Roadblock recognition system for assisting driving of heavy truck

Abstract

The invention discloses a roadblock recognition system for assisting driving of a heavy truck, relates to the technical field of intelligent traffic systems and vehicle safety, and solves the problem that in the prior art, after a roadblock recognition system recognizes the roadblock type, the roadblock type is directly provided for a driver; the technical problem that the roadblock identification system is low in efficiency due to the lack of intelligent consideration of roadblock types and whether a vehicle will cause harm to the driving process is solved. Inputting the detection image into a barricade identification model to obtain a category label of the barricade; searching a danger score corresponding to the category label in a category danger score table; calculating an influence coefficient of the roadblock according to the detection image; generating a danger coefficient according to the danger score, the influence coefficient, the heavy truck data and the roadblock data; the danger degree of the roadblock to the heavy truck is calculated by integrating multivariate data, intelligent danger early warning prompt is achieved, and the accuracy and directivity of danger early warning are higher; and the efficiency of the roadblock identification system is improved.

Description

Technical Field

[0001] The present application belongs to the field of intelligent transportation system and vehicle safety technology, and specifically is a roadblock recognition system for assisting heavy truck driving. Background Art

[0002] Heavy trucks are referred to as heavy trucks, which usually refer to trucks and semi-trailer tractors with large load capacity. Heavy trucks are a very important means of transportation, mainly used for long-distance transportation of goods and special-purpose vehicles. With the rapid development of the transportation industry, the role of heavy trucks in the national economy has become increasingly prominent. As the main force of road transportation, the driving safety of heavy trucks is directly related to the overall safety of road traffic.

[0003] However, due to the characteristics of heavy trucks such as long body, heavy weight, and long braking distance, it is often difficult to respond in time when facing sudden roadblocks, thus causing traffic accidents. In the existing technology, the roadblock recognition system directly provides the roadblock type to the driver after identifying the roadblock type, lacks intelligent consideration of the roadblock type and whether the vehicle itself will cause harm to the driving process, resulting in low efficiency of the roadblock recognition system. Therefore, it is particularly important to develop an efficient and intelligent roadblock recognition system to assist heavy truck driving. Summary of the invention

[0004] The present application aims to solve at least one of the technical problems existing in the prior art; to this end, the present application proposes a roadblock identification system for assisting the driving of heavy trucks, which is used to solve the technical problem that the roadblock identification system in the prior art directly provides the roadblock type to the driver after identifying the roadblock type, lacks intelligent consideration of the roadblock type and whether the vehicle itself will cause harm to the driving process, thereby resulting in low efficiency of the roadblock identification system.

[0005] To achieve the above-mentioned purpose, the present application provides a roadblock identification system for assisting heavy truck driving, comprising: a data acquisition module, a data analysis module, a warning module and a database;

[0006] The data acquisition module: acquires the detection image and heavy truck data in real time through the data acquisition device, and performs data preprocessing on the detection image to obtain the preprocessed detection image;

[0007] The data analysis module: inputs the detection image into the roadblock recognition model to obtain the category label of the roadblock; searches the category label in the category risk score table for the risk score corresponding to the category label; calculates the impact coefficient of the roadblock according to the detection image; obtains the roadblock data through the corresponding data acquisition device, generates the risk coefficient according to the risk score, the impact coefficient, the heavy truck data and the roadblock data; generates an alarm signal according to the risk coefficient;

[0008] The warning module performs corresponding operations according to the warning signal and provides corresponding information to the driver.

[0009] The present application acquires detection images and heavy truck data in real time through data acquisition equipment, and performs data preprocessing on the detection images to obtain preprocessed detection images; inputs the detection images into a roadblock recognition model to obtain a category label of the roadblock; searches for the danger score corresponding to the category label in a category danger score table through the category label; calculates the influence coefficient of the roadblock according to the detection image; acquires the roadblock data through corresponding data acquisition equipment, and generates a danger coefficient according to the danger score, influence coefficient, heavy truck data and roadblock data; generates an alarm signal according to the danger coefficient; performs corresponding operations according to the alarm signal, calculates the danger degree of the roadblock to the heavy truck by integrating multivariate data, realizes intelligent danger warning prompts, and makes the accuracy and direction of the danger warning higher; improves the efficiency of the roadblock recognition system.

[0010] Furthermore, the data preprocessing includes: image denoising processing and image enhancement processing.

[0011] Furthermore, the roadblock recognition model is constructed by an artificial intelligence model, including:

[0012] Acquire a number of historical data from a database; the historical data includes a number of category labels and their corresponding roadblock images; the category labels are set by an expert according to the roadblocks in the roadblock images;

[0013] Divide historical data into training data, verification data, and test data according to category labels; perform data preprocessing on the training data, verification data, and test data to obtain a training set, a verification set, and a test set;

[0014] Select an AI model as the base model;

[0015] Train the basic model with the training set, and adjust the learning rate or other hyperparameters on the validation set to obtain the pre-trained model;

[0016] By verifying the pre-trained model on the test set, we finally get a roadblock recognition model that inputs a detection image and outputs a category label.

[0017] Furthermore, searching the category risk score table for the risk score corresponding to the category label by the category label includes:

[0018] Get the category label;

[0019] Find the roadblock category corresponding to the current category label in the category hazard score table;

[0020] The danger score corresponding to the roadblock category is selected as the danger score of the current category label; the danger score is an average value of the scores given by multiple experts to the danger of each roadblock category.

[0021] Furthermore, the calculation of the influence coefficient of the roadblock according to the detection image includes:

[0022] Obtain two detection images separated by a time interval of T, denoted as JT1 and JT2, where JT2 is the detection image after a time interval of T from JT1; extract the roadblock sizes W1 and W2 with the same category labels in the two detection images; W1 includes a height WG1 and a width WK1, and W2 includes a height WG2 and a width WK2;

[0023] Calculate the corresponding roadblock area Si through the formula;

[0024] Si=WGi×WKi; where i=1,2, i represents the number of the detected image;

[0025] Determine whether S2 is greater than S1;

[0026] Yes, the influence coefficient of the roadblock YX is calculated by the formula;

[0027] YX=k1×arctan(exp(-T / DT)×(S2-S1) / S1); where k1 is the proportionality coefficient, k1∈(0, 2 / π); DT is the unit time;

[0028] No, the influence coefficient YX=0.

[0029] Furthermore, generating a risk factor according to the risk score, the impact coefficient, the heavy truck data and the roadblock data includes:

[0030] Obtain the danger score WD, influence coefficient YX, braking distance Zd in the heavy truck data, and distance d in the roadblock data; the distance d is the distance between the heavy truck and the roadblock at the moment when the JT2 image is detected;

[0031] The risk factor WX is calculated by the formula; WX = WD × YX × exp (α1 × (Zd-d) / Dd);

[0032] Among them, α1 is the proportional coefficient, α1∈(0,1); Zd is the braking distance of the current heavy truck, and Dd is the unit distance.

[0033] Further, generating an alarm signal according to the risk factor includes:

[0034] Get the risk factor;

[0035] Determine whether the risk factor is greater than the risk threshold; if yes, generate a risk alarm signal; if no, when the risk factor is greater than D times the risk threshold, generate a risk warning signal; otherwise, do nothing; where D is the proportional coefficient, D∈(0,1).

[0036] Compared with the prior art, the beneficial effects of this application are:

[0037] 1. The present application acquires detection images and heavy truck data in real time through data acquisition equipment, and performs data preprocessing on the detection images to obtain preprocessed detection images; inputs the detection images into a roadblock recognition model to obtain a category label of the roadblock; searches for the hazard score corresponding to the category label in a category hazard score table through the category label; calculates the influence coefficient of the roadblock based on the detection image; acquires roadblock data through corresponding data acquisition equipment, and generates a hazard coefficient based on the hazard score, influence coefficient, heavy truck data and roadblock data; generates an alarm signal based on the hazard coefficient to remind the driver; calculates the degree of danger of the roadblock to the heavy truck by integrating multivariate data, realizes intelligent hazard warning prompts, and improves the efficiency of the roadblock recognition system. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 This is a schematic diagram of the principle of a roadblock recognition system for assisting heavy truck driving in the present application;

[0040] Figure 2 This is a flow chart of a roadblock identification system for assisting heavy truck driving in the present application. DETAILED DESCRIPTION

[0041] The technical solution of the present application will be described clearly and completely below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without creative work are within the scope of protection of the present application.

[0042] See also Figure 1-Figure 2 , the embodiment of the present application provides a roadblock identification system for assisting heavy truck driving, including: a data acquisition module, a data analysis module, a warning module and a database;

[0043] Data acquisition module: The detection image and heavy truck data are acquired in real time through the data acquisition equipment, and the detection image is preprocessed to obtain the preprocessed detection image; the data acquisition equipment includes high-precision cameras, radars, and laser radars, etc.; the detection image is an image taken by a high-precision camera in the forward direction of the heavy truck; the heavy truck data is the parameter data of the heavy truck itself, including braking distance, etc.;

[0044] Data analysis module: input the detection image into the roadblock recognition model to obtain the category label of the roadblock; search the hazard score corresponding to the category label in the category hazard score table through the category label; calculate the impact coefficient of the roadblock according to the detection image; obtain the roadblock data through the corresponding data acquisition equipment, and generate the hazard coefficient according to the hazard score, impact coefficient, heavy truck data and roadblock data; generate an alarm signal according to the hazard coefficient; the hazard score can indicate the degree of danger of safety hazards caused by the roadblock; the impact coefficient can indicate the possibility of the roadblock causing harm to the heavy truck; the roadblock data includes other data such as the distance between the high-precision camera and the roadblock; the hazard coefficient can indicate the possibility of danger between the roadblock and the heavy truck;

[0045] Warning module: take corresponding actions according to the alarm signals and provide corresponding information to the driver; the alarm signals include danger alarm signals and danger warning signals.

[0046] This embodiment acquires detection images and heavy truck data in real time through data acquisition equipment, and performs data preprocessing on the detection images to obtain preprocessed detection images; inputs the detection images into a roadblock recognition model to obtain a category label of the roadblock; searches for the danger score corresponding to the category label in a category danger score table through the category label; calculates the influence coefficient of the roadblock according to the detection image; acquires the roadblock data through corresponding data acquisition equipment, and generates a danger coefficient according to the danger score, influence coefficient, heavy truck data and roadblock data; generates an alarm signal according to the danger coefficient; performs corresponding operations according to the alarm signal, calculates the danger degree of the roadblock to the heavy truck by integrating multivariate data, realizes intelligent danger warning prompts, and makes the accuracy and direction of the danger warning higher; improves the efficiency of the roadblock recognition system.

[0047] Data preprocessing includes: image denoising and image enhancement;

[0048] Image denoising: denoise the detected image and select the median filter method to remove the noise in the image to obtain a denoised image;

[0049] Image enhancement processing: Enhancement processing is performed on the denoised image after denoising, and the clarity and contrast of the detection image are improved by the contrast enhancement method to obtain the preprocessed detection image.

[0050] The roadblock recognition model is constructed through an artificial intelligence model, including:

[0051] Acquire a number of historical data from a database; the historical data includes a number of category labels and their corresponding roadblock images; the category labels are set by experts according to the roadblocks in the roadblock images;

[0052] The historical data is divided into training data, verification data and test data according to the category labels; the training data, verification data and test data are preprocessed to obtain the training set, verification set and test set; the ratio between the training set, test set and verification set is 7:2:1;

[0053] Select an artificial intelligence model as the basic model; artificial intelligence models include Yolo model, etc.;

[0054] Train the basic model with the training set, and adjust the learning rate or other hyperparameters on the validation set to obtain the pre-trained model;

[0055] By verifying the pre-trained model on the test set, we finally get a roadblock recognition model that inputs detection images and outputs category labels.

[0056] Searching for a danger score corresponding to the category label in a category danger score table according to the category label, including: obtaining the category label; searching for a roadblock category corresponding to the current category label in the category danger score table;

[0057] The danger score corresponding to the roadblock category is selected as the danger score of the current category label; the danger score is the average value of the scores given by multiple experts to the danger of each roadblock category; the danger score table is composed of roadblock categories and their corresponding danger scores.

[0058] Calculate the impact coefficient of the roadblock based on the detection image, including:

[0059] Obtain two detection images with an interval of T time, which are recorded as JT1 and JT2, respectively, where JT2 is the detection image after JT1 has passed T time; extract the roadblock sizes W1 and W2 with the same category label in the two detection images; W1 includes height WG1 and width WK1, and W2 includes height WG2 and width WK2; the interval time T is set according to experience, and in this embodiment, the interval time T is set to 0.1 second; in another embodiment, the interval time can be set according to the driving speed of the truck, and the faster the driving speed, the shorter the interval time;

[0060] Calculate the corresponding roadblock area Si through the formula;

[0061] Si=WGi×WKi; where i=1,2, i represents the number of the detection image; as the height and width of the detection image increase, the corresponding roadblock area will also increase;

[0062] Determine whether S2 is greater than S1;

[0063] Yes, the influence coefficient of the roadblock YX is calculated by the formula;

[0064] YX=k1×arctan(exp(-T / DT)×(S2-S1) / S1); k1 is a proportionality coefficient, k1∈(0,2 / π), and the specific value is set based on experience; DT is a unit time, which is set based on experience. In this embodiment, DT is set to 1 second; after T time, S2 is increasingly greater than S1, indicating that the distance between the roadblock and the heavy truck is gradually shrinking, and the influence of the roadblock on the heavy truck will increase accordingly, so the influence coefficient will also increase accordingly; when the area of ​​the roadblock changes at the same level, the shorter the interval time, the faster the distance between the obstacle and the truck is shortened, and the influence of the roadblock on the heavy truck will increase accordingly;

[0065] No, the influence coefficient YX=0.

[0066] Generate hazard factors based on hazard scores, impact factors, heavy truck data, and roadblock data, including:

[0067] Obtain the danger score WD, influence coefficient YX, braking distance Zd in the heavy truck data, and distance d in the roadblock data; distance d is the distance between the heavy truck and the roadblock at the moment of JT2 detection image;

[0068] The risk factor WX is calculated by the formula; WX = WD × YX × exp (α1 × (Zd-d) / Dd); if the impact coefficient of the roadblock is 0, it means that the roadblock has no dangerous effect on the heavy truck, so the risk factor is also 0; as the impact coefficient increases, the possibility of the roadblock causing danger to the heavy truck will increase, so the risk factor will also increase; when the distance between the roadblock and the heavy truck is less than the braking distance, the safety distance when the heavy truck needs to take the braking method is not enough, which will lead to safety hazards, so as the braking distance increases or the distance between the roadblock and the middle decreases, the risk factor will increase;

[0069] Among them, α1 is the proportional coefficient, α1∈(0, 1), and the specific value is set according to experience; Zd is the braking distance of the current heavy truck, and Dd is the unit distance, which is set according to experience. In this embodiment, the unit distance is set to 1 meter; the braking distance will vary depending on the speed, weight and braking acceleration of the heavy truck.

[0070] Generates alarm signals based on risk factors, including:

[0071] Get the risk factor;

[0072] Determine whether the risk factor is greater than the risk threshold, which is set based on experience; if yes, generate a risk alarm signal; if no, when the risk factor is greater than D times the risk threshold, generate a risk warning signal; otherwise, do nothing; wherein D is the proportional coefficient, D∈(0,1), which is set based on experience. In this embodiment, D is set to 0.7.

[0073] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0074] The working principle of the present application is as follows: a detection image and heavy truck data are acquired in real time through a data acquisition device, and data preprocessing is performed on the detection image to obtain a preprocessed detection image; the detection image is input into a roadblock recognition model to obtain a category label of the roadblock; the category label is used to find the danger score corresponding to the category label in the category danger score table; the influence coefficient of the roadblock is calculated according to the detection image; the roadblock data is acquired through a corresponding data acquisition device, and a danger coefficient is generated according to the danger score, the influence coefficient, the heavy truck data and the roadblock data; an alarm signal is generated according to the danger coefficient; corresponding operations are performed according to the alarm signal, and corresponding information is provided to the driver, and the degree of danger of the roadblock to the heavy truck is calculated by integrating multivariate data, so as to realize intelligent danger warning prompts, so that the accuracy and directionality of the danger warning are higher; the efficiency of the roadblock recognition system is improved, and the problem that the roadblock recognition system in the prior art directly provides the roadblock type to the driver after identifying the roadblock type is avoided, and the roadblock type and whether the vehicle itself will cause harm to the driving process are not intelligently considered, which leads to the problem of low efficiency of the roadblock recognition system.

[0075] The above embodiments are only used to illustrate the technical method of the present application and are not intended to limit it. Although the present application has been described in detail with reference to the preferred embodiments, a person of ordinary skill in the art should understand that the technical method of the present application may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present application.

Claims

1. A roadblock recognition system for assisting heavy truck driving, characterized in that: include: Data collection module, data analysis module, early warning module and database; The data acquisition module: acquires the detection image and heavy truck data in real time through the data acquisition device, and performs data preprocessing on the detection image to obtain the preprocessed detection image; The data analysis module: inputs the detection image into the roadblock recognition model to obtain the category label of the roadblock; searches the category label in the category risk score table for the risk score corresponding to the category label; calculates the impact coefficient of the roadblock according to the detection image; obtains the roadblock data through the corresponding data acquisition device, generates the risk coefficient according to the risk score, the impact coefficient, the heavy truck data and the roadblock data; generates an alarm signal according to the risk coefficient; The warning module performs corresponding operations according to the warning signal and provides corresponding information to the driver.

2. A roadblock recognition system for assisting heavy truck driving according to claim 1, characterized in that: The data preprocessing includes: image denoising processing and image enhancement processing.

3. The roadblock recognition system for assisting heavy truck driving according to claim 1, characterized in that: The roadblock recognition model is constructed through an artificial intelligence model, including: Acquire a number of historical data from a database; the historical data includes a number of category labels and their corresponding roadblock images; the category labels are set by an expert according to the roadblocks in the roadblock images; Divide historical data into training data, verification data, and test data according to category labels; perform data preprocessing on the training data, verification data, and test data to obtain a training set, a verification set, and a test set; Select an AI model as the base model; Train the basic model with the training set, and adjust the learning rate or other hyperparameters on the validation set to obtain the pre-trained model; By verifying the pre-trained model on the test set, we finally get a roadblock recognition model that inputs a detection image and outputs a category label.

4. The roadblock recognition system for assisting heavy truck driving according to claim 1, characterized in that: The step of searching the category risk score table for the risk score corresponding to the category label by using the category label includes: Get the category label; Find the roadblock category corresponding to the current category label in the category hazard score table; The danger score corresponding to the roadblock category is selected as the danger score of the current category label; the danger score is an average value of the scores given by multiple experts to the danger of each roadblock category.

5. The roadblock recognition system for assisting heavy truck driving according to claim 1, characterized in that: The calculating the influence coefficient of the roadblock according to the detection image comprises: Obtain two detection images separated by a time interval of T, denoted as JT1 and JT2, where JT2 is the detection image after a time interval of T from JT1; extract the roadblock sizes W1 and W2 with the same category labels in the two detection images; W1 includes a height WG1 and a width WK1, and W2 includes a height WG2 and a width WK2; Calculate the corresponding roadblock area Si through the formula; Si=WGi×WKi; where i=1,2, i represents the number of the detected image; Determine whether S2 is greater than S1; Yes, the influence coefficient of the roadblock YX is calculated by the formula; YX=k1×arctan(exp(-T / DT)×(S2-S1) / S1); where k1 is the proportionality coefficient, k1∈(0, 2 / π); DT is the unit time; No, the influence coefficient YX=0.

6. The roadblock recognition system for assisting heavy truck driving according to claim 1, characterized in that: The generating of the risk factor according to the risk score, the impact factor, the heavy truck data and the roadblock data includes: Obtain the danger score WD, influence coefficient YX, braking distance Zd in the heavy truck data, and distance d in the roadblock data; the distance d is the distance between the heavy truck and the roadblock at the moment when the JT2 image is detected; The risk factor WX is calculated by the formula; WX = WD × YX × exp (α1 × (Zd-d) / Dd); Among them, α1 is the proportional coefficient, α1∈(0,1); Zd is the braking distance of the current heavy truck, and Dd is the unit distance.

7. The roadblock recognition system for assisting heavy truck driving according to claim 1, characterized in that: The step of generating an alarm signal according to the risk factor comprises: Get the risk factor; Determine whether the risk factor is greater than the risk threshold; if yes, generate a risk alarm signal; if no, when the risk factor is greater than D times the risk threshold, generate a risk warning signal; otherwise, do nothing; where D is the proportional coefficient, D∈(0,1).

Images