, and the inconsistency signal, along with the corresponding difference value, difference matrix, and inconsistency location index, is sent to the compensation execution module.

6. The vision-based adaptive road condition compensation brake system according to claim 1, characterized in that, The compensation execution module is specifically used for: The system acquires the inconsistency signal, the actual road surface condition information, the expected road surface condition, and the current braking parameters of the brake-by-wire module. A compensation input feature vector is constructed based on the inconsistency signal, the actual road surface condition information, the expected road surface condition, and the current braking parameters; The compensation input feature vector is input into the braking parameter compensation model to obtain the braking parameter adjustment amount; Based on the braking parameter adjustment amount, the brake force distribution parameter, brake pressure increase parameter, and anti-lock braking trigger parameter are constrained and calculated to obtain the target braking parameters; The target braking parameters are sent to the brake-by-wire module to perform the dynamic compensation.

7. The vision-based adaptive road condition compensation brake system according to claim 6, characterized in that, The compensation execution module is also specifically used for: After the brake-by-wire module performs the dynamic compensation according to the target braking parameters, it collects the vehicle deceleration, wheel slip ratio, vehicle attitude change and current vehicle speed. The consistency of the vehicle deceleration, the wheel slip ratio, the change in vehicle posture, the current vehicle speed, the braking parameter adjustment, and the inconsistency signal is verified to obtain the compensation execution result and the compensation deviation identifier. The model parameters of the braking parameter compensation model are updated based on the compensation execution result and the compensation deviation identifier, and the updated model parameters are written into the compensation model storage area.

8. The vision-based adaptive road condition compensation brake system according to claim 1, characterized in that, The brake-by-wire module is specifically used for: Receive the target braking parameters sent by the compensation execution module, and perform parameter integrity verification and boundary constraint verification on the target braking parameters; Based on the verified target braking parameters, electronic braking control commands are generated for the corresponding wheels. The electronic braking control commands are sent to the braking execution unit to control the establishment, maintenance, or release of braking pressure on each wheel; During the electronic braking control process, the execution status, pressure establishment status, and command execution status are fed back in real time, and the execution status is sent to the compensation execution module and the comparison judgment module.

9. The vision-based adaptive road condition compensation brake system according to claim 1, characterized in that, It also includes a collaborative control module, which is specifically used for: The outputs of the visual detection module, the expected data acquisition module, the comparison and judgment module, and the compensation execution module are synchronized in time to obtain a synchronization control sequence. The synchronization control sequence is subjected to integrity verification to determine whether there is data loss, timing limit violation or signal conflict; A degradation control signal is generated when there is missing data, timing limit exceedance, or signal conflict. The braking parameter adjustment amount is frozen according to the degradation control signal, and the brake-by-wire module executes electronic braking control according to the most recent valid target braking parameters. After the synchronization control sequence is restored to meet the integrity check, the degradation control signal is released and the dynamic compensation is restored.

Images