Driving risk monitoring method for emergent target

By establishing an impulse-passing model of sudden targets on intelligent vehicles, using the decision tree algorithm to build a driving risk field, combining real-time environmental data and vehicle status information, driving risk warning in a visual blind spot environment is achieved, and drivers cannot discover it in a timely manner Traffic accidents of pedestrians or non-motor vehicles have improved driving safety.

CN118928384AActive Publication Date: 2024-11-12HEFEI UNIV OF TECH
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411354771.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-25
Publication Date
2024-11-12
Estimated Expiration
2044-09-25

AI Technical Summary

Technical Problem

In a visual blind spot environment, drivers are unable to detect and avoid emergencies of pedestrians or non-motor vehicles in a timely manner, resulting in frequent traffic accidents. The existing technology lacks effective road driving risk warning methods.

Method used

Using the impulsive travel model of sudden targets, a driving risk field is constructed through a decision tree algorithm, combining real-time environmental data and vehicle status information, to judge driving risks and adopt early warning strategies, including on-board sensors to obtain visibility, time period, traffic, traffic and environment Data such as temperature, generate a sample set, train a decision tree model, output the probability state of sudden impulsive travel of the target, judge driving risks and provide early warnings.

Benefits of technology

Without the support of a third-party device, it can warn and defensive driving in advance in a visual blind spot environment, effectively responding to collision accidents where the vehicle does not have time to brake when it detects obstacles, especially in the case of fast vehicle speeds, reducing the incidence rate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118928384A_ABST
    Figure CN118928384A_ABST
Patent Text Reader

Abstract

The invention provides a driving risk monitoring method for an emergent target, and belongs to the field of intelligent driving. The monitoring method comprises the following steps: acquiring historical environment data associated with an emergent target, establishing an emergent target impulse traversal model, and predicting the probability of emergent target impulse traversal; the driving risk field is constructed by combining the emergent target impulse passing probability and other state information of the intelligent vehicle, and the driving risk is judged according to the value of the driving risk field. According to the method, the driving risk field is constructed, the risk threshold value of safe driving of the intelligent vehicle is provided, once the driving risk field value of the intelligent vehicle exceeds the risk field threshold value, the intelligent vehicle can give out an early warning prompt, and collision accidents occurring when an emergent target travels in an impulsive mode in a visual blind area environment are effectively avoided.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of intelligent driving, and in particular relates to a driving risk monitoring method for emergent targets. Background Art

[0002] With the rapid growth of car ownership, traffic safety issues have become increasingly serious. According to the latest report of the World Health Organization, the number of deaths in traffic accidents worldwide exceeds 3,200 per day, and the number of deaths in traffic accidents exceeds 2 per minute. Traffic accidents are an important cause of death for children and young people aged 5-29. At the same time, the mortality rate of traffic accidents involving pedestrians, non-motor vehicles and motor vehicles is relatively high, among which "ghost peeking" is a frequent case of such traffic accidents. When a vehicle is driving normally, the driver's vision is blocked by side parking, buildings, bus stops at bus stops, stone piers of overpasses and green belts, forming a driver's visual blind spot environment. When pedestrians cross the road in front, the driver cannot find them in time and make timely and effective avoidance, resulting in a collision between people and vehicles, commonly known as "ghost peeking". Due to the suddenness of such accidents and the vulnerability of pedestrians and non-motor vehicles, such accidents often cause a large amount of property losses and serious casualties. At present, there is little research on the risk theory of sudden target obstacles in the visual blind spot environment, and there is even less research on the risk warning method of road driving with sudden targets in the visual blind spot environment. Summary of the invention

[0003] In view of this, the present invention proposes a driving risk monitoring method for emergent targets. The driving risk warning method aims to provide a driving risk field that can quantify the visual blind spot environment that the vehicle is subjected to during driving, and use the risk field to determine whether the current driving speed of the vehicle is safe.

[0004] The purpose of the present invention is achieved in this way. The present invention provides a driving risk monitoring method for emergent targets, wherein the emergent targets are defined as traffic participants that suddenly appear and pose a major threat to the safe driving of vehicles, including but not limited to suddenly appearing pedestrians, animals and non-motor vehicles; the driving risk monitoring method includes establishing an emergent target impulsive travel model, predicting the probability state of the emergent target impulsive travel ahead through the emergent target impulsive travel model; constructing a driving risk field in combination with the probability state of the emergent target impulsive travel, judging the driving risk according to the driving risk field value and taking corresponding early warning strategies, and the specific steps are as follows:

[0005] Step 1, obtaining historical environmental data of scenes where the emergent target may appear, including visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6, and generating a sample set D;

[0006] Step 2: Establish a decision tree model, use the sample set D as the decision tree training set, perform classification training based on the C4.5 decision tree algorithm, and obtain an emergent target impulsive travel model;

[0007] Step 3, obtaining real-time environmental data including visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6, and inputting the emergent target impulsive travel model obtained in step 2, and outputting the probability state of the emergent target impulsive travel;

[0008] The probability state of the emergent target impulsive travel includes three states, which are respectively recorded as a high probability state, a medium probability state and a low probability state;

[0009] Step 4: Given a value of the probability δ of an emergent target impulsive travel, a driving risk field is constructed in combination with the probability δ of the emergent target impulsive travel, the driving risk is determined according to the value of the driving risk field, and a corresponding early warning strategy is adopted;

[0010] Step 4.1, the vehicle-mounted sensor obtains vehicle status information and real-time environmental information;

[0011] The vehicle-mounted sensor is installed at the center of the front bumper of the smart vehicle; the vehicle status information includes the speed and position of the smart vehicle;

[0012] Step 4.2, giving the value of the probability δ: high probability state, δ = 0.8; medium probability state, δ = 0.5; low probability state, δ = 0.2;

[0013] Step 4.3, record the current time when the intelligent vehicle is driving as time t, and establish the driving risk field at time t;

[0014] Step 4.3.1: The time when the front of the intelligent vehicle reaches the conflict area at time t is recorded as t c1 (t), which is expressed as follows:

[0015]

[0016] The conflict area refers to the line segment in the emerging target walking route that overlaps with the intelligent vehicle driving route. l is the distance between the front of the intelligent vehicle and the conflict area along the vehicle's driving direction at time t, v c (t) is the speed of the smart vehicle at time t;

[0017] Step 4.3.2, determine whether the on-board sensor of the intelligent vehicle detects an emergent target at time t. If no emergent target is detected, execute step 4.3.3. If an emergent target is detected, jump to step 4.3.4;

[0018] Step 4.3.3, establish the first driving risk field of the intelligent vehicle in the visual blind spot environment at time t, and record it as the first driving risk field E(t)1, which is expressed as follows:

[0019]

[0020] Among them, θ(t) is the blind spot detection angle of the vehicle-mounted sensor, d l ×tanθ(t) is the width of the lateral field of view of the intelligent vehicle;

[0021] Go to step 5.1;

[0022] Step 4.3.4, establish the second driving risk field of the intelligent vehicle in the visual blind spot environment at time t, and record it as the second driving risk field E(t)2;

[0023] The calculated angle when the onboard sensor of the intelligent vehicle detects the emergent target is θ r , select as follows:

[0024] When the onboard sensor of the intelligent vehicle detects the emergent target for the first time, take θ r =θ(t);

[0025] When the onboard sensor of the intelligent vehicle detects an emergent target for the first time, take θ r =θ′(t), where θ′(t) is the detection angle when the onboard sensor of the intelligent vehicle detects the emergent target;

[0026] Calculate the time t at which the emerging target arrives at the conflict area r1 (θ r ), which is calculated as follows:

[0027]

[0028] Where W c is the width of the smart vehicle, v p The speed of emerging targets;

[0029] The conflict time difference is Δt, and its expression is:

[0030] Δt=t c1 (t)-t r1 (θ r )

[0031] The expression of the second driving risk field E(t)2 is as follows:

[0032]

[0033] Go to step 5.2;

[0034] Step 5: Risk warning;

[0035] Step 5.1: Assume that the probability of the emergent target impulsive travel is 0.5, and the vehicle speed is v c The risk value generated when (t) is equal to the preset vehicle speed and the lateral field of view width is equal to the preset width is the first risk threshold E thre (t) 1. Make judgments and risk warnings in the following ways:

[0036] When E(t)1≥E thre At (t)1, the intelligent vehicle moves at a speed v c (t) Driving is high risk, and the warning indicates high risk;

[0037] When E(t)1 <E thre At (t)1, the intelligent vehicle moves at a speed v c (t) The driving risk is low and no warning is required;

[0038] Step 5.2, let the vehicle speed v c The risk value generated when (t) is equal to the preset vehicle speed and the conflict time difference Δt is equal to the preset time difference is the second risk threshold E thre (t)2. Make judgments and risk warnings in the following ways:

[0039] When E(t)2≥E thre At (t)2, the intelligent vehicle moves at a speed v c (t) Driving is high risk, and the warning indicates high risk;

[0040] When E(t)2 <E thre At (t)2, the intelligent vehicle moves at a speed v c (t) The driving risk is low and no warning is required.

[0041] Preferably, the implementation process of step 2 is as follows:

[0042] Step 2.1: Take visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6 as nodes and record them as node A. i , i = 1, 2, 3, 4, 5, 6, establish the node attribute set A of the decision tree, A = {A1, A2, A3, A4, A5, A6};

[0043] According to the node attributes, each node A i are divided into V probability fluctuation threshold intervals;

[0044] Step 2.2, the probability of an emergent target appearing in the samples of the same probability fluctuation threshold interval is recorded as ó, and the V probability fluctuation threshold intervals are divided into high probability intervals, medium probability intervals and low probability intervals according to the value of ó, as follows:

[0045] When ó>0.1, high probability interval;

[0046] When 0.1≥ó≥0.01, medium probability interval;

[0047] When ó < 0.01, low probability interval;

[0048] Step 2.3, set the probability division rules for the emergence of emergent targets during the driving process of the intelligent vehicle, as follows:

[0049] If there is a node attribute in all nodes whose probability fluctuation threshold interval is a high probability interval, and the probability fluctuation threshold intervals of other node attributes are high probability intervals, medium probability intervals, or low probability intervals, then the probability of impulsive crossing of a sudden target during the driving of the intelligent vehicle is high probability;

[0050] If there is a node attribute in all nodes whose probability fluctuation threshold interval is in the medium probability interval, and the probability fluctuation threshold intervals of other node attributes are in the low probability interval or the medium probability interval, then the probability of impulsive crossing of the emergent target during the driving of the intelligent vehicle is medium probability;

[0051] If the probability fluctuation threshold interval of the node attributes of all nodes is a low probability interval, then the probability of impulsive crossing of the emergent target during the driving of the intelligent vehicle is low probability;

[0052] In step 2.4, the sample set D is used as the decision tree training set, and classification training is performed based on the C4.5 decision tree algorithm. The decision tree model generated by the training is used as the emergent target impulsive travel model.

[0053] Preferably, in step 2.1, node A is i The specific situation of being divided into V probability fluctuation threshold intervals is as follows:

[0054] The visibility A1 (km) is divided into four probability fluctuation threshold intervals, namely [0, 0.1], [0.1, 1], [1, 10], and [10, +∞];

[0055] The time period A2 is divided into 6 probability fluctuation threshold intervals, namely [0:00,7:00], [7:00,9:00], [9:00,12:00], [12:00,17:00], [17:00,19:00], and [19:00,24:00];

[0056] The passenger flow A3 (person / minute) is divided into five probability fluctuation threshold intervals, namely [0, 1], [1, 5], [5, 20], [20, 50], and [50, +∞];

[0057] The precipitation A4 (mm) is divided into five probability fluctuation threshold intervals, namely [0, 10], [10, 20], [20, 50], [50, 150], and [150, +∞];

[0058] The traffic volume A5 (vehicles / hour) is divided into four probability fluctuation threshold intervals, namely [0, 50], [50, 300], [300, 500], and [500, +∞];

[0059] The ambient temperature A6 (°C) is divided into five probability fluctuation threshold intervals, namely [-∞, 0], [0, 10], [10, 20], [20, 30], and [30, +∞].

[0060] Preferably, the classification training based on the C4.5 decision tree algorithm in step 2.4 includes:

[0061] Calculate the information gain rate Gr(D, A) of node attributes respectively i ), select the node feature with the largest information gain rate as the root node; in the branch nodes generated after the root node is divided, the branch nodes are divided by calculating the information gain rate; when the branch node and the parent node are of the same category, stop dividing and use the node as a leaf node.

[0062] Preferably, the information gain rate Gr(D,A i ) is calculated as follows:

[0063] Record any one of the high probability interval, medium probability interval and low probability interval defined in step 2.2 as the kth probability interval, from high to low, k = 1, 2, 3;

[0064] Calculate the information entropy E(D) of node attributes, and its expression is:

[0065]

[0066] Among them, p k Indicates the proportion of samples in different probability intervals in the sample set D;

[0067] Calculate the information gain G(D, A) of node attributes i ) and the dynamic gain of node attributes I(A i ), whose expressions are as follows:

[0068]

[0069] Wherein, v is any probability fluctuation threshold interval among V probability fluctuation threshold intervals, v = 1, 2, ..., V; |D v| represents the number of samples in the sample set D that are within the vth probability fluctuation threshold interval of the node attribute; E(D v ) is |D v |Information entropy;

[0070] Calculate the information gain rate Gr(D, A) of the node attribute i ), the calculation formula is:

[0071]

[0072] Preferably, the scenarios where the emergent target may appear include a smart vehicle, a current lane, a zebra crossing and a right roadside. The smart vehicle is traveling in the current lane, a stationary obstacle is parked on the right roadside in front of the smart vehicle, and there is a zebra crossing in front of the stationary obstacle. The emergent target may appear on the zebra crossing at any time.

[0073] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0074] 1. The present invention does not require the blind spot information obtained by third-party devices such as roadside stopped vehicle camera information and red light running monitoring processing devices in zebra crossing areas. The application scenario of the risk warning method of the present invention is not limited by third-party devices.

[0075] 2. Once the present invention finds that there is a blind spot in the field of vision, it starts the risk warning mode to remind the driver to drive defensively. It does not remind the driver only when the camera detects a prominent target. It can effectively deal with collision accidents in which the vehicle has no time to brake when it detects an obstacle, especially when the vehicle is driving at a high speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 is a top view of a pop-up target scene in an embodiment of the present invention;

[0077] Figure 2 is a diagram of the installation positions of the on-board sensors of the intelligent vehicle in an embodiment of the present invention;

[0078] Figure 3 is a flow chart of a driving risk monitoring method for an emergent target in an embodiment of the present invention;

[0079] Figure 4 is a scene graph in which the vehicle-mounted sensor does not detect an emergent target at time t in an embodiment of the present invention;

[0080] Figure 5 It is a scene graph of a sudden target detected by the vehicle-mounted sensor at time t in an embodiment of the present invention. DETAILED DESCRIPTION

[0081] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0082] Figure 1 is a top view of a scene where the emergent target described in the embodiment of the present invention may appear. Figure 1 It can be seen that the scenarios involved in the driving risk monitoring method include the current lane, zebra crossing and right roadside. The smart vehicle is driving in the current lane. There is a stationary obstacle parked on the right roadside in front of the smart vehicle. There is a zebra crossing in front of the stationary obstacle. Sudden targets may appear on the zebra crossing at any time.

[0083] The emergent target is defined as a traffic participant that appears suddenly and poses a significant threat to the safe driving of the vehicle, including but not limited to suddenly appearing pedestrians, animals and non-motor vehicles.

[0084] Figure 2 FIG. 1 is a diagram showing the installation positions of the onboard sensors of the intelligent vehicle in the embodiment of the present invention. Figure 2 It can be seen that the smart vehicle is equipped with an on-board sensor, which is installed at the center of the front bumper of the smart vehicle.

[0085] The visual blind spot environment is caused by a stationary obstacle blocking the field of view of the on-board sensor of the smart vehicle, including but not limited to a visual blind spot caused by a stationary obstacle on the right side of the road and a visual blind spot caused by a stationary obstacle on the left side of the road.

[0086] Figure 3 : is a flow chart of a driving risk monitoring method for an emergent target in an embodiment of the present invention. As can be seen from the figure, the driving risk monitoring method includes establishing an emergent target impulsive travel model, predicting the probability state of an emergent target impulsive travel ahead through the emergent target impulsive travel model; constructing a driving risk field in combination with the probability state of the emergent target impulsive travel, judging the driving risk according to the driving risk field value and taking corresponding early warning strategies. The specific steps are as follows:

[0087] Step 1, obtain historical environmental data of the scene where the emergent target may appear, including visibility A1, time period A2, pedestrian flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6, and generate a sample set D.

[0088] Step 2: Establish a decision tree model, use the sample set D as the decision tree training set, perform classification training based on the C4.5 decision tree algorithm, and obtain the emergent target impulsive travel model.

[0089] Step 3, obtaining real-time environmental data including visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6, and inputting the emergent target impulsive travel model obtained in step 2, and outputting the probability state of the emergent target impulsive travel;

[0090] The probability state of the emergent target impulsive travel includes three states, which are respectively recorded as a high probability state, a medium probability state and a low probability state.

[0091] Step 4: Given the value of the probability δ of the emergent target impulsive travel, a driving risk field is constructed in combination with the probability δ of the emergent target impulsive travel, the driving risk is judged according to the driving risk field value, and a corresponding early warning strategy is adopted.

[0092] Step 4.1, the vehicle-mounted sensor obtains vehicle status information and real-time environmental information;

[0093] The vehicle-mounted sensor is installed at the center of the front bumper of the smart vehicle; the vehicle status information includes the speed and position of the smart vehicle.

[0094] Step 4.2, giving the value of the probability δ: high probability state, δ = 0.8; medium probability state, δ = 0.5; low probability state, δ = 0.2.

[0095] Step 4.3, record the current time when the intelligent vehicle is driving as time t, and establish the driving risk field at time t;

[0096] Step 4.3.1: The time when the front of the intelligent vehicle reaches the conflict area at time t is recorded as t c1 (t), which is expressed as follows:

[0097]

[0098] The conflict area refers to the line segment in the emerging target walking route that overlaps with the intelligent vehicle driving route. l is the straight-line distance from the front of the intelligent vehicle to the conflict area along the vehicle's driving direction at time t, v c (t) is the speed of the smart vehicle at time t;

[0099] Step 4.3.2, determine whether the on-board sensor of the intelligent vehicle detects an emergent target at time t. If no emergent target is detected, execute step 4.3.3. If an emergent target is detected, jump to step 4.3.4;

[0100] Step 4.3.3, establish the first driving risk field of the intelligent vehicle in the visual blind spot environment at time t, and record it as the first driving risk field E(t)1, which is expressed as follows:

[0101]

[0102] Among them, θ(t) is the blind spot detection angle of the vehicle-mounted sensor, and d1×tanθ(t) is the width of the lateral field of view of the intelligent vehicle.

[0103] Go to step 5.1;

[0104] Step 4.3.4, establish the second driving risk field of the intelligent vehicle in the visual blind spot environment at time t, and record it as the second driving risk field E(t)2;

[0105] The calculated angle when the onboard sensor of the intelligent vehicle detects the emergent target is θ r , select as follows:

[0106] When the onboard sensor of the intelligent vehicle detects the emergent target for the first time, take θ r =θ(t);

[0107] When the onboard sensor of the intelligent vehicle detects an emergent target for the first time, take θ r =θ′(t), where θ′(t) is the detection angle when the onboard sensor of the intelligent vehicle detects the emergent target;

[0108] Calculate the time t at which the emerging target arrives at the conflict area r1 (θ r ), which is calculated as follows:

[0109]

[0110] Where W c is the width of the smart vehicle, v p The speed of the emerging target;

[0111] The conflict time difference is Δt, and its expression is:

[0112] Δt=t c1 (t)-t r1 (θ r )

[0113] The expression of the second driving risk field E(t)2 is as follows:

[0114]

[0115] Proceed to step 5.2.

[0116] Step 5: Risk warning;

[0117] Step 5.1: Assume that the probability of the emergent target impulsive travel is 0.5, and the vehicle speed is v c The risk value generated when (t) is equal to the preset vehicle speed and the lateral field of view width is equal to the preset width is the first risk threshold E thre (t) 1. Make judgments and risk warnings in the following ways:

[0118] When E(t)1≥E thre At (t)1, the intelligent vehicle moves at a speed v c (t) Driving is high risk, and the warning indicates high risk;

[0119] When E(t)1 <E thre At (t)1, the intelligent vehicle moves at a speed v c (t) The driving risk is low and no warning is required.

[0120] Assume that the preset vehicle speed is 10m / s and the preset width is 4m.

[0121] Step 5.2, let the vehicle speed v c The risk value generated when (t) is equal to the preset vehicle speed and the conflict time difference Δt is equal to the preset time difference is the second risk threshold E thre (t)2. Make judgments and risk warnings in the following ways:

[0122] When E(t)2≥E thre At (t)2, the intelligent vehicle moves at a speed v c (t) Driving is high risk, and the warning indicates high risk;

[0123] When E(t)2 <E thre At (t)2, the intelligent vehicle moves at a speed v c (t) The driving risk is low and no warning is required.

[0124] Assume that the preset vehicle speed is 10m / s and the preset time difference is 1.2s.

[0125] Figure 4 This is a scene diagram of a vehicle sensor not detecting an emergent target in a visual blind spot environment at time t in an embodiment of the present invention. Figure 4 In , line segment CD is the emerging target walking route, and line segment AB is the conflict area. Figure 4 In the figure, point 0 is the position of the on-board sensor, point 0 is connected with the left front corner of the static obstacle and extended to intersect with line segment CD at point C, and the angle θ(t) between line segment OC and the speed direction of the controlled vehicle is the blind spot detection angle of the on-board sensor of the controlled vehicle at time t. Figure 5 is a scene graph of a sudden target detected by the vehicle-mounted sensor at time t in an embodiment of the present invention;

[0126] In this embodiment, the implementation process of step 2 is as follows:

[0127] Step 2.1: Take visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6 as nodes and record them as node A. i , i=1,2,3,4,5,6, establish the node attribute set A of the decision tree, A={A1,A2,A3,A4,A5,A6}.

[0128] According to the node attributes, each node A iThey are divided into V probability fluctuation threshold intervals, the details are as follows:

[0129] The visibility A1 (km) is divided into four probability fluctuation threshold intervals, namely [0, 0.1], [0.1, 1], [1, 10], and [10, +∞];

[0130] The time period A2 is divided into 6 probability fluctuation threshold intervals, namely [0:00,7:00], [7:00,9:00], [9:00,12:00], [12:00,17:00], [17:00,19:00],

[0131] [19:00,24:00];

[0132] The passenger flow A3 (person / minute) is divided into five probability fluctuation threshold intervals, namely [0, 1], [1, 5], [5, 20], [20, 50], and [50, +∞];

[0133] The precipitation A4 (mm) is divided into five probability fluctuation threshold intervals, namely [0, 10], [10, 20], [20, 50], [50, 150], and [150, +∞];

[0134] The traffic volume A5 (vehicles / hour) is divided into four probability fluctuation threshold intervals, namely [0, 50], [50, 300], [300, 500], and [500, +∞];

[0135] The ambient temperature A6 (°C) is divided into five probability fluctuation threshold intervals, namely [-∞, 0], [0, 10], [10, 20], [20, 30], and [30, +∞].

[0136] Step 2.2, the probability of an emergent target appearing in the samples of the same probability fluctuation threshold interval is recorded as ó, and the V probability fluctuation threshold intervals are divided into high probability intervals, medium probability intervals and low probability intervals according to the value of ó, as follows:

[0137] When ó>0.1, high probability interval;

[0138] When 0.1≥ó≥0.01, medium probability interval;

[0139] When ó<0.01, low probability interval.

[0140] Step 2.3, set the probability division rules for the emergence of emergent targets during the driving process of the intelligent vehicle, as follows:

[0141] If there is a node attribute in all nodes whose probability fluctuation threshold interval is a high probability interval, and the probability fluctuation threshold intervals of other node attributes are high probability intervals, medium probability intervals, or low probability intervals, then the probability of impulsive crossing of a sudden target during the driving of the intelligent vehicle is high probability;

[0142] If there is a node attribute in all nodes whose probability fluctuation threshold interval is in the medium probability interval, and the probability fluctuation threshold intervals of other node attributes are in the low probability interval or the medium probability interval, then the probability of impulsive crossing of the emergent target during the driving of the intelligent vehicle is medium probability;

[0143] If the probability fluctuation threshold interval of the node attributes of all nodes is a low probability interval, the probability of impulsive crossing of a sudden target during the driving of the intelligent vehicle is low probability.

[0144] In step 2.4, the sample set D is used as the decision tree training set, and classification training is performed based on the C4.5 decision tree algorithm. The decision tree model generated by the training is used as the emergent target impulsive travel model.

[0145] The classification training based on the C4.5 decision tree algorithm includes:

[0146] Calculate the information gain rate Gr(D, A) of node attributes respectively i ), select the node feature with the largest information gain rate as the root node; in the branch nodes generated after the root node is divided, the branch nodes are divided by calculating the information gain rate; when the branch node and the parent node are of the same category, stop dividing and use the node as a leaf node.

[0147] The information gain rate Gr(D, A) of the node attribute i ) is calculated as follows:

[0148] Record any one of the high probability interval, medium probability interval and low probability interval defined in step 2.2 as the kth probability interval, from high to low, k = 1, 2, 3.

[0149] Calculate the information entropy E(D) of node attributes, and its expression is:

[0150]

[0151] Among them, p k It represents the proportion of samples in different probability intervals in the sample set D.

[0152] Calculate the information gain G(D, A) of node attributes i ) and the dynamic gain of node attributes I(A i ), whose expressions are as follows:

[0153]

[0154] Wherein, v is any probability fluctuation threshold interval among V probability fluctuation threshold intervals, v = 1, 2, ..., V; |D v | represents the number of samples in the sample set D that are within the vth probability fluctuation threshold interval of the node attribute; E(D v ) is |D v |Information entropy.

[0155] Calculate the information gain rate Gr(D, A) of the node attribute i ), the calculation formula is:

[0156]

Claims

1. A driving risk monitoring method for emergent targets, wherein the emergent targets are defined as traffic participants that suddenly appear and pose a significant threat to the safe driving of vehicles, including but not limited to suddenly appearing pedestrians, animals and non-motor vehicles; characterized in that: The driving risk monitoring method includes establishing an emergent target impulsive travel model, predicting the probability state of the emergent target impulsive travel ahead through the emergent target impulsive travel model; constructing a driving risk field in combination with the probability state of the emergent target impulsive travel, judging the driving risk according to the driving risk field value and taking corresponding early warning strategies, and the specific steps are as follows: Step 1, obtaining historical environmental data of scenes where the emergent target may appear, including visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6, and generating a sample set D; Step 2: Establish a decision tree model, use the sample set D as the decision tree training set, perform classification training based on the C4.5 decision tree algorithm, and obtain an emergent target impulsive travel model; Step 3, obtaining real-time environmental data including visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6, and inputting the emergent target impulsive travel model obtained in step 2, and outputting the probability state of the emergent target impulsive travel; The probability state of the emergent target impulsive travel includes three states, which are respectively recorded as a high probability state, a medium probability state and a low probability state; Step 4: Given a value of the probability δ of an emergent target impulsive travel, a driving risk field is constructed in combination with the probability δ of the emergent target impulsive travel, the driving risk is determined according to the value of the driving risk field, and a corresponding early warning strategy is adopted; Step 4.1, the vehicle-mounted sensor obtains vehicle status information and real-time environmental information; The vehicle-mounted sensor is installed at the center of the front bumper of the smart vehicle; the vehicle status information includes the speed and position of the smart vehicle; Step 4.2, giving the value of the probability δ: high probability state, δ = 0.8; Medium probability state, δ = 0.5; low probability state, δ = 0.2; Step 4.3, record the current time when the intelligent vehicle is driving as time t, and establish the driving risk field at time t; Step 4.3.1: The time when the front of the intelligent vehicle reaches the conflict area at time t is recorded as t c1 (t), which is expressed as follows: The conflict area refers to the line segment in the emerging target walking route that overlaps with the intelligent vehicle driving route. l is the distance between the front of the intelligent vehicle and the conflict area along the vehicle's driving direction at time t, v c (t) is the speed of the smart vehicle at time t; Step 4.3.2, determine whether the on-board sensor of the intelligent vehicle detects an emergent target at time t. If no emergent target is detected, execute step 4.3.

3. If an emergent target is detected, jump to step 4.3.4; Step 4.3.3, establish the first driving risk field of the intelligent vehicle in the visual blind spot environment at time t, and record it as the first driving risk field E(t)1, which is expressed as follows: Among them, θ(t) is the blind spot detection angle of the vehicle-mounted sensor, and d1×tanθ(t) is the width of the lateral field of view of the intelligent vehicle; Go to step 5.1; Step 4.3.4, establish the second driving risk field of the intelligent vehicle in the visual blind spot environment at time t, and record it as the second driving risk field E(t)2; The calculated angle when the onboard sensor of the intelligent vehicle detects the emergent target is θ r , select as follows: When the onboard sensor of the intelligent vehicle detects the emergent target for the first time, take θ r =θ(t); When the onboard sensor of the intelligent vehicle detects an emergent target for the first time, take θ r =θ′(t), where θ′(t) is the detection angle when the onboard sensor of the intelligent vehicle detects the emergent target; Calculate the time t at which the emerging target arrives at the conflict area r1 (θ r ), which is calculated as follows: Where W c is the width of the smart vehicle, v p The speed of emerging targets; The conflict time difference is Δt, and its expression is: Δt=t c1 (t)-t r1 (θ r ) The expression of the second driving risk field E(t)2 is as follows: Go to step 5.2; Step 5: Risk warning; Step 5.1: Assume that the probability of the emergent target impulsive travel is 0.5, and the vehicle speed is v c The risk value generated when (t) is equal to the preset vehicle speed and the lateral field of view width is equal to the preset width is the first risk threshold E thre (t) 1. Make judgments and risk warnings in the following ways: When E(t)1≥E thre At (t)1, the intelligent vehicle moves at a speed v c (t) Driving is high risk, and the warning indicates high risk; When E(t)1 <E thre At (t)1, the intelligent vehicle moves at a speed v c (t) The driving risk is low and no warning is required; Step 5.2, let the vehicle speed v c The risk value generated when (t) is equal to the preset vehicle speed and the conflict time difference Δt is equal to the preset time difference is the second risk threshold E thre (t)2. Make judgments and risk warnings in the following ways: When E(t)2≥E thre At (t)2, the intelligent vehicle moves at a speed v c (t) Driving is high risk, and the warning indicates high risk; When E(t)2 <E thre At (t)2, the intelligent vehicle moves at a speed v c (t) The driving risk is low and no warning is required.

2. A method for monitoring driving risk of an emerging target according to claim 1, characterized in that: The implementation process of step 2 is as follows: Step 2.1: Take visibility A1, time period A2, passenger flow A3, precipitation A4, vehicle flow A5, and ambient temperature A6 as nodes and record them as node A. i , i = 1, 2, 3, 4, 5, 6, establish the node attribute set A of the decision tree, A = {A1, A2, A3, A4, A5, A6}; According to the node attributes, each node A i are divided into V probability fluctuation threshold intervals; Step 2.2, the probability of an emergent target appearing in the samples of the same probability fluctuation threshold interval is recorded as ó, and the V probability fluctuation threshold intervals are divided into high probability intervals, medium probability intervals and low probability intervals according to the value of ó, as follows: When ó>0.1, high probability interval; When 0.1≥ó≥0.01, medium probability interval; When ó < 0.01, low probability interval; Step 2.3, set the probability division rules for the emergence of emergent targets during the driving process of the intelligent vehicle, as follows: If there is a node attribute in all nodes whose probability fluctuation threshold interval is a high probability interval, and the probability fluctuation threshold intervals of other node attributes are high probability intervals, medium probability intervals, or low probability intervals, then the probability of impulsive crossing of a sudden target during the driving of the intelligent vehicle is high probability; If there is a node attribute in all nodes whose probability fluctuation threshold interval is in the medium probability interval, and the probability fluctuation threshold intervals of other node attributes are in the low probability interval or the medium probability interval, then the probability of impulsive crossing of the emergent target during the driving of the intelligent vehicle is medium probability; If the probability fluctuation threshold interval of the node attributes of all nodes is a low probability interval, then the probability of impulsive crossing of the emergent target during the driving of the intelligent vehicle is low probability; In step 2.4, the sample set D is used as the decision tree training set, and classification training is performed based on the C4.5 decision tree algorithm. The decision tree model generated by the training is used as the emergent target impulsive travel model.

3. A method for monitoring driving risk of an emerging target according to claim 2, characterized in that: Step 2.1: Node A is converted according to the node attributes i The specific situation of being divided into V probability fluctuation threshold intervals is as follows: The visibility A1 (km) is divided into four probability fluctuation threshold intervals, namely [0, 0.1], [0.1, 1], [1, 10], and [10, +∞]; The time period A2 is divided into 6 probability fluctuation threshold intervals, namely [0:00,7:00], [7:00,9:00], [9:00,12:00], [12:00,17:00], [17:00,19:00], and [19:00,24:00]; The passenger flow A3 (person / minute) is divided into five probability fluctuation threshold intervals, namely [0, 1], [1, 5], [5, 20], [20, 50], and [50, +∞]; The precipitation A4 (mm) is divided into five probability fluctuation threshold intervals, namely [0, 10], [10, 20], [20, 50], [50, 150], and [150, +∞]; The traffic volume A5 (vehicles / hour) is divided into four probability fluctuation threshold intervals, namely [0, 50], [50, 300], [300, 500], and [500, +∞]; The ambient temperature A6 (°C) is divided into five probability fluctuation threshold intervals, namely [-∞, 0], [0, 10], [10, 20], [20, 30], and [30, +∞].

4. The method for monitoring driving risk of an emerging target according to claim 2, characterized in that: The classification training based on the C4.5 decision tree algorithm described in step 2.4 includes: Calculate the information gain rate Gr(D,A) of node attributes respectively i ), select the node feature with the largest information gain rate as the root node; in the branch nodes generated after the root node is divided, the branch nodes are divided by calculating the information gain rate; when the branch node and the parent node are of the same category, stop dividing and use the node as a leaf node.

5. The method for monitoring driving risk of an emerging target according to claim 4, characterized in that: The information gain rate Gr(D, A) of the node attribute i ) is calculated as follows: Record any one of the high probability interval, medium probability interval and low probability interval defined in step 2.2 as the kth probability interval, from high to low, k = 1, 2, 3; Calculate the information entropy E(D) of node attributes, and its expression is: Among them, p k Indicates the proportion of samples in different probability intervals in the sample set D; Calculate the information gain G(D, A) of node attributes i ) and the dynamic gain of node attributes I(A i ), whose expressions are as follows: Wherein, v is any probability fluctuation threshold interval among V probability fluctuation threshold intervals, v = 1, 2, ..., V; |D v | represents the number of samples in the sample set D that are within the vth probability fluctuation threshold interval of the node attribute; E(D v ) is |D v |Information entropy; Calculate the information gain rate Gr(D, A) of the node attribute i ), the calculation formula is:

6. The method for monitoring driving risk of an emerging target according to claim 1, characterized in that: The scenarios where the emergent target may appear include the smart vehicle, the current lane, the zebra crossing and the right side of the road. The smart vehicle is traveling in the current lane, a stationary obstacle is parked on the right side of the road in front of the smart vehicle, and there is a zebra crossing in front of the stationary obstacle. The emergent target may appear on the zebra crossing at any time.

Citation Information

Patent Citations

  • Systems, devices, and methods for predictive risk aware driving

    CN114248800A

  • Dangerous event chain extraction method and system based on complex traffic scene

    CN115238958A

  • Unmanned vehicle control method and system based on pedestrian existence prediction in sidewalk blind area

    CN116639151A

  • Macroscopic area vehicle early warning method and system based on safety risk field

    CN118116236A

  • Intelligent vehicle defensive driving behavior modeling method

    CN118419007A