Vehicle steering lamp self-starting method and system

Abstract

The invention relates to the technical field of intelligent auxiliary driving, and provides a vehicle steering lamp self-starting method and system.Fusion algorithms such as a Bayesian network and a Kalman filter are built in a steering decision model based on a deep learning algorithm, and when sight line features, steering wheel features and vehicle speed features are extracted, the vehicle steering lamp self-starting method and system are provided. According to reliability and correlation of different data, probability relations among various factors are considered, weights of the factors in a final result are dynamically adjusted, different turning intentions of a user are represented and predicted through a probability model, and therefore a more reliable prediction result is provided so as to improve accuracy and reliability of turning intention recognition. Through cooperative work of a driver monitoring system and rotation monitoring (vehicle speed data and steering wheel data), the steering lamp can be automatically turned on according to the sight direction intention of a driver, so that the operation convenience and automation level are improved, traffic accidents caused by misoperation of the steering lamp are reduced, and road traffic safety is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of intelligent assisted driving technology, and in particular to a vehicle turn signal self-starting method and system. Background Art

[0002] In modern traffic environments, vehicle turn signals are a vital safety device. Their main function is to inform other road users (including vehicle drivers, pedestrians, etc.) that the vehicle is about to change direction. For example, when the vehicle is turning, changing lanes, or pulling over, turning on the turn signal in advance can give surrounding traffic participants enough time to react accordingly, thereby avoiding traffic accidents.

[0003] However, the operation of vehicle turn signals currently relies mainly on manual activation by the driver, which has many defects. On the one hand, manually operated turn signals require the driver to pay extra attention to operate the turn signal control device while performing driving operations (such as turning or changing lanes), which affects driving efficiency to a certain extent. For example, in complex traffic conditions (such as busy intersections or emergency lane changes when driving fast), the driver may forget or delay turning on the turn signal because of his focus on the road conditions, which may mislead other road users and increase the risk of traffic accidents. On the other hand, since the turn signal operation relies entirely on the driver's subjective consciousness, some drivers may not use the turn signal due to negligence or bad driving habits, which also poses a serious threat to traffic safety.

[0004] In summary, the existing manual operation method of vehicle turn signals is inefficient and easily ignored by drivers, and there are traffic safety problems caused by improper operation of turn signals. Summary of the invention

[0005] The present invention provides a vehicle turn signal self-starting method and system, which solves the technical problems that the existing manual operation mode of vehicle turn signal is inefficient and easily ignored by the driver, and there is a potential traffic safety hazard due to improper operation of the turn signal.

[0006] In order to solve the above technical problems, the present invention provides a vehicle turn signal self-starting method, comprising:

[0007] Obtain user status information in real time based on the driver monitoring system;

[0008] Collect vehicle status information of the current vehicle in real time;

[0009] Building a steering decision model based on a deep learning algorithm, inputting the user state information and the vehicle state information into the steering decision model to perform feature fusion processing and classification prediction to obtain the user's steering intention;

[0010] Whether to execute the turn signal automatic start control is determined according to the user's turning intention.

[0011] In a further embodiment, obtaining user status information in real time based on a driver monitoring system includes:

[0012] Divide the space around the vehicle into regions and obtain target regions corresponding to different steering intentions;

[0013] Based on the driver monitoring system, the user's head image is collected in real time, image recognition is performed to realize line of sight tracking, and area matching is performed with the target area to determine the user's line of sight information;

[0014] The user's line of sight information includes one or more of a line of sight angle, a line of sight dwelling time, a line of sight moving speed, and a line of sight moving direction.

[0015] This solution builds a driver's gaze tracking process based on the gaze interaction function of the DMS system, so as to accurately track the driver's eye movement during vehicle driving, identify the driver's gaze direction in real time by analyzing the user's gaze point and gaze time, and automatically start the turn signal with contactless automatic detection and identification, completely freeing both hands and thus improving driving safety; at the same time, the space around the vehicle is divided into areas for gaze tracking, and target areas corresponding to different turning intentions are obtained, so that the user's intention can be quickly determined by matching the gaze with the target area (such as whether to look at the left or right rearview mirror, and the left or right front area, so as to obtain information related to the driver's possible turning or lane changing intention), further improving the sensitivity of the device.

[0016] In a further embodiment, it also includes: using infrared sensors to perform pupil positioning and corneal reflection point detection to assist image recognition in achieving line of sight tracking; and / or, obtaining the user's head posture and using a deep learning algorithm to estimate the three-axis posture of the user's head, and then dynamically compensating for the line of sight tracking achieved by image recognition in combination with the three-axis posture.

[0017] When performing eye tracking, this solution simultaneously incorporates other auxiliary means, including but not limited to using infrared sensors to perform pupil positioning and corneal reflection point detection to assist image recognition in achieving eye tracking; and / or, obtaining the user's head posture and using a deep learning algorithm to estimate the three-axis posture of the user's head, and then dynamically compensating the eye tracking achieved by image recognition in combination with the three-axis posture, thereby improving the accuracy and reliability of the eye tracking module.

[0018] In a further implementation scheme, the vehicle status information of the current vehicle is collected in real time, specifically: the vehicle body domain control system is used to continuously monitor the vehicle and obtain corresponding vehicle body information, and the vehicle speed data and steering wheel data are obtained from the vehicle body information as the vehicle status information of the current vehicle.

[0019] This solution uses the body domain control system to continuously monitor the vehicle and obtain body information such as speed data and steering wheel data. Based on the changing parameters of the equipment on the vehicle, it further identifies the user's steering intention from the on-board operation, combines the user status information and the vehicle status information for mutual verification, and realizes multi-factor fusion intention recognition with higher recognition accuracy.

[0020] In a further embodiment, the steering decision model includes an input layer, a convolutional layer, a pooling layer, a fully connected layer, and an activation function;

[0021] The input layer is used to normalize the user state information and the vehicle state information to obtain corresponding input features;

[0022] The convolution layer is used to perform a convolution operation on the input features through a plurality of convolution kernels to realize feature extraction to obtain corresponding sight features, steering wheel features and vehicle speed features;

[0023] The pooling layer is used to downsample the sight features, steering wheel features and vehicle speed features to reduce feature dimensions and output a corresponding feature sequence;

[0024] The fully connected layer has a built-in fusion algorithm for performing feature fusion processing and classification prediction according to the sight features, steering wheel features and vehicle speed features to obtain the user's steering intention.

[0025] This solution uses a deep learning algorithm to establish a steering decision model. Through feature extraction, feature fusion, classification prediction and other processes, it can adapt to different users, with strong pertinence and good adaptability. The deep learning algorithm can learn the user's preferences and behavior patterns, so as to more accurately predict the user's next operation, thereby ensuring the accuracy of the prediction of the user's steering intention and providing a good turn signal self-start service.

[0026] In a further embodiment, when the fusion algorithm is a Bayesian network, feature fusion processing and classification prediction are performed according to the sight features, steering wheel features and vehicle speed features to obtain the user's steering intention, including:

[0027] Classify scenes based on different driving scenarios to set corresponding steering intentions;

[0028] The sight line features, steering wheel features and vehicle speed features obtained according to the user state information and the vehicle state information are used as observation variables, and the prior probability corresponding to each turning intention is calculated in combination with the Bayesian theorem;

[0029] The prior probabilities of all the turning intentions are compared, and the turning intention with the highest probability value is taken as the classification prediction result and output.

[0030] This solution has built-in fusion algorithms, such as Bayesian networks and Kalman filters, in the steering decision model based on deep learning algorithms. When extracting line of sight features, steering wheel features, and vehicle speed features, it considers the probabilistic relationship between various factors based on the reliability and relevance of different data, and dynamically adjusts their weights in the final result. Then, the probabilistic model is used to characterize and predict different user steering intentions, thereby providing more reliable prediction results to improve the accuracy and reliability of steering intention recognition.

[0031] In a further embodiment, determining whether to execute the turn signal self-start control according to the turning intention of the user includes:

[0032] The user's turning intention and the corresponding prior probability are obtained from the turning decision model, and it is determined whether the prior probability is greater than a preset threshold. If so, it is confirmed that the turn signal self-start control is executed, and then the corresponding control instruction is generated based on the turning intention to drive the corresponding left / right turn signal to start.

[0033] Based on the actual application environment, this solution sets a probability threshold as a preset threshold. By comparing whether the prior probability of the output turning intention is greater than the preset threshold, it is determined whether to execute the turn signal self-start control, thereby reducing the probability of false triggering of the turn signal and improving driving safety.

[0034] In a further embodiment, the steering decision model is further trained, and the model training includes:

[0035] Collect driving data and annotate the data, including sight features, steering wheel features, vehicle speed features and corresponding steering intention labels;

[0036] Divide the driving data into training set, validation set and test set according to the preset ratio;

[0037] Use the training set data to train the steering decision model, optimize the model parameters θ and minimize the loss function through the back propagation algorithm;

[0038] The trained steering decision model is evaluated using a test set, wherein the evaluation criteria of the model evaluation include one or more indicators of accuracy, recall, and F1 value.

[0039] This solution collects a large amount of labeled driving data for training. The availability of big data enables the model to learn the underlying patterns and reduce overfitting, thereby being able to handle more complex input data and task scenarios, achieving higher prediction accuracy and classification effects.

[0040] The present invention also provides a vehicle turn signal self-starting system, comprising a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a vehicle turn signal self-starting method as described above is implemented.

[0041] In a further embodiment, a data acquisition module connected to the processor is further included, wherein the data acquisition module includes an image acquisition module, a vehicle speed acquisition module and a turning angle acquisition unit;

[0042] The image acquisition module is used to collect user status information in real time;

[0043] A vehicle speed collection unit, used for collecting vehicle speed data;

[0044] The turning angle acquisition unit is used to collect the steering wheel data of the vehicle.

[0045] This solution is based on the vehicle's own driver monitoring system and body domain control system, and defines an image acquisition module, a vehicle speed acquisition module, and a turning angle acquisition unit for data acquisition. It does not require the installation of additional equipment and can be adapted to different types of vehicles, with good applicability.

[0046] The beneficial effects of the present invention are as follows:

[0047] 1. Improve the automation of turn signal operation

[0048] The present invention can automatically turn on the turn signal according to the driver's line of sight through the coordinated work of the driver monitoring system (DMS) and the rotation monitoring (vehicle speed data and steering wheel data). This greatly reduces the number of manual operations by the driver and improves the convenience and automation level of the operation.

[0049] 2. Enhance traffic safety

[0050] By reducing the manual operation of the turn signal, the driver can avoid the situation where the turn signal is not turned on due to forgetfulness or untimely operation. When the vehicle turns or changes lanes, correctly turning on the turn signal is crucial to convey the driving intention to other road users. This technology can effectively reduce traffic accidents caused by improper operation of the turn signal and ensure road traffic safety.

[0051] 3. Improve the comfort of driving experience

[0052] The driver does not need to be distracted to manually operate the turn signal, especially in complex road conditions that require frequent turns or lane changes. He can focus more on driving operations, such as controlling vehicle speed and maintaining distance between vehicles. By realizing automated turn signal operation, the driver's operating burden is reduced and the driving experience is more comfortable.

[0053] 4. Improve the accuracy of turning intention recognition

[0054] A variety of data fusion algorithms are used, such as Kalman filter, Bayesian network and other fusion algorithms, which can dynamically adjust the weights according to the reliability and relevance of different data. In complex scenarios, machine learning or deep learning algorithms can also be used to train classifiers (such as support vector machines, decision trees or neural networks) to distinguish different turning intentions. This effectively improves the recognition accuracy of the driver's turning intention, reduces the possibility of misjudgment, and further improves the reliability of the automatic turn signal function. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 This is a system flow chart of a vehicle turn signal self-starting method provided by an embodiment of the present invention;

[0056] Figure 2 is a schematic diagram of algorithm execution provided by an embodiment of the present invention;

[0057] Figure 3 is a sight range area map provided by an embodiment of the present invention;

[0058] Figure 4 It is a schematic diagram of sight line acquisition provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following specifically illustrates the implementation mode of the present invention in conjunction with the accompanying drawings. The embodiments are provided for illustrative purposes only and cannot be understood as limiting the present invention. The accompanying drawings are provided for reference and illustration only and do not constitute a limitation on the scope of patent protection of the present invention, because many changes can be made to the present invention without departing from the spirit and scope of the present invention.

[0060] An embodiment of the present invention provides a method for automatically starting a vehicle turn signal lamp, such as Figure 1 to Figure 4 As shown, in this embodiment, it includes:

[0061] S1. Acquire user status information in real time based on the driver monitoring system; collect vehicle status information of the current vehicle in real time;

[0062] In this embodiment, the user status information is obtained in real time based on the driver monitoring system, including:

[0063] a1. Divide the space around the vehicle into regions and obtain target regions corresponding to different steering intentions;

[0064] See also Figure 3 The space around the vehicle is divided into regions based on the driver's field of vision, and the target areas obtained include the forward field of vision area, the left forward field of vision area, the right forward field of vision area, the left rearview mirror field of vision area, and the right rearview mirror field of vision area.

[0065] a2. Based on the driver monitoring system, the user's head image is collected in real time, image recognition is performed to realize line of sight tracking, and area matching is performed with the target area to determine the user's line of sight information;

[0066] The user's line of sight information includes one or more of a line of sight angle, a line of sight dwelling time, a line of sight moving speed, and a line of sight moving direction.

[0067] See also Figure 4 The diagram of line of sight collection is a schematic diagram of a driver's line of sight tracking mechanism built based on the line of sight interaction function of the DMS system. Then, during vehicle driving, the driver's eye movement is accurately tracked and gaze point detection is performed, and the driver's line of sight focus direction is identified in real time, such as whether he is looking at the left or right rearview mirror, and the left or right front area.

[0068] This solution builds a driver's gaze tracking process based on the gaze interaction function of the DMS system, so as to accurately track the driver's eye movement during vehicle driving, identify the driver's gaze direction in real time by analyzing the user's gaze point and gaze time, and automatically start the turn signal with contactless automatic detection and identification, completely freeing both hands and thus improving driving safety; at the same time, the space around the vehicle is divided into areas for gaze tracking, and target areas corresponding to different turning intentions are obtained, so that the user's intention can be quickly determined by matching the gaze with the target area (such as whether to look at the left or right rearview mirror, and the left or right front area, so as to obtain information related to the driver's possible turning or lane changing intention), further improving the sensitivity of the device.

[0069] In this embodiment, the collection of user status information also includes:

[0070] Fusion of data from other sensors, such as the use of infrared sensors to perform pupil localization and corneal reflection point detection under low-light conditions, can assist image recognition in achieving line of sight tracking. Among them, combining multi-task learning to simultaneously perform pupil localization and corneal reflection point detection can improve the accuracy of line of sight estimation.

[0071] And / or, when the driver's head moves freely, the system can dynamically compensate for the line of sight estimation, obtain the user's head posture and use a deep learning algorithm to estimate the three-axis posture (Yaw, Pitch, Roll) of the user's head, and then dynamically compensate for the line of sight tracking achieved by image recognition in combination with the three-axis posture (i.e., adjust the estimation of the line of sight direction according to these posture information).

[0072] When performing eye tracking, this solution simultaneously incorporates other auxiliary means, including but not limited to using infrared sensors to perform pupil positioning and corneal reflection point detection to assist image recognition in achieving eye tracking; and / or, obtaining the user's head posture and using a deep learning algorithm to estimate the three-axis posture of the user's head, and then dynamically compensating the eye tracking achieved by image recognition in combination with the three-axis posture, thereby improving the accuracy and reliability of the eye tracking module.

[0073] In this embodiment, the vehicle status information of the current vehicle is collected in real time, specifically: the vehicle body domain control system is used to continuously monitor the vehicle and obtain corresponding vehicle body information, and the vehicle speed data and steering wheel data are obtained from the vehicle body information as the vehicle status information of the current vehicle.

[0074] Among them, the body domain control system is responsible for continuously monitoring the vehicle's rotation during driving, accurately detecting data such as the steering angle of the steering wheel and the vehicle's driving speed, and providing necessary data support for judging the driver's steering intention.

[0075] This solution uses the body domain control system to continuously monitor the vehicle and obtain body information such as speed data and steering wheel data. Based on the changing parameters of the equipment on the vehicle, it further identifies the user's steering intention from the on-board operation, combines the user status information and the vehicle status information for mutual verification, and realizes multi-factor fusion intention recognition with higher recognition accuracy.

[0076] S2. Building a steering decision model based on a deep learning algorithm, inputting the user status information and the vehicle status information into the steering decision model to perform feature fusion processing and classification prediction to obtain the user's steering intention;

[0077] In this embodiment, the steering decision model includes an input layer, a convolution layer, a pooling layer, a fully connected layer and an activation function;

[0078] The input layer is used to normalize the user state information and the vehicle state information to obtain corresponding input features; the normalization process is used to map feature values ​​of different dimensions and ranges to the same scale interval, such as [0,1], so that subsequent algorithms can process more effectively.

[0079] For example, the dimension of the normalized feature sequence of the user state information and the vehicle state information is (T, F), where T is the time series length and F is the feature dimension.

[0080] The convolution layer is used to perform a convolution operation on the input features through a plurality of convolution kernels to realize feature extraction to obtain corresponding sight features, steering wheel features and vehicle speed features;

[0081] Among them, the convolution kernel size is (k, F), the step size is 1, and the padding method is the same, in order to keep the length of the output feature sequence the same as the input. The output feature dimension of the convolution layer is (T, C), where C is the number of convolution kernels, and k is the width of the convolution kernel on the time series, which determines the range covered by the convolution kernel on the time series.

[0082] The pooling layer is used to downsample the sight features, steering wheel features and vehicle speed features to reduce feature dimensions and output a corresponding feature sequence;

[0083] Among them, the pooling layer adopts maximum pooling, the pooling window size is 2, and the step size is 2. The output feature dimension of the pooling layer is (T / 2, C).

[0084] The fully connected layer has a built-in fusion algorithm for performing feature fusion processing and classification prediction according to the sight features, steering wheel features and vehicle speed features to obtain the user's steering intention.

[0085] That is, the fully connected layer is used to flatten the feature sequence output by the pooling layer into a one-dimensional vector, and perform feature fusion and classification through the fully connected layer. The output dimension of the fully connected layer is the number of turning intention categories, such as left turn, right turn, and straight driving.

[0086] The input data can be represented as a vector:

[0087] T=[Aleft,Aright,B,C]

[0088] Aleft and Aright represent the attention characteristics of the driver's sight on the left and right sides respectively, B represents the steering wheel angle characteristics, and C represents the vehicle speed characteristics.

[0089] Activation function: Add activation function after the convolutional layer and the fully connected layer to introduce nonlinear factors and enhance the expressiveness of the model.

[0090] This solution uses a deep learning algorithm to establish a steering decision model. Through feature extraction, feature fusion, classification prediction and other processes, it can adapt to different users, with strong pertinence and good adaptability. The deep learning algorithm can learn the user's preferences and behavior patterns, so as to more accurately predict the user's next operation, thereby ensuring the accuracy of the prediction of the user's steering intention and providing a good turn signal self-start service.

[0091] In this embodiment, the steering decision model is further trained, and the model training includes:

[0092] Collect driving data and annotate the data, including sight features, steering wheel features, vehicle speed features and corresponding steering intention labels;

[0093] The driving data is divided into a training set, a validation set, and a test set according to a preset ratio; the preset ratio is set according to actual needs, for example, 7:2:1.

[0094] Use the training set data to train the steering decision model, optimize the model parameters θ and minimize the loss function through the back propagation algorithm;

[0095] The trained steering decision model is evaluated using a test set, and the evaluation criteria of the model evaluation include one or more indicators of accuracy, recall rate, and F1 value to ensure that the model has good generalization ability.

[0096] This solution collects a large amount of labeled driving data for training. The availability of big data enables the model to learn the underlying patterns and reduce overfitting, thereby being able to handle more complex input data and task scenarios, achieving higher prediction accuracy and classification effects.

[0097] The real-time collected user status information and the vehicle status are input into the trained steering decision model, and the model outputs the corresponding steering intention prediction result:

[0098] X=f(T;θ)

[0099] Where X is the predicted steering intention, f is the steering decision model, T is the input data, and θ is the model parameter.

[0100] In this embodiment, Kalman filter, Bayesian network or other fusion algorithms can be used as required. The algorithm is optimized by a large amount of experimental data and the algorithm parameters are adjusted. These algorithms can dynamically adjust their weights in the final results according to the reliability and relevance of different data to improve the accuracy and reliability of turning intention recognition. If complex scene requirements are added, a classifier such as a decision tree or a neural network can also be trained using machine learning or deep learning algorithms to distinguish different turning intentions.

[0101] In this embodiment, after the user state information and the vehicle state information are input into the steering decision model, feature extraction is first performed, and the features obtained are as follows:

[0102] Sight features: Extract the time the sight stays in the rearview mirror or the side front direction, the speed and direction of sight movement, and record this information A(D SightLine ).

[0103] Steering wheel features: Extract the size and rate of change of the steering wheel angle and record this information B(D SteeringWheel ).

[0104] Vehicle speed characteristics: Extract the characteristics of vehicle speed change trend and record this information C(D VehicleSpeed ).

[0105] In this embodiment, when the fusion algorithm is a Bayesian network, feature fusion processing and classification prediction are performed according to the sight features, steering wheel features and vehicle speed features to obtain the user's turning intention, including:

[0106] b1. Classify scenes based on different driving scenarios to set corresponding steering intentions;

[0107] The driving scenarios include left lane change scenario, right lane change scenario, left U-turn scenario, right U-turn scenario and straight driving scenario. The left lane change scenario and left U-turn scenario correspond to "left" turning intention, the right lane change scenario and right U-turn scenario correspond to "right" turning intention, and the straight driving scenario corresponds to "non" turning intention.

[0108] b2. The sight line features, steering wheel features and vehicle speed features obtained according to the user state information and the vehicle state information are used as observation variables, and the prior probability corresponding to each turning intention is calculated in combination with the Bayesian theorem. The calculation formula of the prior probability is as follows:

[0109]

[0110] Among them, the sight feature A (such as sight angle), steering wheel feature B (such as steering wheel angle) and vehicle speed feature C are observed variables, and X is the turning intention. This formula means calculating the probability of turning intention X given the observed values ​​A, B and C. P(X) is the prior probability of turning intention X, which can be obtained by counting the frequency of different turning intentions (such as left turn, right turn, and straight driving) in a large amount of driving data. For example, if left turn occurs n1 times in N driving records, then

[0111] Among them, the Bayesian network is a probabilistic graphical model used to represent the dependency relationship between variables.

[0112] P(A,B,C|X) is the joint conditional probability of sight feature A, steering wheel feature B and vehicle speed feature C given the steering intention X. This needs to be learned from a large amount of driving data. Assume that the sight feature A can be discretized into [a1,a2,…,am], the steering wheel feature B can be discretized into [b1,b2,…,bn], the vehicle speed feature C can be discretized into [c1,c2,…,co], and the steering intention X can be discretized into [x1,x2,…,xp]. Then P(A=ai,B=bj,C=ck|X=xl) represents the probability that the sight angle is ai, the steering wheel angle is bj and the vehicle speed is ck when the steering intention is xl.

[0113] P(A,B,C) is the joint probability of sight angle A, steering wheel angle B and vehicle speed C, which can be expressed by the total probability formula:

[0114] P(A,B,C)=∑ x∈x P(A,B,C|x)P(x)

[0115] b3. Compare the prior probabilities of all the turning intentions, take the turning intention with the highest probability value as the classification prediction result and output it.

[0116] This solution has built-in fusion algorithms, such as Bayesian networks and Kalman filters, in the steering decision model based on deep learning algorithms. When extracting line of sight features, steering wheel features, and vehicle speed features, it considers the probabilistic relationship between various factors based on the reliability and relevance of different data, and dynamically adjusts their weights in the final result. Then, the probabilistic model is used to characterize and predict different user steering intentions, thereby providing more reliable prediction results to improve the accuracy and reliability of steering intention recognition.

[0117] S3, confirming whether to execute the turn signal self-start control according to the user's turning intention, including:

[0118] The user's turning intention and the corresponding prior probability are obtained from the turning decision model, and it is determined whether the prior probability is greater than a preset threshold. If so, it is confirmed that the turn signal self-start control is executed, and then the corresponding control instruction is generated based on the turning intention to drive the corresponding left / right turn signal to start.

[0119] Specifically, a probability threshold δ is set as a preset threshold. When the probability of a certain turning intention predicted by the turning decision model is greater than δ, it is determined that the driver has the turning intention. If the driving behavior and environmental conditions of left and right turns are similar, the same characteristic value threshold δ can be used, and δ=δleft=δright can be set; if the experimental data show that there are significant differences in the driving behavior of left and right turns, different thresholds δleft and δright can be set.

[0120] Then, the turn signal on condition:

[0121] Turn on the left turn signal, if Xleft>δleft

[0122] Turn on the right turn signal, if Xright>δright

[0123] Do not turn on the turn signal, if X<δ

[0124] Among them, Xleft and Xright are the left-turn and right-turn intention probabilities predicted by the model, respectively.

[0125] When it is determined that the conditions for turning on the turn signal are met, this intention is sent as an in-vehicle signal to the vehicle control module responsible for the turn signal; after receiving the signal, the turn signal control module determines that the current turn signal is not turned on, and automatically turns on the turn signal on the corresponding side (that is, the side that the driver's line of sight is focused on and the steering wheel has the intention to turn).

[0126] Based on the actual application environment, this solution sets a probability threshold as a preset threshold. By comparing whether the prior probability of the output turning intention is greater than the preset threshold, it is determined whether to execute the turn signal self-start control, thereby reducing the probability of false triggering of the turn signal and improving driving safety.

[0127] The present invention also provides a vehicle turn signal self-starting system, comprising a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a vehicle turn signal self-starting method as described above is implemented.

[0128] In this embodiment, a data acquisition module connected to the processor is also included, and the data acquisition module includes an image acquisition module, a vehicle speed acquisition module and a turning angle acquisition unit;

[0129] The image acquisition module is used to collect user status information in real time;

[0130] A vehicle speed collection unit, used for collecting vehicle speed data;

[0131] The turning angle acquisition unit is used to collect the steering wheel data of the vehicle.

[0132] This solution is based on the vehicle's own driver monitoring system and body domain control system, and defines an image acquisition module, a vehicle speed acquisition module, and a turning angle acquisition unit for data acquisition. It does not require the installation of additional equipment and can be adapted to different types of vehicles, with good applicability.

[0133] Specifically, the self-starting working principle of the turn signal lamp of the present invention is as follows:

[0134] The driver monitoring system (DMS) has a line of sight interaction function. During driving, it can identify the driver's line of sight in real time. When it is detected that the driver's line of sight is focused on the corresponding left or right rearview mirror, and the left front or right front area, this may indicate that the driver intends to turn to that side or change lanes. At the same time, the present invention further monitors other rotation information of the vehicle as an auxiliary. If it is detected that the steering wheel has the intention to turn to the corresponding side during the period when the line of sight is focused on the corresponding side (such as by monitoring the steering angle of the steering wheel, etc.) and combined with the current vehicle speed change trend, the turn signal on the corresponding side will be automatically turned on, thereby realizing the automatic turn signal function. The present invention does not require the driver to manually operate the turn signal, which improves the degree of automation of the turn signal operation and helps to reduce traffic safety problems caused by improper turn signal operation.

[0135] The beneficial effects of the present invention are as follows:

[0136] 1. Improve the automation of turn signal operation

[0137] The present invention can automatically turn on the turn signal according to the driver's line of sight through the coordinated work of the driver monitoring system (DMS) and the rotation monitoring (vehicle speed data and steering wheel data). This greatly reduces the number of manual operations by the driver and improves the convenience and automation level of the operation.

[0138] 2. Enhance traffic safety

[0139] By reducing the manual operation of the turn signal, the driver can avoid the situation where the turn signal is not turned on due to forgetfulness or untimely operation. When the vehicle turns or changes lanes, correctly turning on the turn signal is crucial to convey the driving intention to other road users. This technology can effectively reduce traffic accidents caused by improper operation of the turn signal and ensure road traffic safety.

[0140] 3. Improve the comfort of driving experience

[0141] The driver does not need to be distracted to manually operate the turn signal, especially in complex road conditions that require frequent turns or lane changes. He can focus more on driving operations, such as controlling vehicle speed and maintaining distance between vehicles. By realizing automated turn signal operation, the driver's operating burden is reduced and the driving experience is more comfortable.

[0142] 4. Improve the accuracy of turning intention recognition

[0143] A variety of data fusion algorithms are used, such as Kalman filter, Bayesian network and other fusion algorithms, which can dynamically adjust the weights according to the reliability and relevance of different data. In complex scenarios, machine learning or deep learning algorithms can also be used to train classifiers (such as support vector machines, decision trees or neural networks) to distinguish different turning intentions. This effectively improves the recognition accuracy of the driver's turning intention, reduces the possibility of misjudgment, and further improves the reliability of the automatic turn signal function.

[0144] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for self-starting a vehicle turn signal lamp, characterized in that: include: Obtain user status information in real time based on the driver monitoring system; Collect vehicle status information of the current vehicle in real time; Building a steering decision model based on a deep learning algorithm, inputting the user state information and the vehicle state information into the steering decision model to perform feature fusion processing and classification prediction to obtain the user's steering intention; Whether to execute the turn signal self-start control is determined according to the user's turning intention.

2. A vehicle turn signal self-starting method as claimed in claim 1, characterized in that: Real-time acquisition of user status information based on the driver monitoring system, including: Divide the space around the vehicle into regions and obtain target regions corresponding to different steering intentions; Based on the driver monitoring system, the user's head image is collected in real time, image recognition is performed to realize line of sight tracking, and area matching is performed with the target area to determine the user's line of sight information; The user's line of sight information includes one or more of a line of sight angle, a line of sight dwelling time, a line of sight moving speed, and a line of sight moving direction.

3. A vehicle turn signal self-starting method as claimed in claim 2, characterized in that: It also includes: using infrared sensors to perform pupil positioning and corneal reflection point detection to assist image recognition in achieving line of sight tracking; and / or, obtaining the user's head posture and using a deep learning algorithm to estimate the three-axis posture of the user's head, and then dynamically compensating the line of sight tracking achieved by image recognition in combination with the three-axis posture.

4. A vehicle turn signal self-starting method as claimed in claim 1, characterized in that: The vehicle status information of the current vehicle is collected in real time, specifically: the vehicle body domain control system is used to continuously monitor the vehicle and obtain the corresponding vehicle body information, and the vehicle speed data and steering wheel data are obtained from the vehicle body information as the vehicle status information of the current vehicle.

5. A vehicle turn signal self-starting method as claimed in claim 1, characterized in that: The steering decision model includes an input layer, a convolution layer, a pooling layer, a fully connected layer and an activation function; The input layer is used to normalize the user state information and the vehicle state information to obtain corresponding input features; The convolution layer is used to perform a convolution operation on the input features through a plurality of convolution kernels to realize feature extraction to obtain corresponding sight features, steering wheel features and vehicle speed features; The pooling layer is used to downsample the sight features, steering wheel features and vehicle speed features to reduce feature dimensions and output a corresponding feature sequence; The fully connected layer has a built-in fusion algorithm for performing feature fusion processing and classification prediction according to the sight features, steering wheel features and vehicle speed features to obtain the user's steering intention.

6. A vehicle turn signal self-starting method as claimed in claim 5, characterized in that: When the fusion algorithm is a Bayesian network, feature fusion processing and classification prediction are performed according to the sight features, steering wheel features and vehicle speed features to obtain the user's steering intention, including: Classify scenes based on different driving scenarios to set corresponding steering intentions; The sight line features, steering wheel features and vehicle speed features obtained according to the user state information and the vehicle state information are used as observation variables, and the prior probability corresponding to each turning intention is calculated in combination with the Bayesian theorem; The prior probabilities of all the turning intentions are compared, and the turning intention with the highest probability value is taken as the classification prediction result and output.

7. A vehicle turn signal self-starting method as claimed in claim 6, characterized in that: Confirming whether to execute the turn signal self-start control according to the user's turning intention includes: The user's turning intention and the corresponding prior probability are obtained from the turning decision model, and it is determined whether the prior probability is greater than a preset threshold. If so, it is confirmed that the turn signal self-start control is executed, and then the corresponding control instruction is generated based on the turning intention to drive the corresponding left / right turn signal to start.

8. A vehicle turn signal self-starting method as claimed in claim 6, characterized in that: The invention also includes performing model training on the steering decision model, wherein the model training includes: Collect driving data and annotate the data, including sight features, steering wheel features, vehicle speed features and corresponding steering intention labels; Divide the driving data into training set, validation set and test set according to the preset ratio; Use the training set data to train the steering decision model, optimize the model parameters θ and minimize the loss function through the back propagation algorithm; The trained steering decision model is evaluated using a test set, wherein the evaluation criteria of the model evaluation include one or more indicators of accuracy, recall, and F1 value.

9. A vehicle turn signal self-starting system, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, a vehicle turn signal self-starting method as described in any one of claims 1 to 8 is implemented.

10. A vehicle turn signal self-starting system as claimed in claim 9, characterized in that: It also includes a data acquisition module connected to the processor, the data acquisition module includes an image acquisition module, a vehicle speed acquisition module and a turning angle acquisition unit; The image acquisition module is used to collect user status information in real time; A vehicle speed collection unit, used for collecting vehicle speed data; The turning angle acquisition unit is used to collect the steering wheel data of the vehicle.

Images