Intelligent automobile obstacle avoidance control method and system based on driver behavior habits

Abstract

The invention provides an intelligent automobile obstacle avoidance control method and system based on driver behavior habits. The method comprises the steps that initialization operation is conducted on a sensing device of an intelligent automobile; key parameters are collected and monitored through a sensing device; performing sliding window division on the key parameters, and adjusting the neighborhood radius based on dynamic indexes in a sliding window; a driving scene is determined according to the real-time key parameters, and corresponding weights are given to all the sensing devices based on the data features; building an obstacle avoidance path evaluation function according to the automobile and road safety indexes to determine an obstacle avoidance path evaluation function value; on the basis of the obstacle avoidance path evaluation function value, a planned path conforming to the driving behavior of the driver is calculated through a # imgabs0 # algorithm; and executing the calculated planning path through a feedback module. According to the method, the data dynamics can be adjusted in real time, the situation that the radius is too large or too small due to the fixed neighborhood radius is avoided, the weight of each dimension can be adjusted according to the driving scene, and the importance of the dimensions can be automatically calculated through statistical indexes.

Description

Technical Field

[0001] The present invention relates to the technical field of automotive intelligent control systems, and particularly to an intelligent vehicle obstacle avoidance control method and system based on driver behavior habits. Background Art

[0002] The intelligent vehicle obstacle avoidance control system is a technical system integrating core modules such as environmental perception, real-time decision-making, path planning, and motion control, aiming to endow the vehicle with the ability to autonomously avoid obstacles in complex environments. The vehicle obstacle avoidance system is crucial for ensuring driving safety and is a key link in avoiding collisions and reducing accidents.

[0003] However, existing obstacle avoidance systems generally adopt a general mode in strategy formulation and do not fully consider the individual characteristics of different drivers. These systems usually operate based on preset fixed parameters: setting a unified safety distance threshold, adopting a standardized speed adjustment curve, and executing a fixed steering avoidance logic. This rigid processing method ignores the significant differences among different drivers in driving styles (such as aggressive / conservative), reaction times, steering preferences, etc. Summary of the Invention Aiming at the deficiencies of the prior art, the purpose of the present invention is to provide an intelligent vehicle obstacle avoidance control method based on driver behavior habits, aiming to solve the technical problems mentioned in the background art.

[0004] To achieve the above purpose, the present invention is realized through the following technical solutions: An intelligent vehicle obstacle avoidance control method based on driver behavior habits, comprising the following steps: Performing an initialization operation on each sensor of the intelligent vehicle; Real-time collecting and monitoring key parameters during driving through each sensor; Dividing a sliding window for the key parameters and adjusting the neighborhood radius within the sliding window based on the dynamic index within the sliding window; Determining the driving scenario according to the key parameters and assigning corresponding weights to each sensor based on the data characteristics of the key parameters; Based on the assigned weights and constructing an obstacle avoidance path evaluation function according to vehicle and road safety indicators to determine the obstacle avoidance path evaluation function value; Based on the obstacle avoidance path evaluation function value, calculating a planned path that conforms to the driver's driving behavior based on the A-Star Algorithm; Executing the calculated planned path through the feedback module.

[0005] According to one aspect of the above technical solution, the sensor device includes an accelerator pedal sensor, a brake pedal sensor, a steering wheel sensor, an in-vehicle camera, an environmental perception camera, a vehicle state sensor, a millimeter wave radar, and a lidar, and the initialization operation is to perform a self-check on the sensor device.

[0006] Before the step of dividing the key parameters into sliding windows and adjusting the neighborhood radius based on the dynamic index within the sliding window according to the above technical solution, the following steps are further included: Use the mean filtering algorithm to filter and smooth the data of the key parameters; ; Wherein, represents the original data at time t, represents the filtered data of, G is the number of data points; Perform zero correction and gain correction on the data of the key parameters after smoothing processing; ; ; Wherein, represents the calibration value after zero correction and gain correction, represents the original value, c represents the zero offset, and e is the gain coefficient; Perform normalization processing on the data of the key parameters after zero correction and gain correction; ; Wherein, represents the normalized value, represents the minimum value in the data, represents the maximum value in the data.

[0007] According to one aspect of the above technical solution, the step of dividing the key parameters into sliding windows and adjusting the neighborhood radius based on the dynamic index within the sliding window specifically includes: Divide the key parameters into multiple sliding windows with a fixed length according to a continuous time series; ; Wherein, X k represents the data range of the kth sliding window, S represents the sliding step, W represents the window size; Calculate the dynamic index of the sliding window; ; ; Wherein, denotes the variance of the data within the k-th sliding window, denotes the mean of the data within the k-th sliding window, denotes the i-th data point under this sliding window, denotes the mean of the multi-dimensional range within the k-th sliding window, d denotes the data dimension coefficient, denotes within the k-th sliding window the l data set of the denotes within the k-th sliding window the l maximum value of the denotes within the k-th sliding window the l minimum value of the linearly adjust the neighborhood radius within the said sliding window based on the variance index and the range index; ; ; wherein, denotes the dynamic neighborhood radius of the k-th sliding window, denotes the basic neighborhood radius, denotes the variance scaling coefficient, denotes the global variance, denotes the minimum protection radius, denotes the weight coefficient of the range; smooth the linearly adjusted neighborhood radius; ; denotes the neighborhood radius after smoothing, denotes the neighborhood radius after smoothing in the previous sliding window, denotes the smoothing factor.

[0008] According to one aspect of the above technical solution, determining the driving scenario based on the real-time key parameters and assigning corresponding weights to each of the sensor devices based on data characteristics specifically includes: Determine the driving scenario according to the real-time key parameters, where the driving scenario includes a highway scenario, a congestion scenario, and a curve scenario; When the mean speed and the speed variance , determine that the driving scenario is a highway scenario; When the braking frequency , determine that the driving scenario is a congestion scenario, where γ represents the number of braking times, denotes the sliding window duration; When the angular change rate of the steering wheel , determine that the driving scenario is a curve scenario; Calculate the variance weight based on the variance within the sliding window; ; Among them, represents the variance weight of the l th dimension, represents the variance of the data of the l th dimension within the kth sliding window, represents the sum of variances of all dimensions within the sliding window, l and j represents the dimension index; Calculate the entropy weight within the sliding window based on the information entropy algorithm; ; Among them, represents the information entropy of the l th dimension, , represents the probability of the l th dimension coefficient in the kth sliding window, represents the entropy weight of the ith dimension, B represents the total number of sliding windows, represents the information entropy of the jth dimension; Define the scenario rule weight based on the determined driving scenario; Fuse and normalize the variance weight, the entropy weight, and the scenario rule weight to obtain the comprehensive weight; ; Among them, w l represents the comprehensive weight, and represent the fusion coefficients, , ∈[0,1], represents the scenario rule weight of the l th dimension; Perform smoothing processing on the comprehensive weight to obtain the final weight; ; Among them, represents the final weight of the l th dimension; Calculate the similarity between any two data points in the sliding window based on the final weight, and perform clustering on the data points using the DBSCAN algorithm based on the smoothed neighborhood radius; ; Among them, m and n represent two data points respectively, d(m, n)Indicates the similarity between two data points m and n.

[0009] According to one aspect of the above technical solution, the expression of the obstacle avoidance path evaluation function is as follows: ; Wherein, , , , All represent weight coefficients, Represents the path length, Represents the safety distance, Represents the obstacle avoidance speed, Represents the change in steering angle.

[0010] According to one aspect of the above technical solution, based on the value of the obstacle avoidance path evaluation function, a planned path that conforms to the driver's driving behavior is calculated through the A-Star Algorithm, specifically including: Introduce the A-Star Algorithm, and weight the value of the obstacle avoidance path evaluation function through a heuristic function; ; Wherein, H p Represents the weighted value of the obstacle avoidance path evaluation function; Determine the driving habit according to the value of the obstacle avoidance path evaluation function; When > 1, it is determined that the driving habit is aggressive. When = 1, it is determined that the driving habit is stable. When < 1, it is determined that the driving habit is conservative; Based on the driving habit and using the A-Star Algorithm, calculate a planned path that conforms to the driver's driving behavior; Real-time monitor the relative position, speed of the intelligent vehicle and the obstacle, and the driver's operation behavior, and record the real-time value of the obstacle avoidance path evaluation function; ; Wherein, Represents the change amount of the obstacle avoidance path evaluation function value, Represents the evaluation function value of the current path, Represents the evaluation function value of the planned path; If > the preset threshold , then recalculate the planned path.

[0011] The present invention also provides an intelligent vehicle obstacle avoidance control system based on the driver's behavior habit, including: Initialization module: used to perform initialization operations on various sensor components of the intelligent vehicle; The sensor components include an accelerator pedal sensor, a brake pedal sensor, a steering wheel sensor, an in-vehicle camera, an environmental perception camera, a vehicle status sensor, a millimeter wave radar, and a lidar. The initialization operation is to perform self-check on the sensor components; Acquisition module: used to collect and monitor key parameters during driving in real time through various sensor components; Preprocessing module: used to perform filtering and smoothing processing on the data of the key parameters by using the mean filtering algorithm; ; Among them, represents the original data at time t, represents the filtered data of, G is the number of data points; Perform zero-point correction and gain correction on the data of the key parameters after smoothing processing; ; ; Among them, represents the calibration value after zero-point correction and gain correction, represents the original value, c represents the zero-point offset, and e is the gain coefficient; Perform normalization processing on the data of the key parameters after zero-point correction and gain correction; ; Among them, represents the normalized value, represents the minimum value in the data, represents the maximum value in the data; Adjustment module: used to divide the key parameters into sliding windows and adjust the neighborhood radius within the sliding window based on the dynamic index within the sliding window; The adjustment module is specifically used for: dividing the key parameters into multiple sliding windows with a fixed length according to a continuous time series; ; Among them, X k represents the data range of the kth sliding window, S represents the sliding step, W represents the window size; Calculate the dynamic index of the sliding window; ; ; Among them, represents the variance of the data within the k-th sliding window, represents the mean of the data within the k-th sliding window, represents the i-th data point under this sliding window, represents the mean of the multi-dimensional range within the k-th sliding window, d represents the data dimension coefficient, represents within the k-th sliding window the l data set of the represents within the k-th sliding window the l maximum value of the represents within the k-th sliding window the l minimum value of the Based on the variance index and the range index, linearly adjust the neighborhood radius within the said sliding window; ; ; wherein, represents the dynamic neighborhood radius of the k-th sliding window, represents the basic neighborhood radius, represents the variance scaling coefficient, represents the global variance, represents the minimum protection radius, represents the weight coefficient of the range; Perform smoothing processing on the linearly adjusted neighborhood radius; ; represents the neighborhood radius after smoothing processing, represents the neighborhood radius after smoothing processing in the previous sliding window, represents the smoothing factor; Weight assignment module: used to determine the driving scenario according to the said key parameters, and assign corresponding weights to each of the said sensor devices based on the data characteristics of the said key parameters; The said weight assignment module is specifically used for: determining the driving scenario according to the real-time said key parameters, and the driving scenario includes a high-speed scenario, a congestion scenario and a curve scenario; When the mean speed and the speed variance , determine that the driving scenario is a high-speed scenario; When the braking frequency , determine that the driving scenario is a congestion scenario, where γ represents the number of braking times, represents the sliding window duration; When the steering angle change rate of the steering wheel , determine that the driving scenario is a curve scenario; Calculate the variance weight based on the variance within the sliding window; ; wherein, represents the variance weight of the l th dimension, represents the variance of the data of the l th dimension within the kth sliding window, represents the sum of the variances of all dimensions within the sliding window, l and j represents the dimension index; Calculate the entropy weight within the sliding window based on the information entropy algorithm; ; wherein, represents the information entropy of the lth dimension, , represents the probability of the l th dimension coefficient in the kth sliding window, represents the entropy weight of the ith dimension, B represents the total number of sliding windows, represents the information entropy of the jth dimension; Define the scenario rule weight based on the determined driving scenario; Fuse and normalize the variance weight, the entropy weight, and the scenario rule weight to obtain the comprehensive weight; ; wherein, w l represents the comprehensive weight, and represent the fusion coefficients, , ∈[0,1], represents the scenario rule weight of the l th dimension; Perform smoothing processing on the comprehensive weight to obtain the final weight; ; wherein, represents the final weight of the l th dimension; Calculate the similarity between any two data points in the sliding window based on the final weight, and cluster the data points using the DBSCAN algorithm based on the smoothed neighborhood radius; ; wherein, m and n respectively represent two data points, d(m, n) represents the similarity between the two data points m and n; Evaluation module: used to construct an obstacle avoidance path evaluation function based on the given weights and according to vehicle and road safety indicators to determine the value of the obstacle avoidance path evaluation function; The expression of the obstacle avoidance path evaluation function is as follows: ; Among them, , , , all represent weight coefficients, represents the path length, represents the safety distance, represents the obstacle avoidance speed, represents the change in steering angle; Planning module: used to calculate a planned path that conforms to the driver's driving behavior based on the value of the obstacle avoidance path evaluation function using the A-Star Algorithm; Specifically, the planning module is used to introduce the A-Star Algorithm and weight the value of the obstacle avoidance path evaluation function through a heuristic function; ; Among them, H p represents the weighted value of the obstacle avoidance path evaluation function; Determine the driving habit according to the value of the obstacle avoidance path evaluation function; When > 1, it is determined that the driving habit is aggressive. When = 1, it is determined that the driving habit is stable. When < 1, it is determined that the driving habit is conservative; Calculate a planned path that conforms to the driver's driving behavior based on the driving habit and using the A-Star Algorithm; Real-time monitor the relative position, speed, and driver's operation behavior of the intelligent vehicle and the obstacle, and record the real-time value of the obstacle avoidance path evaluation function; ; Among them, represents the change amount of the obstacle avoidance path evaluation function value, represents the evaluation function value of the current path, represents the evaluation function value of the planned path; If > the preset threshold , then recalculate the planned path; Execution module: used to execute the calculated planned path through the feedback module.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: The neighborhood radius of the traditional algorithm needs to be preset manually and cannot adapt to the dynamic changes of data distribution. However, the improved algorithm of the present invention can be adjusted in real time based on the data dynamics within the sliding window, avoiding being too large or too small due to a fixed neighborhood radius. The traditional algorithm uses Euclidean distance, and all dimensions contribute equally to the distance calculation, ignoring the importance differences of different sensor data in different scenarios. However, the improved algorithm of the present invention can dynamically adjust the weights of each dimension according to the driving scenario (such as highway, congestion, curve) and automatically calculate the dimension importance through statistical indicators such as variance and entropy value. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a flowchart of the intelligent vehicle obstacle avoidance control method based on driver behavior habits in the first embodiment of the present invention; Figure 2 It is a clustering result diagram obtained by applying the traditional DBSCAN clustering algorithm; Figure 3 It is a clustering result diagram obtained by applying the improved DBSCAN clustering algorithm proposed by the present invention; Figure 4 It is a schematic diagram of the obstacle avoidance planning path for various driving habits in the present invention; Figure 5 It is a structural block diagram of the intelligent vehicle obstacle avoidance control system based on driver behavior habits in the second embodiment of the present invention; The following specific embodiments will further illustrate the present invention in conjunction with the above-mentioned drawings. SPECIFIC EMBODIMENTS

[0014] For the convenience of understanding the present invention, the present invention will be described more comprehensively below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0015] It should be noted that when an element is referred to as being "fixedly provided on" another element, it can be directly on the other element or there may also be an intermediate element. When an element is considered to be "connected" to another element, it can be directly connected to the other element or there may be an intermediate element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are only for the purpose of illustration.

[0016] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the technical field to which this invention belongs. The terms used in the description of this invention herein are for the purpose of describing specific embodiments only and are not intended to limit the invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0017] Please refer to Figures 1 to 4 , which shows an intelligent vehicle obstacle avoidance control method based on driver behavior habits in the first embodiment of the present invention, including the following steps: S10, perform initialization operations on each sensor component of the intelligent vehicle; S20, collect and monitor key parameters in real time during driving through each sensor component; S30, divide the sliding window for the key parameters, and adjust the neighborhood radius within the sliding window based on the dynamic index within the sliding window; S40, determine the driving scenario according to the key parameters, and assign corresponding weights to each sensor component based on the data characteristics of the key parameters; S50, based on the assigned weights, construct an obstacle avoidance path evaluation function according to vehicle and road safety indicators to determine the obstacle avoidance path evaluation function value; S60, based on the obstacle avoidance path evaluation function value, calculate a planned path that conforms to the driver's driving behavior based on the A-Star Algorithm; S70, execute the calculated planned path through the feedback module.

[0018] It can be understood that the neighborhood radius of the traditional algorithm needs to be preset manually and cannot adapt to the dynamic changes of data distribution. However, the improved algorithm of the present invention can be adjusted in real time based on the data dynamics within the sliding window, avoiding the situation that the radius is too large or too small due to a fixed neighborhood radius; The traditional algorithm uses Euclidean distance, and all dimensions contribute equally to the distance calculation, ignoring the importance differences of different sensor data in different scenarios. However, the improved algorithm of the present invention can dynamically adjust the weights of each dimension according to the driving scenario (such as highway, congestion, curve) and automatically calculate the dimension importance through statistical indicators such as variance and entropy value.

[0019] Furthermore, the sensor components include an accelerator pedal sensor, a brake pedal sensor, a steering wheel sensor, an in-vehicle camera, an environmental perception camera, a vehicle status sensor, a millimeter wave radar, and a lidar. The initialization operation is to perform self-check on the sensor components, and this step can ensure the normal operation of all sensor components.

[0020] Further, for step S20, during the vehicle driving process, the data acquisition module continuously and synchronously collects various types of data. Among them, the driving operation sensors real-time monitor the operations of the driver on the accelerator, brake, and steering wheel, including the changes in parameters such as operation force, angle, and speed. The in-vehicle camera in the visual monitoring system continuously captures the driver's facial expressions, line of sight directions, and eye states, while the environmental perception camera continuously scans the road conditions, traffic signs, other vehicles, pedestrians, and obstacle information around the vehicle. The vehicle state sensors accurately obtain data such as the vehicle speed and acceleration, and other auxiliary sensors (such as seat pressure sensors) also work synchronously to collect relevant information. These data are transmitted to the data processing module in real time at a high frequency.

[0021] Further, step S20 specifically includes: Using the mean filtering algorithm to perform filtering and smoothing processing on the data of the key parameters; ; Among them, represents the original data at time t, represents the filtered data of, G is the number of data points; Performing zero-point correction and gain correction on the data of the key parameters after the smoothing processing; ; ; Among them, represents the calibrated value after zero-point correction and gain correction, represents the original value, c represents the zero-point offset, and e is the gain coefficient; Performing normalization processing on the data of the key parameters after the zero-point correction and gain correction; ; Among them, represents the normalized value, represents the minimum value in the data, represents the maximum value in the data. This step is to uniformly map data with different dimensions to a specific standard range (such as the [0, 1] interval).

[0022] Further, step S30 specifically includes: Dividing the key parameters into multiple sliding windows with a fixed length according to a continuous time series; ; Among them, X k represents the data range of the k-th sliding window, S represents the sliding step, WRepresents the window size; Calculate the dynamic index of the sliding window; ; ; Wherein, Represents the variance of the data within the k-th sliding window, Represents the mean value of the data within the k-th sliding window, Represents the i-th data point under this sliding window, Represents the multi-dimensional range mean within the k-th sliding window (this index can reflect the overall dynamic range of the data within the sliding window), d Represents the data dimension coefficient (such as speed, acceleration dimension), Represents the l -th dimension data set within the k-th sliding window, Represents the l -th dimension maximum value within the k-th sliding window, Represents the l -th dimension minimum value within the k-th sliding window; Linearly adjust the neighborhood radius within the sliding window based on the variance index and the range index; ; ; Wherein, Represents the dynamic neighborhood radius of the k-th sliding window, Represents the basic neighborhood radius, Represents the variance scaling coefficient, Represents the global variance, Represents the minimum protection radius, Represents the weight coefficient of the range; Smooth the linearly adjusted neighborhood radius; ; Represents the smoothed neighborhood radius, Represents the smoothed neighborhood radius in the previous sliding window, Represents the smoothing factor. Smoothing is to avoid sudden changes in the neighborhood radius.

[0023] Furthermore, the step S40 specifically includes: Determine the driving scenario according to the real-time key parameters, and the driving scenario includes highway scenario, congestion scenario and curve scenario; When the mean speed and the speed variance , determine that the driving scenario is a highway scenario; When the braking frequency , determine that the driving scenario is a congestion scenario, where γ represents the number of braking times, represents the sliding window duration; When the steering angle change rate of the steering wheel , determine that the driving scenario is a curve scenario; Calculate the variance weight based on the variance within the sliding window; ; Among them, represents the variance weight of the l th dimension, represents the variance of the l th dimension data within the kth sliding window, represents the sum of variances of all dimensions within the sliding window, l and j represents the dimension index; Calculate the entropy weight within the sliding window based on the information entropy algorithm; ; Among them, represents the information entropy of the lth dimension, , represents the probability of the l th dimension coefficient in the kth sliding window, represents the entropy weight of the ith dimension, B represents the total number of sliding windows, represents the information entropy of the jth dimension; Define the scenario rule weight based on the determined driving scenario; In the high-speed scenario, the weight of speed is increased; in the congestion scenario, the weight of braking is increased; in the curve scenario, the weight of the steering wheel is increased; Fuse and normalize the variance weight, the entropy weight, and the scenario rule weight to obtain the comprehensive weight; ; Among them, w l represents the comprehensive weight, and represent the fusion coefficients, , ∈[0,1], represents the l th dimension of the scenario rule weight; the comprehensive weight satisfies, ; Perform smoothing processing on the comprehensive weight to obtain the final weight; ; Among them, represents the lThe final weights of each dimension; Based on the final weights, calculate the similarity between any two data points in the sliding window, and based on the smoothed neighborhood radius, use the DBSCAN algorithm to cluster the data points; ; where m and n respectively represent two data points, d(m, n) represents the similarity between two data points m and n.

[0024] It can be understood that in this step, the current driving scenario is judged through the statistical features of the data within the sliding window, the adjustment direction of the weights is determined, and then according to the type of the scenario and the features of the data within the window, the representation ability of each dimension for the current scenario is quantified. During the operation of the algorithm, the number of data points within the neighborhood of each data point is calculated. For example, for a data point with a large change in throttle pressure and frequent braking operations, if there are a sufficient number (greater than or equal to the minimum number of points) of similar data points (similar in terms of throttle, brake, other driving behaviors, and state dimensions) within its neighborhood radius, then this data point is identified as a core point. By continuously expanding the neighborhood of the core point, clusters are formed. If some data points are within the neighborhood of a certain core point but the number of data points within their own neighborhoods is less than the minimum number of points, they are marked as border points and belong to a certain cluster. In this way, different driving behavior clusters are identified. It is possible to cluster data points with similar aggressive driving characteristics into one category, representing aggressive driving behavior. Cluster data points with similar smooth driving characteristics into one category, representing smooth driving behavior. Cluster data points with similar conservative driving characteristics into one category, representing conservative driving behavior.

[0025] Furthermore, the expression of the obstacle avoidance path evaluation function is as follows: ; where, , , , all represent weight coefficients, represents the path length, represents the safety distance, represents the obstacle avoidance speed, represents the change in steering angle.

[0026] Furthermore, the specific steps of step S60 include: Introduce the A-Star Algorithm, and weight the value of the obstacle avoidance path evaluation function through a heuristic function; ; where, H p represents the value of the weighted obstacle avoidance path evaluation function; Determine the driving habit according to the value of the obstacle avoidance path evaluation function; When > 1, it is judged that the driving habit is aggressive. When = 1, it is judged that the driving habit is steady. When < 1, it is judged that the driving habit is conservative; Based on the driving habit and using the A-Star Algorithm, calculate the path that conforms to the driver's driving behavior planning. Real-time monitor the relative position, speed of the intelligent vehicle and the obstacle, and the driver's operation behavior, and record the real-time value of the obstacle avoidance path evaluation function; ; Among them, represents the change amount of the obstacle avoidance path evaluation function value, represents the evaluation function value of the current path, represents the evaluation function value of the planned path; During the obstacle avoidance process, the relative position, speed of the vehicle and the obstacle, and the driver's operations (such as steering wheel angle, braking force, etc.) are monitored in real time. If a path deviation or a change in the obstacle position is detected, the path is replanned and the weight coefficient is adjusted.

[0027] If > the preset threshold , then recalculate the planned path. is a preset threshold, which is used to judge whether the path deviation is large enough to require replanning the path.

[0028] It can be understood that in terms of personalized obstacle avoidance path planning, if it is an aggressive driver, when using the algorithm for path search, adjust the cost of nodes with larger steering angles or higher speed changes according to the aggressive coefficient, so that the planned path is more inclined to meet its fast driving style. For steady drivers, increase the cost of such nodes according to the steady coefficient to ensure the comfort and stability of the path. At the same time, if the driver is inattentive, appropriately increase the safety distance margin.

[0029] Figure 4 This is the schematic diagram of the obstacle avoidance planning path for various driving habits in the present invention. This simulation diagram is drawn using MatlabR2023a version. As shown in the figure, the three dotted lines respectively represent the planned paths of aggressive, steady and conservative driving habits, and the black area is the obstacle area. Among them, the aggressive driver chooses to directly accelerate quickly when encountering an obstacle vehicle, the steady driver chooses to accelerate slowly, and the conservative driver chooses to decelerate and let the obstacle vehicle go first when encountering an obstacle vehicle.

[0030] Finally, according to step S70, the execution feedback module accurately conveys the control instruction generated by the obstacle avoidance strategy generation module to the actuator of the vehicle, completing the obstacle avoidance of the vehicle.

[0031] Compare Figure 2 and Figure 3 It can be seen that the improved algorithm has a better clustering effect, and the data is denser with the clustering center.

[0032] In summary, the intelligent vehicle obstacle avoidance control method based on driver behavior habits in the above embodiments of the present invention can be dynamically adjusted in real time based on the data within the sliding window, avoiding the radius being too large or too small due to a fixed neighborhood radius, and can dynamically adjust the weights of each dimension according to the driving scenario (such as highway, congestion, curve) and automatically calculate the dimension importance through statistical indicators such as variance and entropy value. Please refer to Figure 5 , which shows the intelligent vehicle obstacle avoidance control system based on driver behavior habits in the second embodiment of the present invention, including: Initialization module 11: used to perform initialization operations on each sensor of the intelligent vehicle; The sensors include an accelerator pedal sensor, a brake pedal sensor, a steering wheel sensor, an in-vehicle camera, an environmental perception camera, a vehicle state sensor, a millimeter-wave radar, and a lidar. The initialization operation is to perform self-check on the sensors; Acquisition module 12: used to collect and monitor key parameters during driving in real time through each sensor; Preprocessing module: used to filter and smooth the data of the key parameters by using the mean filtering algorithm; ; Among them, represents the original data at time t, represents the filtered data of, G is the number of data points; Perform zero-point correction and gain correction on the data of the key parameters after smoothing processing; ; ; Among them, represents the calibration value after zero-point correction and gain correction, represents the original value, c represents the zero-point offset, and e is the gain coefficient; Perform normalization processing on the data of the key parameters after zero-point correction and gain correction; ; Among them, represents the normalized value, Represents the minimum value in the data, represents the maximum value in the data; Adjustment module 13: used to perform a sliding window division on the key parameter and adjust the neighborhood radius within the sliding window based on the dynamic index within the sliding window; The adjustment module 13 is specifically configured to: divide the key parameter into multiple sliding windows of a fixed length according to a continuous time series; ; wherein, X k represents the data range of the k-th sliding window, S represents the sliding step size, W represents the window size; Calculate the dynamic index of the sliding window; ; ; wherein, represents the variance of the data within the k-th sliding window, represents the mean value of the data within the k-th sliding window, represents the i-th data point under this sliding window, represents the mean value of the multi-dimensional range within the k-th sliding window, d represents the data dimension coefficient, represents the l -th dimension data set within the k-th sliding window, represents the l -th dimension maximum value within the k-th sliding window, represents the l -th dimension minimum value within the k-th sliding window; Linearly adjust the neighborhood radius within the sliding window based on the variance index and the range index; ; ; wherein, represents the dynamic neighborhood radius of the k-th sliding window, represents the basic neighborhood radius, represents the variance scaling coefficient, represents the global variance, represents the minimum protection radius, represents the weight coefficient of the range; Smooth the linearly adjusted neighborhood radius; ; represents the neighborhood radius after smoothing, Represents the smoothed neighborhood radius in the previous sliding window, represents the smoothing factor; Weighting module 14: used to determine the driving scenario according to the key parameters, and assign corresponding weights to each sensor device based on the data characteristics of the key parameters; The weighting module 14 is specifically used to: determine the driving scenario according to the real-time key parameters, and the driving scenario includes a high-speed scenario, a congestion scenario, and a curve scenario; When the average speed and the speed variance , determine that the driving scenario is a high-speed scenario; When the braking frequency , determine that the driving scenario is a congestion scenario, where γ represents the number of braking times, represents the sliding window duration; When the rate of change of the steering wheel angle , determine that the driving scenario is a curve scenario; Calculate the variance weight based on the variance within the sliding window; ; Among them, represents the variance weight of the l th dimension, represents the variance of the data of the l th dimension within the kth sliding window, represents the sum of the variances of all dimensions within the sliding window, l and j represents the dimension index; Calculate the entropy weight within the sliding window based on the information entropy algorithm; ; Among them, represents the information entropy of the lth dimension, , represents the probability of the l th dimension coefficient in the kth sliding window, represents the entropy weight of the ith dimension, B represents the total number of sliding windows, represents the information entropy of the jth dimension; Define the scenario rule weight based on the determined driving scenario; Fuse and normalize the variance weight, the entropy weight, and the scenario rule weight to obtain the comprehensive weight; ; Among them, w l represents the comprehensive weight, and denotes the fusion coefficient, and ∈[0,1], denotes the weight of the scenario rule for the l th dimension; Perform smoothing processing on the comprehensive weight to obtain the final weight; ; wherein, denotes the final weight for the l th dimension; Based on the final weight, calculate the similarity between any two data points in the sliding window, and cluster the data points using the DBSCAN algorithm based on the smoothed neighborhood radius; ; where m and n respectively represent two data points, d(m, n) denotes the similarity between two data points m and n; Evaluation module 15: Used to construct an obstacle avoidance path evaluation function based on the assigned weight and according to the vehicle and road safety indicators to determine the obstacle avoidance path evaluation function value; The expression of the obstacle avoidance path evaluation function is as follows: ; wherein, , , , all denote weight coefficients, denotes the path length, denotes the safety distance, denotes the obstacle avoidance speed, denotes the steering angle change; Planning module 16: Used to calculate a planned path that conforms to the driver's driving behavior based on the obstacle avoidance path evaluation function value using the A-Star Algorithm; The planning module 16 is specifically used for: introducing the A-Star Algorithm and weighting the obstacle avoidance path evaluation function value through a heuristic function; ; wherein, H p denotes the weighted obstacle avoidance path evaluation function value; Determine the driving habit according to the obstacle avoidance path evaluation function value; When >1, judge that the driving habit is aggressive. When =1, judge that the driving habit is stable. When When it is < 1, it is determined that the driving habit is conservative; Based on the driving habit and using the A-Star Algorithm, calculate a planned path that conforms to the driver's driving behavior; Real-time monitor the relative position, speed of the intelligent vehicle and the obstacle, and the driver's operation behavior, and record the real-time obstacle avoidance path evaluation function value; ; Among them, represents the change amount of the obstacle avoidance path evaluation function value, represents the evaluation function value of the current path, represents the evaluation function value of the planned path; If > the preset threshold , then recalculate the planned path; Execution module 17: used to execute the calculated planned path through the feedback module.

[0033] In the description of this specification, the descriptions with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0034] The above-described embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention should be subject to the appended claims.

Claims

1. An intelligent vehicle obstacle avoidance control method based on driver behavior habits, characterized in that, It includes the following steps: Perform initialization operations on each sensor component of the intelligent vehicle; Collect and monitor key parameters during driving in real time through each sensor component; Divide the key parameters into sliding windows, and adjust the neighborhood radius within the sliding window based on the dynamic index within the sliding window; Determine the driving scenario based on the key parameters, and assign corresponding weights to each sensor component based on the data characteristics of the key parameters; Based on the assigned weights, construct an obstacle avoidance path evaluation function according to the vehicle and road safety indicators to determine the obstacle avoidance path evaluation function value; Based on the obstacle avoidance path evaluation function value, calculate a planned path that conforms to the driver's driving behavior based on the A-Star Algorithm; Execute the calculated planned path through the feedback module.

2. The intelligent vehicle obstacle avoidance control method based on driver behavior habits according to claim 1, wherein, The sensor components include an accelerator pedal sensor, a brake pedal sensor, a steering wheel sensor, an in-vehicle camera, an environmental perception camera, a vehicle state sensor, a millimeter wave radar, and a lidar. The initialization operation is to perform self-checks on the sensor components.

3. The intelligent vehicle obstacle avoidance control method based on driver behavior habits according to claim 2, wherein Before the step of dividing the key parameters into sliding windows and adjusting the neighborhood radius based on the dynamic index within the sliding window, it further includes: Use the mean filtering algorithm to filter and smooth the data of the key parameters; ； Among them, represents the original data at time t, represents the filtered data of, and G is the number of data points; Perform zero point correction and gain correction on the data of the key parameters after smoothing processing; ； ； Among them, represents the calibration value after zero point correction and gain correction, represents the original value, c represents the zero point offset, and e is the gain coefficient; Perform normalization processing on the data of the key parameters after zero point correction and gain correction; ； Among them, represents the normalized value, represents the minimum value in the data, represents the maximum value in the data.

4. The intelligent vehicle obstacle avoidance control method based on driver behavior habits according to claim 1, wherein The step of dividing the key parameters into sliding windows and adjusting the neighborhood radius based on the dynamic index within the sliding window specifically includes: Divide the key parameters into multiple sliding windows of fixed length according to a continuous time series; ； Among them, X k represents the data range of the k-th sliding window, S represents the sliding step size, W represents the window size; Calculate the dynamic index of the sliding window; ； ； Among them, represents the variance of the data within the k-th sliding window, represents the mean of the data within the k-th sliding window, represents the i-th data point under this sliding window, represents the mean of the multi-dimensional range within the k-th sliding window, d represents the data dimension coefficient, represents within the k-th sliding window the l data set of the -th dimension, l represents the maximum value of the -th dimension within the k-th sliding window, l represents the minimum value of the -th dimension within the k-th sliding window; Linearly adjust the neighborhood radius within the sliding window based on the variance index and the range index; ； ； Among them, represents the dynamic neighborhood radius of the k-th sliding window, represents the basic neighborhood radius, represents the variance scaling coefficient, represents the global variance, represents the minimum protection radius, represents the weight coefficient of the range; Perform smoothing processing on the linearly adjusted neighborhood radius; ； denotes the neighborhood radius after smoothing processing denotes the neighborhood radius after smoothing processing in the previous sliding window denotes the smoothing factor 5. The intelligent vehicle obstacle avoidance control method based on driver behavior habits according to claim 4, characterized in that The step of determining the driving scenario based on the real-time key parameters and assigning corresponding weights to each sensor component based on the data characteristics specifically includes: Determine the driving scenario based on the real-time key parameters, and the driving scenario includes a highway scenario, a congestion scenario, and a bend scenario; When the average speed and the speed variance , it is determined that the driving scenario is a highway scenario; When the braking frequency , it is determined that the driving scenario is a congested scenario, where γ represents the number of braking times, represents the sliding window duration; When the steering wheel angle change rate , it is determined that the driving scenario is a curve scenario; Calculate the variance weight based on the variance within the sliding window; ； Among them, represents the variance weight of the l th dimension, represents the variance of the data of the l th dimension within the kth sliding window, represents the sum of variances of all dimensions within the sliding window, l and j represents the dimension index; Calculate the entropy weight within the sliding window based on the information entropy algorithm; ； Among them, represents the information entropy of the l th dimension, , represents the probability of the l th dimension coefficient in the kth sliding window, represents the entropy weight of the ith dimension, B represents the total number of sliding windows, represents the information entropy of the jth dimension; Define the scenario rule weight based on the determined driving scenario; Fuse and normalize the variance weight, the entropy weight, and the scenario rule weight to obtain a comprehensive weight; ； Among them, w l represents the comprehensive weight, and represents the fusion coefficient, 、 ∈[0,1], represents the weight of the scenario rule for the l th dimension; Perform smoothing processing on the comprehensive weight to obtain the final weight; ； Among them, represents the final weight of the l th dimension; Based on the final weight, calculate the similarity between any two data points in the sliding window, and perform clustering on the data points based on the smoothed neighborhood radius using the DBSCAN algorithm; ； where m and n respectively represent two data points, d(m, n) represents the similarity between two data points m and n.

6. The intelligent vehicle obstacle avoidance control method based on driver behavior habits according to claim 1, wherein The expression of the obstacle avoidance path evaluation function is as follows: ； Among them, , , , all represent weight coefficients, represents the path length, represents the safety distance, represents the obstacle avoidance speed, represents the steering angle change.

7. The intelligent vehicle obstacle avoidance control method based on driver behavior habits according to claim 1, wherein, The step of calculating a planned path that conforms to the driver's driving behavior based on the obstacle avoidance path evaluation function value through the A-Star Algorithm specifically includes: Introduce the A-Star Algorithm, and weight the value of the obstacle avoidance path evaluation function through a heuristic function; ； Among them, H p represents the value of the obstacle avoidance path evaluation function after weighting; Determine the driving habit according to the value of the obstacle avoidance path evaluation function; When > 1, it is determined that the driving habit is aggressive. When = 1, it is determined that the driving habit is stable. When < 1, it is determined that the driving habit is conservative; Based on the driving habit and using the A-Star Algorithm, calculate the planned path that conforms to the driver's driving behavior; Real-time monitor the relative position, speed and driver's operation behavior between the intelligent vehicle and the obstacle, and record the real-time value of the obstacle avoidance path evaluation function; ； Among them, represents the change amount of the obstacle avoidance path evaluation function value, represents the evaluation function value of the current path, represents the evaluation function value of the planned path; If is greater than a preset threshold , recalculate the planned path.

8. An intelligent vehicle obstacle avoidance control system based on driver behavior habits, characterized in that, Include: Initialization module: used to perform initialization operations on each sensor of the intelligent vehicle; Acquisition module: used to collect and monitor key parameters during driving in real time through each sensor; Adjustment module: used to divide the sliding window for the key parameters, and adjust the neighborhood radius within the sliding window based on the dynamic index within the sliding window; Weight assignment module: used to determine the driving scenario according to the key parameters, and assign corresponding weights to each sensor based on the data characteristics of the key parameters; Evaluation module: used to construct an obstacle avoidance path evaluation function based on the assigned weights and according to the vehicle and road safety indicators to determine the value of the obstacle avoidance path evaluation function; Planning module: used to calculate the planned path that conforms to the driver's driving behavior based on the value of the obstacle avoidance path evaluation function and based on the A-Star Algorithm; Execution module: used to execute the calculated planned path through the feedback module.

Images