An automatic driving steering angle tracking control optimization method and system based on segmented proportional feedback

Abstract

The application discloses a kind of automatic driving steering angle tracking control optimization method and system based on segmented proportional feedback, it is related to steering angle tracking control technical field, including obtaining vehicle actual steering angle and target steering angle signal, error time sequence is analyzed and is divided into working section;For minimum error section, set basic proportional feedback gain, generate first gain progressive ratio in combination with speed curvature experience mapping, and form multiple sets of second gain progressive ratio control amount by random disturbance;It is applied to fine tuning compensation, collects control characteristics, and predicts fine tuning effect based on machine learning training compensation evaluation model;Finally, construct control amount-compensation evaluation diagram, filter the best gain control amount in combination with actual constraint, realize the optimization of automatic driving steering angle tracking control and response performance improvement, solve the problem that response lag, overshoot and fine tuning compensation deficiency caused by unbalanced gain setting in automatic driving steering angle tracking control.

Description

Technical Field

[0001] This invention relates to the field of steering angle tracking control technology, and more specifically, to an optimization method and system for steering angle tracking control of autonomous driving based on segmented proportional feedback. Background Technology

[0002] In existing autonomous driving technologies, segmented proportional feedback-based steering angle tracking control is a commonly used method. This method divides the steering angle error range into several working segments and sets a proportional feedback gain for each segment, thereby achieving fine-grained control of the vehicle's steering system. The segmented proportional feedback method can employ different gain strategies within different error ranges, improving control accuracy and vehicle handling stability, and is a common steering control strategy in current advanced driver assistance and autonomous driving systems.

[0003] In the aforementioned segmented proportional feedback control method, the gain increment ratio is a key parameter determining the magnitude of proportional gain variation between adjacent working segments. The gain increment ratio controls the gradual increase or decrease of the proportional gain with each error segment, thus affecting the sensitivity of the steering angle response and the system stability. In applications such as asset walls, by appropriately setting the gain increment ratio, a smooth response in small error segments and rapid correction in large error segments can be achieved, enabling the vehicle to maintain good steering tracking performance under complex road conditions.

[0004] However, both excessively high and excessively low prediction gain step ratios can lead to different drawbacks, negatively impacting the performance of autonomous driving steering angle tracking control based on piecewise proportional feedback. An excessively low step ratio may result in sluggish steering response, making it difficult for the vehicle to adjust quickly when errors are large, increasing trajectory deviation; while an excessively high step ratio may cause the controller output to be too strong, generating oscillations or overshoot, affecting handling stability.

[0005] When the gain increment ratio is too low, the proportional feedback gain change is insufficient when the vehicle enters a large error operating range, resulting in weakened fine-tuning compensation capability. In this case, the steering angle response lags, and the time it takes for the vehicle trajectory to deviate from the target path is prolonged, thus affecting overall tracking accuracy and driving safety. A low increment ratio may also prevent the control system from quickly adapting to error changes under conditions of continuous curves or complex roads, reducing handling sensitivity.

[0006] Conversely, when the gain increment ratio is too high, the proportional feedback gain rises rapidly as the vehicle changes within the error range, potentially leading to over-response in steering angle control. An excessively high increment ratio can cause increased steering torque fluctuations, increased front wheel overshoot, and vehicle oscillations or short-term instability. This high increment ratio negatively impacts the vehicle's steering tracking performance, especially at high speeds or on roads with significant curvature changes, easily reducing driving comfort and safety.

[0007] Existing technologies lack effective means for dynamic control of the gain increment ratio, resulting in drawbacks in practical applications of autonomous driving steering angle tracking control based on piecewise proportional feedback, such as response lag, large oscillation amplitude, and insufficient fine-tuning compensation. The lack of a targeted adjustment mechanism makes it difficult for the control system to achieve balanced optimization across different error ranges, thus limiting the accuracy, stability, and overall safety of autonomous driving steering tracking.

[0008] To address the above problems, this invention proposes a solution. Summary of the Invention

[0009] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide an optimization method and system for autonomous driving steering angle tracking control based on piecewise proportional feedback. By using a gain progressive ratio optimization method based on piecewise proportional feedback, the method addresses the problems of response lag, overshoot, and insufficient fine-tuning compensation caused by unbalanced gain settings in autonomous driving steering angle tracking control.

[0010] To achieve the above objectives, the present invention provides the following technical solution: An optimization method for autonomous driving steering angle tracking control based on piecewise proportional feedback includes the following steps: acquiring the actual steering angle signal and the target steering angle signal of the vehicle steering system within a certain time period, and analyzing the steering angle error time series; dividing the error range into several working segments according to the steering angle error time series; setting a basic proportional feedback gain for the smallest working segment, and generating a first proportional gain set by combining the first gain progressive ratio control quantity set based on the empirical mapping of velocity curvature; randomly perturbing the first gain progressive ratio control quantity to generate several sets of second gain progressive ratio control quantities; applying the second proportional gain set corresponding to the second gain progressive ratio control quantity to the fine-tuning compensation of the driving steering angle control, and acquiring the control characteristics during the compensation process; generating compensation evaluation values ​​by training a compensation evaluation model based on machine learning according to the control characteristics; constructing a control quantity-compensation evaluation graph based on several second gain progressive ratio control quantities and the corresponding compensation evaluation values, and selecting the optimal second gain progressive ratio control quantity based on actual demand constraints for fine-tuning compensation of the driving steering angle control.

[0011] In a preferred embodiment, the step of acquiring the actual steering angle signal and the target steering angle signal of the vehicle steering system within a certain time period and analyzing them to obtain the steering angle error time sequence specifically involves: acquiring the actual steering angle signal and the target steering angle signal of the vehicle within a certain time period, and synchronously sampling both to generate a steering angle signal sequence; performing low-pass filtering on the steering angle signal sequence to remove high-frequency interference; calculating the difference between the actual steering angle and the target steering angle point by point to obtain an initial steering angle error sequence; calculating the mean, variance, and maximum value of the error in each window using a fixed-length sliding window on the initial steering angle error sequence, and recording the sampling point index corresponding to each window; concatenating the mean, variance, and maximum value of all sliding windows in the order of sampling points to generate a complete steering angle error time sequence; labeling the corresponding vehicle speed and curvature for each sampling point in the generated steering angle error time sequence; and performing a continuity check on the generated steering angle error time sequence to obtain the final steering angle error time sequence.

[0012] In a preferred embodiment, dividing the error range into several working segments based on the steering angle error timing sequence specifically involves: extracting an absolute error value sequence based on the final steering angle error timing sequence; sorting the absolute error value sequence to obtain the minimum, maximum, and distribution range of the error; uniformly dividing the distribution range between the minimum and maximum error values ​​into several working segments according to a preset number of segments; recording the corresponding upper and lower error limits for each working segment and generating a working segment index sequence; traversing each sampling point of the steering angle error timing sequence, mapping the sampling point error to its corresponding working segment index, and generating a steering angle error timing sequence with segment markings.

[0013] In a preferred embodiment, the step of setting a basic proportional feedback gain for the minimum working segment and generating a first proportional gain set by combining a first gain progressive ratio control amount set based on the empirical mapping of velocity curvature is as follows: In the working segments, the working segment with the smallest error is selected as the basic working segment; based on the vehicle speed, yaw rate, and steering wheel speed data of the basic working segment, the gain interval with the highest similarity is found using a pre-established basic gain empirical table; within the found gain interval, the basic proportional feedback gain value is calculated by linear interpolation based on the average steering angle error change rate and the maximum residual error of the segment; the basic proportional feedback gain is used as the initial gain; a first gain progressive ratio control amount is set based on the empirical mapping of velocity curvature; the basic proportional feedback gain is combined with the first gain progressive ratio control amount to calculate the proportional feedback gain of the next working segment; the above calculation is repeated for each remaining working segment, and the proportional feedback gain of each working segment is determined by the gain of the previous segment and the corresponding progressive ratio control amount; the proportional feedback gains of each working segment are summarized in sequence to generate the first proportional gain set.

[0014] In a preferred embodiment, the first gain progressive ratio control quantity set based on the empirical mapping of speed curvature specifically involves: acquiring the corresponding vehicle speed and road curvature information point by point in the steering angle error time series marked with working segments; for each sampling point, firstly, using the basic progressive ratio constant as the starting value, adjusting the progressive ratio according to the vehicle speed: when the vehicle speed increases, the progressive ratio decreases proportionally to suppress excessive control response at high speeds; and then amplifying the progressive ratio according to the absolute value of the road curvature: when the curvature increases, the progressive ratio increases proportionally to enhance steering response sensitivity; superimposing the speed adjustment and curvature adjustment effects to obtain the preliminary progressive ratio control quantity for the sampling point; imposing upper and lower limit constraints on the preliminary progressive ratio control quantity for all sampling points, limiting the preliminary progressive ratio control quantity to a preset minimum and maximum range; averaging the preliminary progressive ratio control quantity within each working segment according to the working segment index to obtain the first gain progressive ratio control quantity for each working segment, and sequentially summarizing the segments to form the first gain progressive ratio control quantity.

[0015] In a preferred embodiment, the step of randomly perturbing the first gain increment ratio control quantity to generate several sets of second gain increment ratio control quantities specifically involves: obtaining the corresponding first gain increment ratio value for each working segment from the set of first gain increment ratio control quantities; generating a random perturbation quantity for each working segment's first gain increment ratio value according to a preset perturbation amplitude range, the perturbation quantity including positive and negative directions; superimposing the random perturbation quantity with the first gain increment ratio value to generate a set of second gain increment ratio control quantities for that working segment; repeating the above steps for each working segment to generate multiple sets of second gain increment ratio control quantities, thereby realizing the discrete random expansion of the first gain increment ratio; and summarizing the several sets of second gain increment ratio control quantities for each working segment in sequence to generate several sets of second gain increment ratio control quantities.

[0016] In a preferred embodiment, the step of applying the second proportional gain set corresponding to the second gain progressive ratio control quantity to the fine-tuning compensation of the driving steering angle control and obtaining the control characteristics during the compensation process specifically involves: selecting the corresponding second gain progressive ratio control quantity for each working segment from several sets of second gain progressive ratio control quantities; calculating the proportional feedback gain of the working segment based on the selected control quantity and the proportional feedback gain of the working segment, and inputting it into the fine-tuning compensation loop of the steering angle controller; during the fine-tuning compensation process, collecting the deviation between the actual steering angle and the target steering angle, the steering wheel torque output, and the front wheel steering angle response rate in real time for each sampling point to form the original signal data sequence of the sampling point; calculating the rate of change of the deviation over time for each sampling point based on the deviation sequence between the actual steering angle and the target steering angle; extracting the maximum value of the deviation sequence within each working segment to obtain the maximum deviation within the segment; calculating the magnitude by which the deviation exceeds the target angle to obtain the deviation overshoot; calculating the magnitude of the change of the steering wheel torque output sequence for each sampling point over time to obtain the torque change magnitude; and using the rate of change of the deviation over time, the maximum deviation within the segment, the deviation overshoot, and the torque change magnitude as control characteristics.

[0017] In a preferred embodiment, the step of generating compensation evaluation values ​​based on the control features and training a compensation evaluation model using machine learning specifically involves: pairing the control feature set of the working segment with the corresponding second gain increment ratio control quantity to form a training sample set; standardizing the training sample set; selecting a machine learning model and setting the model input as the control feature vector of the working segment, with the output being the fine-tuning compensation effect score corresponding to that working segment; inputting the standardized training sample set into the machine learning model for training, and iteratively optimizing the model parameters so that the machine learning model can predict the gain increment ratio based on the control features to adjust the quantitative evaluation value of the fine-tuning compensation effect, thereby generating compensation evaluation values; and summarizing the compensation evaluation values ​​of each working segment in order of control quantity to form a sequence of compensation evaluation values ​​for the whole vehicle fine-tuning compensation.

[0018] In a preferred embodiment, the step of constructing a control quantity-compensation evaluation diagram based on several second gain progressive ratio control quantities and corresponding compensation evaluation values, and selecting the optimal second gain progressive ratio control quantity for fine-tuning compensation of driving steering angle control in combination with actual demand constraints, specifically involves: pairing control quantities with compensation evaluation values ​​in several sets of second gain progressive ratio control quantities and their corresponding compensation evaluation values ​​for each working segment to form a control quantity-compensation evaluation vector; initializing an empty control quantity-compensation evaluation display space to map all control quantity-compensation evaluation vectors while maintaining the sequential relationship of control quantities in each working segment; and mapping each control quantity-compensation evaluation vector to its corresponding position in the display space, so that each point... It possesses a matching relationship between the combination of control quantities and the evaluation value of its fine-tuning compensation effect; for the mapped discrete point set, local interpolation processing is performed according to the working segment index to transform the discrete points into a continuous control quantity-compensation evaluation surface or curve region; in the continuous surface or curve region, combined with actual requirement constraints, including the maximum allowable steering angle error, control response speed requirements, and steering wheel torque limits, feasible regions that meet the constraints are selected; in the feasible region, the second gain progressive ratio control quantity corresponding to the region peak is taken as the optimal second gain progressive ratio control quantity; the selected optimal second gain progressive ratio control quantity is applied to the fine-tuning compensation of the driving steering angle control to achieve optimized control of the vehicle's steering performance.

[0019] A system for optimizing steering angle tracking control in autonomous driving based on piecewise proportional feedback includes an error timing module, a working segment module, a gain progressive ratio control quantity generation module, a screening group generation module, a control feature extraction module, an evaluation module, and a control module. The error timing module acquires the actual steering angle signal and the target steering angle signal of the vehicle steering system over a certain period and analyzes them to obtain the steering angle error timing. The working segment module divides the error range into several working segments based on the steering angle error timing. The gain progressive ratio control quantity generation module sets a basic proportional feedback gain for the minimum working segment and generates a first proportional gain set by combining a first gain progressive ratio control quantity set based on an empirical mapping of velocity curvature. The screening group generation module randomly perturbs the first gain progressive ratio control quantity to generate several sets of second gain progressive ratio control quantities. The control feature extraction module is used to apply the second proportional gain set corresponding to the second gain increment ratio control quantity to the fine-tuning compensation of the driving steering angle control, and to obtain the control features during the compensation process. The evaluation module is used to generate compensation evaluation values ​​based on a compensation evaluation model trained using machine learning according to control characteristics. The control module is used to construct a control quantity-compensation evaluation graph based on several second gain progressive ratio control quantities and corresponding compensation evaluation values, and to select the optimal second gain progressive ratio control quantity in combination with actual demand constraints for fine-tuning compensation of driving steering angle control.

[0020] The technical effects and advantages of the present invention regarding an optimization method and system for steering angle tracking control of autonomous driving based on piecewise proportional feedback are as follows: 1. This invention divides the steering angle error timing into several working segments, sets a proportional feedback gain for the basic working segment, generates a gain progression ratio by combining velocity curvature empirical mapping, and then forms multiple sets of candidate gain control quantities through random perturbation, thereby achieving segmented optimization and fine-tuning of steering angle control. This method can flexibly adjust the proportional feedback gain for different error segments, making the steering response more precise, reducing overshoot and steady-state errors, and thus improving the steering tracking accuracy and stability of autonomous vehicles under various driving conditions.

[0021] 2. This invention establishes a compensation evaluation model by fine-tuning the control features collected during the compensation process and combining it with machine learning. Furthermore, it selects the optimal gain control quantity by constructing a control quantity-compensation evaluation graph, thus integrating theoretical optimization with actual driving feedback. This method can intelligently select the optimal gain progression ratio while meeting the limitations of maximum steering angle error, control response speed, and steering wheel torque, significantly improving the overall vehicle steering performance and achieving high-precision, highly controllable, and safe steering tracking control for autonomous vehicles. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating an optimization method for steering angle tracking control of autonomous driving based on piecewise proportional feedback, according to the present invention.

[0023] Figure 2 This is a schematic diagram of the structure of an autonomous driving steering angle tracking control optimization system based on segmented proportional feedback according to the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0025] Example 1, Figure 1 This invention presents an optimization method for steering angle tracking control of autonomous driving based on piecewise proportional feedback, comprising the following steps: S1: Acquire the actual steering angle signal and the target steering angle signal of the vehicle steering system within a certain time period, and analyze the steering angle error timing.

[0026] In this embodiment, the acquisition of the actual steering angle signal and the target steering angle signal of the vehicle steering system over a certain period of time, and the analysis to obtain the steering angle error timing sequence, specifically involves: The actual steering angle signal and the target steering angle signal of the vehicle are acquired within a certain period of time, and the two are sampled synchronously to generate a steering angle signal sequence; The steering angle signal sequence is low-pass filtered to remove high-frequency interference; The difference between the actual steering angle and the target steering angle is calculated for each sampling point to obtain the initial steering angle error sequence; On the initial steering angle error sequence, the mean, variance and maximum value of the error in each window are calculated using a sliding window of fixed length, and the sampling point index corresponding to each window is recorded; The mean, variance, and maximum values ​​of all sliding windows are concatenated in the order of sampling points to generate a complete steering angle error time series. In the generated steering angle error time series, the corresponding vehicle speed and curvature are labeled for each sampling point; The continuity of the generated steering angle error timing sequence is checked to obtain the final steering angle error timing sequence.

[0027] In this embodiment, the actual steering angle signal refers to the sequence of steering wheel deflection angles or wheel rotation angles collected in real time by the steering wheel sensor, steering knuckle sensor, or front wheel angle sensor during vehicle operation. These signals reflect the changes in the vehicle's actual steering state over time and are the fundamental data source for subsequent error calculation and working segment division. Their inherent content includes steering response delay, sensor accuracy, and the impact of vehicle speed on the dynamic characteristics of the collected signals. Therefore, the sampled signals need to be synchronized and filtered before use to ensure the continuity and accuracy of the data.

[0028] In this embodiment, the target steering angle signal is a time-varying sequence of the ideal steering angle generated based on autonomous driving path planning, vehicle control algorithms, or the driver's desired control intentions. This signal is used to compare with the actual steering angle signal to calculate the error sequence. Its inherent content includes not only the theoretically desired steering angle value but also implicitly includes vehicle dynamic constraints, such as minimum turning radius, vehicle speed, and yaw rate limits, thus ensuring that the error sequence accurately reflects the degree to which the vehicle deviates from the expected path.

[0029] In this embodiment, the steering angle error time series is a continuous sequence generated by calculating the difference between the actual steering angle and the target steering angle at each sampling point, summarizing the mean, variance, and maximum value within a sliding window, and then arranging them in chronological order. This time series not only reflects the magnitude of the deviation but also its dynamic trend, including error fluctuations, transient overshoot, and long-term deviations. Its inherent content involves the continuity of the time series, data integrity, and error statistical characteristics, providing a foundation for subsequent working segment division and proportional feedback gain calculation.

[0030] In this embodiment, the sliding window refers to a fixed-length continuous time segment extracted from the error sequence for local statistical analysis of error characteristics. The sliding window can capture the fluctuation characteristics of the steering angle deviation within a local time range, reducing the impact of single-point noise on the error time series. Its inherent content includes the influence of window length on error smoothness, and the contribution of the mean, variance, and maximum value calculated for each window to the dynamic evaluation of the error. These statistics reflect the local deviation characteristics and provide a basis for setting the gain of proportional feedback control.

[0031] In this embodiment, vehicle speed and curvature annotation refers to adding the vehicle's current speed and road curvature information to each sampling point in the steering angle error time series, thus associating the error information with the vehicle's driving state. Speed ​​reflects the magnitude of the vehicle's kinetic energy, while curvature reflects the influence of road geometry on steering response. The underlying principle is that the vehicle's steering dynamics differ significantly under different speed and curvature conditions, and this annotation information provides a decision-making basis for subsequent speed-curvature-based gain progressive ratio mapping.

[0032] It should be noted that although this step describes in detail the acquisition, filtering, and error calculation process of the actual and target steering angles, to ensure the continuity and accuracy of the final steering angle error timing, it is necessary to detect and process any abnormal data (such as missing signals, abrupt changes, or duplicate sampling) that may occur during the sampling process. By removing abnormal data, smoothing jump values, or using adjacent window interpolation to correct missing samples, this embodiment ensures that the generated steering angle error timing can truly and completely reflect the dynamic changes of vehicle steering deviation, providing a reliable basis for subsequent working segment division and proportional feedback gain calculation, thereby avoiding misjudgment of control parameters due to abnormal sampling.

[0033] S2, based on the steering angle error timing sequence, divides the error range into several working sections. In this embodiment, dividing the error range into several working segments based on the steering angle error timing sequence specifically involves: Based on the final steering angle error timing sequence, extract the absolute value sequence of the error; Sort the sequence of absolute error values ​​to obtain the minimum, maximum, and distribution range of the error; Between the minimum and maximum error values, the distribution range is evenly divided into several working sections according to the preset number of sections; Record the corresponding upper and lower error limits for each working segment and generate a working segment index sequence; Traverse each sampling point of the steering angle error timing sequence, map the sampling point error to its corresponding working segment index, and generate a steering angle error timing sequence with segment markings.

[0034] In this embodiment, the error absolute value sequence refers to the sequence formed by taking the absolute value of the difference between the actual steering angle and the target steering angle at each sampling point in the steering angle error time series. This sequence is used to measure the magnitude of the vehicle steering deviation, eliminating the influence of the positive and negative directions of the deviation, so that subsequent working segment division can be uniformly processed based on the magnitude of the deviation.

[0035] In this embodiment, the working section refers to multiple continuous intervals formed by dividing the distribution range of the absolute value of the error according to a preset number. Each section corresponds to a certain range of deviation amplitude, used to set differentiated proportional feedback gain within different deviation ranges, thereby achieving fine control of the steering angle. The division of the working section ensures that small deviation sections can be flexibly adjusted, while large deviation sections can receive enhanced response.

[0036] It should be noted that although the process of uniformly dividing the error range into several segments is described, the actual distribution of deviations under different operating conditions may be uneven, and some segments may have sparse or excessively dense sampling points. To ensure the rationality of the division, this embodiment will count the number of sampling points in each segment after division, and, if necessary, adjust the segment boundaries or merge adjacent segments to ensure that each working segment contains sufficient error samples to support subsequent proportional gain settings and gain progression calculations.

[0037] It should be noted that the working segment index sequence refers to the sequence that labels the working segment to which each sampling point belongs. This labeling allows each sampling point to be clearly mapped to a specific segment in subsequent calculations, thereby setting the corresponding proportional feedback gain according to the characteristics of the segment and achieving the continuity and consistency of fine-grained control.

[0038] S3 sets the basic proportional feedback gain for the minimum working section, and generates the first proportional gain set by combining the first gain progression ratio control value set based on the empirical mapping of velocity curvature. In this embodiment, the step of setting a basic proportional feedback gain for the minimum working section and generating a first proportional gain set by combining a first gain progression ratio control value set based on the empirical mapping of velocity curvature is specifically as follows: Within the working sections, the section with the smallest error is selected as the basic working section; Based on the vehicle speed, yaw rate and steering wheel speed data of the basic working section, the gain interval with the highest similarity is found in combination with the pre-established basic gain experience table. Within the identified gain range, the basic proportional feedback gain value is calculated by linear interpolation based on the average rate of change of steering angle error and the maximum residual error of the segment. Use the base proportional feedback gain as the initial gain; The first gain progressive ratio control value is set based on the empirical mapping of velocity curvature; The proportional feedback gain of the next working segment is calculated by combining the basic proportional feedback gain with the first gain increment ratio control value. Repeat the above calculations for each of the remaining working segments in sequence. The proportional feedback gain of each working segment is determined by the gain of the previous segment and the corresponding progressive ratio control. The proportional feedback gains of each working segment are summarized in order to generate the first proportional gain set.

[0039] In this embodiment, the basic working segment refers to the segment with the smallest absolute value of steering angle error among all the divided working segments. This segment is chosen as the basis for the initial gain calculation because, under small deviation conditions, the controller response is relatively smooth. The gain setting of this segment can provide an initial reference for other segments, giving the progressive gain calculation a reasonable starting point and continuity.

[0040] The baseline gain experience table in this embodiment refers to a reference table formed based on a large number of vehicle handling experiments and historical data. It is used to find the gain range that is most similar to the vehicle state in the current baseline operating range. The table records the appropriate baseline proportional feedback gain range under different vehicle speeds, yaw rates, and steering wheel speeds. It can quantitatively select the most suitable initial gain to ensure that the control output is stable and has appropriate sensitivity.

[0041] In this embodiment, the first gain progression ratio control quantity refers to the gain adjustment rate when the base proportional feedback gain progresses to the next working segment based on vehicle speed and road curvature information. This progression ratio is used to control the rate of gain change between segments, enabling a smooth transition in control response from small deviation segments to large deviation segments, while avoiding over-response under high speed or high curvature conditions.

[0042] It should be noted that the calculation of the segment proportional feedback gain not only relies on the base gain, but also incorporates linear interpolation of the segment's average steering angle error change rate and maximum residual error. This process is to ensure that the gain change conforms to the vehicle's current deviation trend while avoiding abrupt changes, thus guaranteeing control smoothness between consecutive segments.

[0043] It should be noted that the logic of sequentially generating the proportional feedback gain for each working segment in this step ensures the progressiveness and stability of the gain throughout the steering control process. By combining the initial gain and the gain progression ratio, the gain of each segment references both the experience of the previous segment and the characteristics of the current segment, forming a complete first proportional gain set, which serves as the basis for subsequent fine-tuning compensation.

[0044] It should be noted that, to avoid logical loopholes, the generation of gain for each segment should be emphasized as a continuous iterative process. The gain and progression ratio of the previous working segment directly determine the gain of the next segment, rather than being set independently and randomly. This ensures the internal logical consistency of gain settings and the smoothness of control output.

[0045] S4 is the first gain progression ratio control value set based on the empirical mapping of velocity curvature.

[0046] In this embodiment, the first gain progression ratio control amount set based on the empirical mapping of velocity curvature is specifically: In the steering angle error time series with working section markings, the corresponding vehicle speed and road curvature information are obtained point by point; For each sampling point, the basic progressive ratio constant is used as the starting value, and the progressive ratio is adjusted according to the vehicle speed: when the vehicle speed increases, the progressive ratio decreases proportionally to suppress excessive control response at high speeds. The progressive ratio is amplified and adjusted according to the absolute value of the road curvature: when the curvature increases, the progressive ratio increases proportionally to enhance steering response sensitivity. By superimposing the effects of speed adjustment and curvature adjustment, the initial progressive ratio control value of the sampling point is obtained; The initial progressive ratio control value for all sampling points is subject to upper and lower limit constraints, limiting the initial progressive ratio control value to a preset minimum and maximum range. According to the working segment index, the initial progressive ratio control quantity is averaged in each working segment to obtain the first gain progressive ratio control quantity for each working segment. The segments are then summarized sequentially to form the first gain progressive ratio control quantity.

[0047] Here is a feasible example of calculating the first gain increment ratio control value: ; In the formula, Let i be the first gain increment ratio control value for sampling point i. Based on the progressive ratio constant, For vehicle speed, For road curvature, For speed adjustment ratio, This is the curvature adjustment ratio.

[0048] In this embodiment, the initial progressive ratio control value refers to the temporary gain progressive ratio value obtained at each sampling point after adjusting the speed and curvature based on the vehicle speed and road curvature information using the basic progressive ratio. This value reflects the rate of increase of the proportional feedback gain from one working segment to the next under the current vehicle operating conditions, and provides a basis for calculating the average progressive ratio of subsequent segments.

[0049] In this embodiment, the base progression ratio constant refers to the gain progression ratio used as the initial starting point under conditions of no speed or curvature adjustment. It provides a unified reference standard, ensuring a fixed starting value when adjusting at different sampling points, thereby guaranteeing the continuity and controllability of the progression ratio.

[0050] In this embodiment, the speed adjustment ratio refers to a scaling factor applied to the initial progressive ratio based on the vehicle's real-time speed. As the vehicle speed increases, this ratio decreases the progressive ratio to reduce the risk of excessive front wheel response caused by rapid gain increases at high speeds, thus ensuring steering smoothness and handling stability.

[0051] In this embodiment, the curvature adjustment ratio refers to the amplification factor applied to the initial progressive ratio based on the absolute value of the road curvature. As the curvature increases, the gain progressive ratio increases proportionally, making steering control more sensitive, thereby enhancing the vehicle's tracking accuracy and handling response when driving on curves.

[0052] In this embodiment, the first gain increment ratio control quantity refers to the set of segment gain increment ratio values ​​obtained by averaging within the working segment after completing speed and curvature adjustment and undergoing upper and lower limit constraint processing. This set serves as the control basis for the proportional feedback gain increment of each working segment, providing quantitative parameters for subsequent fine-tuning compensation.

[0053] It should be noted that during speed and curvature adjustment, attention should be paid to the inherent logical relationship between the two factors. Speed ​​adjustment and curvature adjustment not only affect the gain progression ratio individually, but also jointly determine the final initial progression ratio, ensuring that the gain progression is both smooth and in line with the vehicle's handling characteristics under conditions of high speed and large curvature or low speed and small curvature.

[0054] It should be noted that the initial gain increment ratio is subject to upper and lower limit constraints before averaging. This step ensures that the first gain increment ratio in each working segment does not exceed the controllable range, thereby preventing the risk of over- or under-response at extreme sampling points. This guarantees the continuity and safety of the entire gain increment control logic.

[0055] S5, randomly perturb the first gain increment ratio control quantity to generate several sets of second gain increment ratio control quantities.

[0056] In this embodiment, the random perturbation of the first gain increment ratio control quantity to generate several sets of second gain increment ratio control quantities specifically involves: In the first gain increment ratio control quantity set, the corresponding first gain increment ratio value is obtained for each working segment; For each working segment, a random disturbance is generated according to a preset disturbance amplitude range, and the disturbance includes both positive and negative directions. The random disturbance is superimposed on the first gain increment ratio value to generate a set of second gain increment ratio control values ​​for this working segment. Repeat the above steps for each working segment to generate multiple sets of second gain progressive ratio control quantities, thereby realizing the discrete random expansion of the first gain progressive ratio. Several sets of second gain progressive ratio control quantities for each working segment are summarized in sequence to generate several sets of second gain progressive ratio control quantities.

[0057] In this embodiment, the step of applying the second proportional gain set corresponding to the second gain increment ratio control quantity to the fine-tuning compensation of the driving steering angle control and obtaining the control characteristics during the compensation process specifically involves: In a set of several sets of second gain progressive ratio control quantities, the corresponding second gain progressive ratio control quantity is selected for each working segment; Based on the selected control quantity and the proportional feedback gain of the working section, calculate the proportional feedback gain of the working section and input it into the fine-tuning compensation loop of the steering angle controller. During the fine-tuning compensation process, the deviation between the actual steering angle and the target steering angle, the steering wheel torque output, and the front wheel steering angle response rate are collected in real time for each sampling point to form the original signal data sequence of the sampling point; Calculate the rate of change of the deviation over time for each sampling point based on the deviation sequence between the actual steering angle and the target steering angle; Extract the maximum value of the deviation sequence within each working segment to obtain the maximum deviation within the segment; Calculate the magnitude by which the deviation exceeds the target angle to obtain the deviation overshoot; The torque change amplitude is obtained by calculating the change amplitude of each sampling point over time for the steering wheel torque output sequence; The rate of change of deviation over time, the maximum deviation within the segment, the deviation overshoot, and the torque variation amplitude are used as control characteristics.

[0058] In this embodiment, the random disturbance refers to a positive or negative adjustment value randomly generated according to a preset upper and lower amplitude range, based on the first gain progression ratio control value in each working segment. This disturbance is used to introduce discrete changes while maintaining the original gain progression trend, making the control value exploratory, thereby providing diversified gain schemes for subsequent fine-tuning compensation.

[0059] In this embodiment, the second gain increment ratio control quantity refers to the set of control quantities obtained by superimposing the random disturbance quantity onto the first gain increment ratio control quantity. Each set of second gain increment ratio control quantities corresponds to a different gain adjustment combination, which can cover multiple possible control strategies, so as to evaluate the impact of different gain increments on steering response and vehicle handling during fine-tuning compensation.

[0060] In this embodiment, fine-tuning compensation refers to the process of inputting the proportional feedback gain corresponding to the second gain increment ratio control quantity into the steering angle controller, and then using the controller output to apply it to the actual steering angle of the vehicle, thereby making a fine adjustment to the target steering angle. This process involves real-time acquisition of vehicle status data, including the actual steering angle, target steering angle deviation, steering wheel torque output, and front wheel steering angle response rate, to characterize the control effect.

[0061] In this embodiment, the rate of change of deviation over time refers to the rate at which the deviation between the actual steering angle and the target steering angle at each sampling point changes over time during the fine-tuning compensation process. This characteristic reflects the vehicle's instantaneous response speed and dynamic tracking capability under progressive gain adjustment, and is a key indicator for evaluating control effectiveness.

[0062] In this embodiment, the maximum deviation within a segment refers to the maximum deviation between the actual steering angle and the target steering angle within each operating segment. This feature is used to quantify the limit of control deviation within a segment and to assess whether gain increments cause overshoot or undershoot response.

[0063] In this embodiment, the deviation overshoot refers to the magnitude by which the deviation exceeds the target steering angle. This feature is used to measure the risk of over-response control and to help identify instability problems that may result from excessively high gain increments.

[0064] In this embodiment, the torque variation refers to the change in steering wheel torque output over time during fine-tuning compensation. This characteristic reflects the dynamic load of the controller output, has a significant impact on driving feel and handling smoothness, and is directly related to the gain progression ratio.

[0065] It should be noted that during random disturbances and fine-tuning compensation, the control quantities in different operating segments should be independent but sequentially continuous. Random disturbances ensure diverse exploration of gain progression strategies, while fine-tuning compensation quantifies the impact of these changes on vehicle response by collecting control characteristics, forming analyzable data support, thereby ensuring the scientific rigor and verifiability of gain optimization.

[0066] It should be noted that the extraction of control features is based entirely on the data sequence collected in real time at the sampling points during the fine-tuning compensation process, rather than on preset or abstract indicators. The deviation change rate, maximum deviation, overshoot, and torque change amplitude are all obtained by processing the actual collected signals, thus ensuring that subsequent evaluation and optimization based on these features have empirical evidence.

[0067] S6, Based on the control characteristics, a compensation evaluation model is trained using machine learning to generate compensation evaluation values.

[0068] In this embodiment, the step of generating compensation evaluation values ​​based on the control features and training a compensation evaluation model using machine learning specifically involves: The control feature set of the working section is paired with the corresponding second gain progressive ratio control quantity to form a training sample set; Standardize the training sample set; Select a machine learning model, set the model input to the control feature vector of the working segment, and the output to the fine-tuning compensation effect score corresponding to the working segment; The standardized training sample set is input into the machine learning model for training. By iteratively optimizing the model parameters, the machine learning model can adjust the quantitative evaluation value of the fine-tuning compensation effect based on the gain increment ratio predicted by the control features, and generate the compensation evaluation value. The compensation assessment values ​​for each working section are summarized in order of control quantity to form a sequence of compensation assessment values ​​for whole vehicle fine-tuning compensation.

[0069] In this embodiment, the control feature set for the working section refers to the set of control feature vectors extracted for each working section during the fine-tuning compensation process. These vectors include the rate of change of deviation over time, the maximum deviation within the section, the deviation overshoot, and the torque variation amplitude. This feature set comprehensively reflects the control response of each working section, providing input information for the quantitative evaluation of the subsequent gain increment ratio adjustment effect.

[0070] In this embodiment, the second gain increment ratio control quantity refers to the set of gain control quantities generated based on the aforementioned random perturbation, with each set of control quantities corresponding to a different fine-tuning strategy. By pairing the control features with the second gain increment ratio control quantities, a training sample set can be formed for training the machine learning model, enabling the model to understand the impact of different gain increments on the fine-tuning compensation effect.

[0071] In this embodiment, the training sample set standardization process refers to scaling the control feature vector and the corresponding second gain increment ratio control quantity for each working segment, ensuring that the features and control quantities are within the same order of magnitude. Standardization ensures the stability and convergence of the machine learning model training, avoiding learning bias caused by numerical differences.

[0072] In this embodiment, the fine-tuning compensation effect score refers to the score obtained by quantitatively evaluating the steering angle control effect based on the control characteristics after applying the second gain progressive ratio control quantity to each working segment. The score comprehensively reflects the accuracy, stability, and handling smoothness of the vehicle's steering response, and is used to guide model learning and subsequent control quantity optimization.

[0073] In this embodiment, the machine learning model refers to a computational model used to establish the mapping relationship between control features and fine-tuning compensation effect scores. The model takes a control feature vector as input and outputs the corresponding compensation evaluation value, realizing the impact of a feature-predicted gain progression strategy on the fine-tuning compensation effect, thereby achieving intelligent optimization of the vehicle steering control.

[0074] It should be noted that the establishment of the training sample set must strictly rely on the actual data collected during the fine-tuning compensation process to ensure that the relationships learned by the model are based on real vehicle responses, rather than theoretical or preset values. The pairing of control characteristics and incremental control quantities in each working segment is continuous and consistent to ensure that the evaluation value sequence can accurately reflect the changing trend of the entire vehicle steering behavior.

[0075] It should be noted that the process of generating the compensation assessment value involves first collecting actual control characteristics and then making predictions using a trained machine learning model, without involving any pre-assumed control effects.

[0076] S7. Based on several second-gain progressive ratio control quantities and corresponding compensation evaluation values, a control quantity-compensation evaluation graph is constructed. The optimal second-gain progressive ratio control quantity is then selected based on actual demand constraints and applied to fine-tuning compensation of the driving steering angle control. In this embodiment, the construction of a control quantity-compensation evaluation graph based on several second gain progressive ratio control quantities and corresponding compensation evaluation values, and the selection of the optimal second gain progressive ratio control quantity for fine-tuning compensation of driving steering angle control in combination with actual demand constraints, specifically involves: In several sets of second gain progressive ratio control quantities and their corresponding compensation evaluation values, the control quantity and the compensation evaluation value are paired in each working segment to form a control quantity-compensation evaluation vector. Initialize an empty control quantity-compensation evaluation display space to map all control quantity-compensation evaluation vectors and maintain the order relationship of control quantities in each working segment; Each control quantity-compensation evaluation vector is mapped to the corresponding position in the display space, so that each point has a matching relationship between the combination of control quantities and its fine-tuning compensation effect evaluation value; For the set of discrete points mapped, local interpolation is performed according to the working segment index to transform the discrete points into a continuous control quantity-compensation evaluation surface or curve region. In continuous curved or curved regions, feasible regions that meet the constraints are selected by combining actual requirements and constraints, including maximum permissible steering angle error, control response speed requirements, and steering wheel torque limits. Within the feasible region, the second gain increment ratio control value corresponding to the regional peak value is taken as the optimal second gain increment ratio control value. The selected optimal second gain progressive ratio control quantity is applied to the fine-tuning compensation of the driving steering angle control to achieve optimized control of the vehicle's steering performance.

[0077] In this embodiment, the control quantity-compensation evaluation vector refers to the vector formed by pairing the second gain increment ratio control quantity of each working segment with its corresponding fine-tuning compensation effect evaluation value. This vector can accurately reflect the fine-tuning compensation performance of the vehicle steering system in that segment under a specific gain increment strategy, providing basic data for the subsequent construction of the control quantity-compensation evaluation map.

[0078] In this embodiment, the control quantity-compensation evaluation display space refers to a multi-dimensional display area used to visualize and map all control quantity-compensation evaluation vectors. The display space ensures that the order of control quantities in each working segment remains unchanged, and each mapping point represents a specific gain combination and its corresponding fine-tuning compensation effect. Through the spatial relationship, the impact of different control strategies on the vehicle's steering response and stability can be intuitively observed.

[0079] In this embodiment, the local interpolation process refers to smoothing the discrete control quantity-compensation evaluation point set mapped to the display space based on the working segment index and the relationship between adjacent points, transforming the discrete points into a continuous surface or curve region. This continuous processing can reveal the trend relationship between the progressive change in gain and the fine-tuning compensation effect, making the feasible control quantity region more continuous and operable.

[0080] In this embodiment, the practical requirement constraint refers to the vehicle performance limitations considered when selecting the optimal second gain progression ratio control quantity, including the maximum permissible steering angle error of the entire vehicle, control response speed requirements, and steering wheel torque output range. The practical requirement constraint ensures that the selected gain control quantity is optimized while meeting the requirements of vehicle handling safety, comfort, and responsiveness.

[0081] In this embodiment, the feasible region refers to the sub-region within the continuous control quantity-compensation evaluation surface or curve region that simultaneously satisfies the aforementioned practical requirement constraints. This region includes all gain-progressive control quantities that can achieve the expected fine-tuning compensation effect within the constraints, providing a candidate set for determining the optimal control quantity.

[0082] In this embodiment, the optimal second gain progression ratio control quantity refers to the control quantity that, within the feasible region, is determined by analyzing the gain combination corresponding to the peak value of the compensation evaluation value to achieve optimal fine-tuning compensation for the vehicle's steering performance. The optimal control quantity can balance the steering response speed and handling stability in each working segment, thereby optimizing the overall vehicle steering control effect.

[0083] It should be noted that the construction of the control quantity-compensation evaluation chart strictly relies on the control characteristics and the second gain progressive ratio control quantity obtained in the aforementioned fine-tuning compensation process, avoiding the use of assumptions or theoretical values ​​to replace actual collected data, thereby ensuring that the selected control quantity has the feasibility and actual effect.

[0084] It should be noted that when selecting the optimal second gain progressive ratio control quantity, continuity, feasibility, and fine-tuning compensation performance must be considered simultaneously to prevent the logical loophole of local optimal gain combination causing suboptimal overall vehicle steering performance, thereby ensuring that the final applied control quantity can achieve the overall optimization goal of vehicle fine-tuning compensation.

[0085] Example 2, Figure 2 This invention presents an optimized steering angle tracking control system for autonomous driving based on piecewise proportional feedback, comprising an error timing module, a working segment module, a gain progressive ratio control quantity generation module, a screening group generation module, a control feature extraction module, an evaluation module, and a control module. The error timing module acquires the actual steering angle signal and the target steering angle signal of the vehicle steering system over a certain period and analyzes them to obtain the steering angle error timing. The working segment module divides the error range into several working segments based on the steering angle error timing. The gain progressive ratio control quantity generation module sets a basic proportional feedback gain for the minimum working segment and combines it with a first gain progressive ratio control quantity set based on an empirical mapping of velocity curvature. The system generates a first proportional gain set; a filtering group generation module, used to randomly perturb the first gain progressive ratio control quantity to generate several sets of second gain progressive ratio control quantities; a control feature extraction module, used to apply the second proportional gain set corresponding to the second gain progressive ratio control quantity to the fine-tuning compensation of the driving steering angle control, and to obtain the control features during the compensation process; an evaluation module, used to generate compensation evaluation values ​​based on the control features and a machine learning-trained compensation evaluation model; and a control module, used to construct a control quantity-compensation evaluation graph based on several second gain progressive ratio control quantities and their corresponding compensation evaluation values, and to select the optimal second gain progressive ratio control quantity for fine-tuning compensation of the driving steering angle control in combination with actual requirement constraints.

[0086] In the embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0087] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0088] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0089] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0090] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0091] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0092] In the embodiments provided in this disclosure, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0093] It should be noted that, in this disclosure, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element limited by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0094] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. An optimization method for steering angle tracking control of autonomous driving based on piecewise proportional feedback, characterized in that, Includes the following steps: The actual steering angle signal and the target steering angle signal of the vehicle steering system are acquired within a certain time period, and the steering angle error timing is analyzed. Based on the steering angle error timing sequence, the error range is divided into several working segments; A basic proportional feedback gain is set for the minimum working section, and a first proportional gain set is generated by combining the first gain progressive ratio control value set based on the empirical mapping of velocity curvature. The first gain increment ratio control quantity is randomly perturbed to generate several sets of second gain increment ratio control quantities. The second proportional gain set corresponding to the second gain increment ratio control quantity is applied to the fine-tuning compensation of the driving steering angle control, and the control characteristics during the compensation process are obtained. Based on the control characteristics, a compensation evaluation model is trained using machine learning to generate compensation evaluation values. A control quantity-compensation evaluation diagram is constructed based on several second-gain progressive ratio control quantities and corresponding compensation evaluation values. The optimal second-gain progressive ratio control quantity is then selected based on actual demand constraints and applied to the fine-tuning compensation of the driving steering angle control.

2. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 1, characterized in that, The process of acquiring the actual steering angle signal and the target steering angle signal of the vehicle steering system over a certain period of time, and analyzing the steering angle error time series, specifically involves: The actual steering angle signal and the target steering angle signal of the vehicle are acquired within a certain period of time, and the two are sampled synchronously to generate a steering angle signal sequence; The steering angle signal sequence is low-pass filtered to remove high-frequency interference; The difference between the actual steering angle and the target steering angle is calculated for each sampling point to obtain the initial steering angle error sequence; On the initial steering angle error sequence, the mean, variance and maximum value of the error in each window are calculated using a sliding window of fixed length, and the sampling point index corresponding to each window is recorded; The mean, variance, and maximum values ​​of all sliding windows are concatenated in the order of sampling points to generate a complete steering angle error time series. In the generated steering angle error time series, the corresponding vehicle speed and curvature are labeled for each sampling point; The continuity of the generated steering angle error timing sequence is checked to obtain the final steering angle error timing sequence.

3. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 2, characterized in that, The error range is divided into several working segments based on the steering angle error timing sequence, specifically as follows: Based on the final steering angle error timing sequence, extract the absolute value sequence of the error; Sort the sequence of absolute error values ​​to obtain the minimum, maximum, and distribution range of the error; Between the minimum and maximum error values, the distribution range is evenly divided into several working sections according to the preset number of sections; Record the corresponding upper and lower error limits for each working segment and generate a working segment index sequence; Traverse each sampling point of the steering angle error timing sequence, map the sampling point error to its corresponding working segment index, and generate a steering angle error timing sequence with segment markings.

4. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 3, characterized in that, The first proportional gain set is generated by setting a basic proportional feedback gain for the minimum working section and combining it with a first gain progression ratio control value set based on the empirical mapping of velocity curvature. Specifically: Within the working sections, the section with the smallest error is selected as the basic working section; Based on the vehicle speed, yaw rate and steering wheel speed data of the basic working section, the gain interval with the highest similarity is found in combination with the pre-established basic gain experience table. Within the identified gain range, the basic proportional feedback gain value is calculated by linear interpolation based on the average rate of change of steering angle error and the maximum residual error of the segment. Use the base proportional feedback gain as the initial gain; The first gain progressive ratio control value is set based on the empirical mapping of velocity curvature; The proportional feedback gain of the next working segment is calculated by combining the basic proportional feedback gain with the first gain increment ratio control value. Repeat the above calculations for each of the remaining working segments in sequence. The proportional feedback gain of each working segment is determined by the gain of the previous segment and the corresponding progressive ratio control. The proportional feedback gains of each working segment are summarized in order to generate the first proportional gain set.

5. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 4, characterized in that, The first gain progression ratio control value set based on the empirical mapping of velocity curvature is specifically as follows: In the steering angle error time series with working section markings, the corresponding vehicle speed and road curvature information are obtained point by point; For each sampling point, the basic progressive ratio constant is used as the starting value, and the progressive ratio is adjusted according to the vehicle speed: when the vehicle speed increases, the progressive ratio decreases proportionally to suppress excessive control response at high speeds. The progressive ratio is amplified and adjusted according to the absolute value of the road curvature: when the curvature increases, the progressive ratio increases proportionally to enhance steering response sensitivity. By superimposing the effects of speed adjustment and curvature adjustment, the initial progressive ratio control value of the sampling point is obtained; The initial progressive ratio control value for all sampling points is subject to upper and lower limit constraints, limiting the initial progressive ratio control value to a preset minimum and maximum range. According to the working segment index, the initial progressive ratio control quantity is averaged in each working segment to obtain the first gain progressive ratio control quantity for each working segment. The segments are then summarized sequentially to form the first gain progressive ratio control quantity.

6. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 5, characterized in that, The process of randomly perturbing the first gain increment ratio control quantity to generate several sets of second gain increment ratio control quantities is as follows: In the first gain increment ratio control quantity set, the corresponding first gain increment ratio value is obtained for each working segment; For each working segment, a random disturbance is generated according to a preset disturbance amplitude range, and the disturbance includes both positive and negative directions. The random disturbance is superimposed on the first gain increment ratio value to generate a set of second gain increment ratio control values ​​for this working segment. Repeat the above steps for each working segment to generate multiple sets of second gain progressive ratio control quantities, thereby realizing the discrete random expansion of the first gain progressive ratio. Several sets of second gain progressive ratio control quantities for each working segment are summarized in sequence to generate several sets of second gain progressive ratio control quantities.

7. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 6, characterized in that, The step of applying the second proportional gain set corresponding to the second gain increment ratio control quantity to the fine-tuning compensation of the driving steering angle control, and obtaining the control characteristics during the compensation process, specifically involves: In a set of several sets of second gain progressive ratio control quantities, the corresponding second gain progressive ratio control quantity is selected for each working segment; Based on the selected control quantity and the proportional feedback gain of the working section, calculate the proportional feedback gain of the working section and input it into the fine-tuning compensation loop of the steering angle controller. During the fine-tuning compensation process, the deviation between the actual steering angle and the target steering angle, the steering wheel torque output, and the front wheel steering angle response rate are collected in real time for each sampling point to form the original signal data sequence of the sampling point; Calculate the rate of change of the deviation over time for each sampling point based on the deviation sequence between the actual steering angle and the target steering angle; Extract the maximum value of the deviation sequence within each working segment to obtain the maximum deviation within the segment; Calculate the magnitude by which the deviation exceeds the target angle to obtain the deviation overshoot; The torque change amplitude is obtained by calculating the change amplitude of each sampling point over time for the steering wheel torque output sequence; The rate of change of deviation over time, the maximum deviation within the segment, the deviation overshoot, and the torque variation amplitude are used as control characteristics.

8. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 7, characterized in that, The process of generating compensation evaluation values ​​based on a machine learning-trained compensation evaluation model using control features specifically involves: The control feature set of the working section is paired with the corresponding second gain progressive ratio control quantity to form a training sample set; Standardize the training sample set; Select a machine learning model, set the model input to the control feature vector of the working segment, and the output to the fine-tuning compensation effect score corresponding to the working segment; The standardized training sample set is input into the machine learning model for training. By iteratively optimizing the model parameters, the machine learning model can adjust the quantitative evaluation value of the fine-tuning compensation effect based on the gain increment ratio predicted by the control features, and generate the compensation evaluation value. The compensation assessment values ​​for each working section are summarized in order of control quantity to form a sequence of compensation assessment values ​​for whole vehicle fine-tuning compensation.

9. The method for optimizing steering angle tracking control of autonomous driving based on piecewise proportional feedback according to claim 8, characterized in that, The process involves constructing a control quantity-compensation evaluation graph based on several second-gain progressive ratio control quantities and corresponding compensation evaluation values. The optimal second-gain progressive ratio control quantity is then selected based on actual demand constraints and applied to the fine-tuning compensation of the driving steering angle control. Specifically: In several sets of second gain progressive ratio control quantities and their corresponding compensation evaluation values, the control quantity and the compensation evaluation value are paired in each working segment to form a control quantity-compensation evaluation vector. Initialize an empty control quantity-compensation evaluation display space to map all control quantity-compensation evaluation vectors and maintain the order relationship of control quantities in each working segment; Each control quantity-compensation evaluation vector is mapped to the corresponding position in the display space, so that each point has a matching relationship between the combination of control quantities and its fine-tuning compensation effect evaluation value; For the set of discrete points mapped, local interpolation is performed according to the working segment index to transform the discrete points into a continuous control quantity-compensation evaluation surface or curve region. In continuous curved or curved regions, feasible regions that meet the constraints are selected by combining actual requirements and constraints, including maximum permissible steering angle error, control response speed requirements, and steering wheel torque limits. Within the feasible region, the second gain increment ratio control value corresponding to the regional peak value is taken as the optimal second gain increment ratio control value. The selected optimal second gain progressive ratio control quantity is applied to the fine-tuning compensation of the driving steering angle control to achieve optimized control of the vehicle's steering performance.

10. A system using the automatic driving steering angle tracking control optimization method based on piecewise proportional feedback as described in any one of claims 1-9, characterized in that, It includes an error timing module, a working segment module, a gain increment ratio control quantity generation module, a screening group generation module, a control feature extraction module, an evaluation module, and a control module; The error timing module is used to acquire the actual steering angle signal and the target steering angle signal of the vehicle steering system within a certain time period, and analyze them to obtain the steering angle error timing. The working section module is used to divide the error range into several working sections according to the steering angle error timing. The gain progression ratio control quantity generation module is used to set the basic proportional feedback gain for the minimum working section, and generate the first proportional gain set by combining the first gain progression ratio control quantity set based on the empirical mapping of velocity curvature. The filter group generation module is used to randomly perturb the first gain increment ratio control quantity to generate several groups of second gain increment ratio control quantities. The control feature extraction module is used to apply the second proportional gain set corresponding to the second gain increment ratio control quantity to the fine-tuning compensation of the driving steering angle control, and to obtain the control features during the compensation process. The evaluation module is used to generate compensation evaluation values ​​based on a compensation evaluation model trained using machine learning according to control characteristics. The control module is used to construct a control quantity-compensation evaluation graph based on several second gain progressive ratio control quantities and corresponding compensation evaluation values, and to select the optimal second gain progressive ratio control quantity in combination with actual demand constraints for fine-tuning compensation of driving steering angle control.

Images