Steering wheel hand-leaving detection method and system, storage medium and intelligent automobile

Abstract

The invention relates to the technical field of vehicle detection, in particular to a steering wheel hand-leaving detection method and system, a storage medium and an intelligent vehicle. Comprising the following steps: acquiring reference capacitance data when a steering wheel is not touched and touch capacitance data when the steering wheel is touched, and acquiring difference data; cleaning the obtained difference data and adding classification labels; training the difference data with the classification labels by using a neural network to obtain a detection model, and deploying the trained detection model into a controller; collecting current reference capacitance data and touch capacitance data to obtain current difference value data, inputting the current difference value data into the detection model to obtain a touch gesture classification result, and judging whether the steering wheel leaves the hand or not according to the touch gesture classification result; according to the scheme, prediction is achieved through a large amount of historical data, a capacitance threshold value does not need to be set manually, errors caused by manual operation are reduced, the accuracy of steering wheel hand leaving detection is improved, and therefore the safety of vehicle driving is improved.

Description

Technical Field

[0001] The present invention relates to the field of vehicle detection technology, and in particular to a method, system, storage medium and intelligent vehicle for detecting hands-off steering wheel. Background Art

[0002] With the accelerated advancement of urbanization and the continuous growth of car ownership, the application of autonomous driving technology is becoming more and more extensive; however, with the development of autonomous driving technology, corresponding safety issues are gradually becoming prominent. In order to ensure driving safety, regulations usually require that even in autonomous driving mode, the driver cannot completely lose control of the vehicle. Therefore, steering wheel hands-off detection becomes particularly important.

[0003] Currently, most of the steering wheel hand-off detection on the market uses a capacitance threshold to determine whether the steering wheel is touched. When the hand touches the capacitive sensor on the steering wheel, the capacitive sensor senses the increase or decrease in capacitance and compares it with the average capacitance when not touched, thereby determining whether there is a hand touching it.

[0004] There are some potential defects in the method of judging by threshold value, which may easily lead to false triggering of different people. The finger capacitance of people of different height and weight is different, and the touch threshold is also very different. If the threshold is set according to a large person, it will cause a small person to find it difficult to reach the set threshold even if he touches it. However, if the threshold is set according to a small person, it will cause a large person to be judged as touching the steering wheel before he touches it. In addition, changes in the environment may cause the capacitance to rise or fall when touching. This situation will not only affect the user experience, but may also bring unnecessary safety hazards.

[0005] Therefore, it is crucial for those skilled in the art to design a more accurate steering wheel hands-off detection method, system, storage medium and smart car. Summary of the invention

[0006] The technical problem to be solved by the embodiments of the present invention is to provide a more accurate steering wheel hands-off detection method, system, storage medium and smart car to solve the problem in the prior art that it is easy to cause false triggering for different people.

[0007] The present invention discloses a method for detecting a hands-off state of a steering wheel, which comprises the following steps:

[0008] Using a capacitive sensor to collect reference capacitance data when the steering wheel is not touched and touch capacitance data generated in different touch modes, and subtracting the touch capacitance data from the reference capacitance data to obtain difference data;

[0009] Clean the acquired difference data and add classification labels to the cleaned difference data;

[0010] A neural network is used to train the difference data with classification labels to obtain a detection model, and the trained detection model is deployed to the controller;

[0011] The current reference capacitance data and touch capacitance data are collected to obtain the current difference data, the current difference data is input into the detection model to obtain the touch gesture classification result, and the steering wheel is judged whether the hands are off the steering wheel according to the touch gesture classification result.

[0012] Optionally, the “collecting reference capacitance data when the steering wheel is not touched and touch capacitance data generated in different touch modes by using a capacitance sensor” comprises the following steps:

[0013] Capacitive sensors are provided in multiple areas of the steering wheel;

[0014] The average capacitance values ​​of the corresponding areas when they are not touched are collected by the capacitance sensors as reference capacitance values;

[0015] The capacitive sensors are used to collect the touch capacitance values ​​generated by different people touching the corresponding areas in different gestures and directions.

[0016] The reference capacitance value and the touch capacitance value are converted into reference capacitance data and touch capacitance data respectively according to a preset conversion formula.

[0017] Optionally, the mode of the capacitive sensor is a mutual capacitance mode, and the conversion formula of the capacitive sensor in a single CDAC mode is:

[0018]

[0019] The conversion formula of the capacitive sensor in dual CDAC mode is:

[0020]

[0021] in,

[0022] Maxcount = N sub *TxCLK Div

[0023] Rawdata is the touch data, Maxcount is the maximum touch capacitance data collected by the capacitance sensor; N sub is the sub-conversion number of the capacitive sensor, TxCLK Div is the Tx clock division frequency of the capacitive sensor, CP is the parasitic capacitance value of the capacitive sensor, CF is the finger capacitance value collected by the capacitive sensor, Cref is the reference capacitance value of the capacitive sensor, and CompClkDiv is the compensation CDAD division frequency of the capacitive sensor.

[0024] Optionally, the step of “cleaning the acquired difference data and adding classification labels to the cleaned difference data” includes the following steps:

[0025] Invalid data and abnormal data in the difference data are filtered out by a digital filter to form a data set;

[0026] The filtered difference data is classified according to the preset classification rules and corresponding classification labels are added to form a label set.

[0027] Optionally, the “using a neural network to train the difference data with classification labels to obtain a detection model, and deploying the trained detection model to the controller” includes the following steps:

[0028] Select part of the data from all the difference data as the training set, and another part of the data as the test set;

[0029] Import the training set and test set in Python, and load the difference data and corresponding label set in the training set;

[0030] The detection model is trained for multiple rounds using the difference data in the training set and the corresponding label set to generate a trained detection model, which is then deployed to the controller.

[0031] Optionally, the detection model is a 5-layer fully connected neural network, which includes an input layer, 3 hidden layers and an output layer, the input layer includes 3 nodes, each hidden layer includes 16 nodes, and the output layer includes 5 nodes, wherein the hidden layer is activated by a Relu activation function, and the output layer is activated by a Softmax activation function.

[0032] Optionally, the Relu activation function is:

[0033] Relu(x)=max(0,x)

[0034] The Softmax activation function is:

[0035]

[0036] The loss function of the fully connected neural network is:

[0037]

[0038] In order to solve the problems existing in the prior art, the present invention further provides a steering wheel hands-off detection system, which is used to implement the steering wheel hands-off detection method as described in any one of the above items, and the solution thereof comprises:

[0039] A data acquisition module, used for acquiring reference capacitance data and touch capacitance data;

[0040] The data processing and labeling module is used to calculate the difference data and classify and label the difference data; the model training and deployment module is used to train the detection model using the difference data with classification labels and deploy the trained detection model to the controller;

[0041] The detection module is used to collect current touch capacitance data, input the current touch capacitance data into the detection model to obtain the touch gesture classification result, and judge whether the steering wheel is released according to the touch gesture classification result.

[0042] In order to solve the problems existing in the prior art, the present invention also provides a computer-readable storage medium, characterized in that a computer program is stored, and when the computer program is executed by a processor, the processor executes the steps of the method as described in any one of claims 1 to 7.

[0043] In order to solve the problems existing in the prior art, the present invention also provides a smart car, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the steering wheel hands-off detection method as described in any of the above items.

[0044] Compared with the prior art, the beneficial effect of the steering wheel hands-off detection method provided by the embodiment of the present invention is that: by collecting a large amount of touch capacitance data of different people at different angles and in different touch methods, and using a neural network to train the large amount of touch capacitance data to obtain a detection model, and judging whether the current steering wheel is off the hand through the detection model and the current touch capacitance data, this scheme uses a large amount of historical data to achieve prediction, does not require manual setting of capacitance thresholds, reduces errors caused by manual operation, reduces the risk of misjudgment, and improves the accuracy of steering wheel hands-off detection, thereby improving the safety of vehicle driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The scheme of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments, in which:

[0046] Figure 1 The process of the steering wheel hands-off detection method provided by the embodiment of the present invention is as follows: Figure 1 ;

[0047] Figure 2 The process of the steering wheel hands-off detection method provided by the embodiment of the present invention is as follows: Figure 2 ;

[0048] Figure 3 is a system block diagram of a steering wheel hands-off detection system provided by an embodiment of the present invention;

[0049] Figure 4 It is a structural block diagram of a smart car provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0050] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. Now, in conjunction with the accompanying drawings, the preferred embodiments of the present invention are described in detail.

[0051] like Figure 1 and Figure 2 As shown, the present invention provides a specific embodiment of a method for detecting hands-off steering wheel.

[0052] A steering wheel hands-off detection method, reference Figure 1 , the steering wheel hands-off detection method comprises the following steps:

[0053] S1, using a capacitance sensor to collect reference capacitance data when the steering wheel is not touched and touch capacitance data generated under different touch modes, and subtracting the touch capacitance data from the reference capacitance data to obtain difference data;

[0054] S2. Clean the acquired difference data and add classification labels to the cleaned difference data;

[0055] S3, using a neural network to train the difference data with classification labels to obtain a detection model, and deploying the trained detection model to the controller;

[0056] S4. Collect current reference capacitance data and touch capacitance data to obtain current difference data, input the current difference data into a detection model to obtain a touch gesture classification result, and determine whether the steering wheel is released according to the touch gesture classification result.

[0057] Specifically, the steering wheel can be divided into three area structures, namely, a left partition, a right partition and a back partition. Pads with capacitive sensors are provided on the left partition, the right partition and the back partition of the steering wheel. When the steering wheel is not touched, the capacitive sensors of different partitions collect the capacitance data of the corresponding partitions as the reference capacitance data of the corresponding partitions. Multiple testers touch the steering wheel with different gestures and in different directions, and the capacitance data generated after touching in different partitions are collected by the capacitive sensors of different partitions as touch capacitance data.

[0058] Among them, since people's height, weight, strength and finger characteristics are different, it is necessary to collect data from multiple testers with different body characteristics, touch the steering wheel with different gestures and in different directions, and collect all the touch capacitance data generated, so that the training results can cover a wider range and cover different user groups.

[0059] Furthermore, since a person can cause the capacitive sensor to generate an increment in finger capacitance when touching the steering wheel, the incremental change in the finger capacitance can be used to determine whether the person has touched the steering wheel, and which specific area has been touched; by subtracting the touch capacitance data from the reference capacitance data, the specific incremental value of the finger capacitance can be obtained as difference data for subsequent training.

[0060] Furthermore, the obtained difference data is cleaned and the noise within a specific frequency range can be filtered out by a digital filter (such as a low-pass filter, a band-pass filter, etc.) to remove invalid data and abnormal data to form a data set for subsequent training.

[0061] Furthermore, the cleaned difference data is classified according to preset classification rules and corresponding classification labels are added to form a label set for subsequent training; the classification labels are used to represent touch gestures as a basis for judging the touch gesture classification results; the classification rules can be "0" as no touch annotation, "1" as left hand touch annotation, "2" as right hand touch annotation, "3" as both hands touch annotation, and "4" as steering wheel back touch annotation; the label set can provide the model with accurate touch gesture classification judgment basis to facilitate the model to learn the recognition of different touch states.

[0062] Furthermore, a part of all the difference data with classification labels is extracted as a training set for training the model, and another part is used as a test set for evaluating the accuracy of the model; wherein, most of the difference data can be used as a training set, and the remaining small part can be used as a test set; the model is trained by the difference data with classification labels in the training set to obtain a detection model,

[0063] In order to process data more quickly and conveniently, the difference data in the training set and the test set can be normalized before training. The normalization process generally includes minimum-maximum scaling, Z-score standardization, normal distribution normalization, etc.; according to the data distribution and model requirements in this embodiment, the minimum-maximum scaling method is selected for normalization.

[0064] Minimum-maximum normalization is to scale the data to a fixed range through linear transformation, usually [0,1] range or [-1,1] range, calculate the minimum min and maximum max of each data, and apply the minimum-maximum scaling formula to each value x of each data. Specifically, the normalization can be obtained according to the following formula, and the difference data is implied to the range of [0,1]:

[0065]

[0066] Among them, X i For each data in the data set; X min is the minimum data in the data set; X max The maximum value in the data set.

[0067] Through minimum-maximum normalization data processing, not only can the original distribution of the data be retained and the data only be scaled to a fixed range, but the impact of different dimensions between different data can also be eliminated, making the data more comparable; when training a neural network model, normalizing the data can speed up the convergence of the model to reduce training time and resource consumption, help the model better learn the relationship between the distribution and features of the data, and improve the performance and generalization ability of the model.

[0068] To use a neural network to train normalized data, you can first set the training rounds and data batches, then import the training set and test set into the neural network model built in Python, and load the difference data and corresponding label set in the training set for training; a training round is the number of times the entire training set is iteratively trained on the neural network. In each training round, the neural network receives the entire data set and updates the weights through forward propagation, loss calculation, back propagation and other steps. The appropriate number of training rounds can be determined based on the training effect of the observed model; in each training round, the data set is divided into multiple batches for training, and each data batch contains a certain amount of sample data. The neural network processes each data batch one by one, calculates the corresponding loss and updates the weights.

[0069] The detection model is trained using a dataset with classification labels to minimize the prediction error through repeated iterations; finally, the trained detection model is converted into a format suitable for running on a microcontroller through compression, quantization, or conversion to a format suitable for the target hardware, and the converted detection model is deployed to the target microcontroller.

[0070] Furthermore, the touch capacitance data in the three partitions of the steering wheel are detected in real time, and the detected touch capacitance data of the three partitions are input into the trained detection model to predict and determine whether it is a touch, and the largest value in the prediction result is selected as the basis for judgment to obtain whether it is a touch and the specific touch gesture classification result.

[0071] In one embodiment, reference Figure 2 “Using a capacitive sensor to collect reference capacitance data when the steering wheel is not touched and touch capacitance data generated under different touch modes” specifically includes the following steps:

[0072] S11, using the capacitance sensor to collect average capacitance values ​​of the corresponding areas when they are not touched, as reference capacitance values;

[0073] S12, collecting, by means of a capacitive sensor, touch capacitance values ​​generated by different people touching in different gestures and directions in multiple groups of corresponding areas;

[0074] S13. Convert the reference capacitance value and the touch capacitance value into reference capacitance data and touch capacitance data respectively according to a preset conversion formula.

[0075] Specifically, the capacitive sensor is in mutual capacitance mode. When the capacitive sensor is in single CDAC mode (only the reference capacitance DAC is present in the capacitive sensor, and the compensation capacitance DAC is in the off state), the reference capacitance data and the touch capacitance data are calculated according to the following conversion formula:

[0076]

[0077] Among them, Rawdata is the touch data, N sub is the sub-conversion number of the capacitive sensor, TxCLK Div is the Tx clock division frequency of the capacitive sensor, CP is the parasitic capacitance value of the capacitive sensor, and CF is the finger capacitance value collected by the capacitive sensor.

[0078] When the capacitance sensor is in dual CDAC mode (both the reference capacitance DAC and the compensation capacitance DAC are enabled in the capacitance sensor), the reference capacitance data and the touch capacitance data are calculated according to the following conversion formula:

[0079]

[0080] in,

[0081] Maxcount = N sub *TxCLK Div ;

[0082] Rawdata is the touch data, Maxcount is the maximum touch capacitance data collected by the capacitance sensor; N sub is the sub-conversion number of the capacitive sensor, TxCLK Div is the Tx clock division frequency of the capacitive sensor, CP is the parasitic capacitance value of the capacitive sensor, CF is the finger capacitance value collected by the capacitive sensor, Ccomp is the compensation capacitance value of the capacitive sensor, CompClkDiv is the compensation CDAD division frequency of the capacitive sensor, and Cref is the reference capacitance value of the capacitive sensor.

[0083] In CDAC, the number of sub-conversions N subRefers to the number of sub-conversion steps required to convert the input capacitance value into a digital signal within a complete conversion cycle. In CDAC, a complete conversion cycle is usually divided into multiple sub-conversion steps, each of which performs certain processing and conversion on the input capacitance to obtain a digital output result.

[0084] The parasitic capacitance CP is an unintentional capacitance existing in the circuit of the capacitive sensor due to structure, wiring or other factors.

[0085] Finger capacitance CF is when a finger approaches the surface of the capacitive sensor. The capacitance around the finger will affect the capacitance value in the capacitive sensor circuit, thereby generating a detectable signal. The finger capacitance CF can be measured and obtained by the capacitive sensors in different partitions of the steering wheel. It can be any of the touch capacitance data, current capacitance data, and reference capacitance data.

[0086] The compensation capacitor Ccomp is used to correct the error or deviation that may exist in the measurement process of the capacitive sensor. The compensation capacitor is usually used to adjust the sensitivity, stability and accuracy of the capacitive sensor to improve the performance of the sensor.

[0087] The role of the compensation divider frequency CompClkDiv here is to divide the compensation capacitor Ccomp to adjust the influence of the compensation capacitor, thereby affecting the calculation. Through the frequency division operation of the compensation divider, the error introduced by the compensation capacitor Ccomp can be corrected to ensure accurate data in the dual CDAC mode.

[0088] The reference capacitance value Cref is used as a reference value to help measure the change of the target capacitance. Specifically, it provides a known fixed value for calibrating and comparing the change of the target capacitance. By comparing with the reference capacitance Cref, the change of the target capacitance can be accurately measured, thereby realizing the detection and recognition of the external environment or touch input.

[0089] Tx clock division frequency TxCLK Div The clock frequency used to control the sensing capacitance measurement affects the accuracy and speed of the measurement. It is usually used to divide the system clock signal to generate the sensing clock signal for measurement. By adjusting the division ratio of the sensing clock divider, the time interval of the sensing capacitance measurement can be controlled, thereby affecting the accuracy and sensitivity of the measurement.

[0090] In one embodiment, the detection model is a 5-layer fully connected neural network, which includes an input layer, 3 hidden layers and an output layer, the input layer includes 3 nodes, each hidden layer includes 16 nodes, and the output layer includes 5 nodes, wherein the hidden layer is activated by a Relu activation function, and the output layer is activated by a Softmax activation function.

[0091] Specifically, the detection model is a 5-layer fully connected neural network. The input layer of the detection model includes 3 nodes, which serve as the input of the fully connected neural network and are used to represent the capacitance signal of the object. The detection model includes three hidden layers, and each hidden layer has 16 nodes, and each hidden layer is activated by a Relu activation function. The Relu activation function is a nonlinear activation function that can help the neural network learn nonlinear relationships to improve the expressive power of the network. The Relu activation function is simple and efficient to calculate, and there is no problem of gradient disappearance. When the input is a negative number, the output of the Relu activation function is 0, which also brings a certain sparsity to the model, which helps the generalization ability of the model.

[0092] The Relu activation function is:

[0093] Relu(x)=max(0,x)

[0094] Among them, if the input x is greater than or equal to 0, the output of the Relu activation function is x, if the output x is less than 0, the output of the Relu activation function is 0.

[0095] Furthermore, the output layer of the detection model includes 5 nodes. These 5 node elements are used to represent the recognized steering wheel touch gesture. The output layer is activated using the Softmax activation function. The Softmax activation function can convert the original output of the neural network into a form representing a probability distribution, so that the output value can be interpreted as the probability of each classification category, so as to facilitate the prediction of multi-classification tasks. During the training process, the output of the Softmax activation function can be used to calculate the loss with the actual classification label to measure the difference between the predicted probability distribution and the actual distribution.

[0096] The Softmax activation function is:

[0097]

[0098] In it, given a vector with multiple categories, the Softmax activation function converts each element x into a value between 0 and 1 to represent the probability of the category.

[0099] The fully connected neural network uses a binary cross entropy loss function for loss calculation. Assuming that for each sample, the true label is y (0 or 1), the output of the model is x (ranging from 0 to 1), the loss function of the fully connected neural network is:

[0100]

[0101] Among them, n is the number of samples, y is the true label of the sample (0 or 1), and x is the model's predicted value for the sample. When the true label is 1, the loss function will punish the model output of 0, so that the model learns in the right direction. When the true label is 0, the loss function will punish the model output of 1. By minimizing the binary cross entropy loss function, the model can better fit the training data and improve the classification accuracy.

[0102] like Figure 3 As shown, the present invention provides a specific embodiment of a steering wheel hands-off detection system.

[0103] A steering wheel hands-off detection system is used to implement any of the above steering wheel hands-off detection methods, referring to Figure 3 The steering wheel hands-off detection system includes: a data acquisition module 100, a data processing and annotation module 200, a model training and deployment module 300 and a detection module 400.

[0104] Specifically, the data acquisition module 100 is used to collect a reference capacitance value when the steering wheel is not touched and a touch capacitance value when the steering wheel is touched, and convert the reference capacitance value and the touch capacitance value into corresponding reference capacitance data and touch capacitance data according to a preset conversion formula.

[0105] Furthermore, the data processing and annotation module 200 is used to calculate difference data based on the reference capacitance data and the touch capacitance data, and to perform data cleaning on the obtained difference data. Specifically, a digital filter can be configured in the data processing and annotation module 200, and a digital filter (such as a low-pass filter, a band-pass filter, etc.) can be used to filter out noise within a specific frequency range to remove invalid data and abnormal data, so as to form a data set for subsequent training.

[0106] The data processing and labeling module 200 is also used to classify the cleaned difference data according to preset classification rules and add corresponding classification labels to form a label set for subsequent training; the classification label is used to represent the touch gesture as a basis for judging the touch gesture classification result; the classification rule can be "0" as the untouched label, "1" as the left hand touch label, "2" as the right hand touch label, "3" as the two-hand touch label, and "4" as the steering wheel back touch label; the label set can provide the model with accurate touch gesture classification judgment basis to facilitate the model to learn the recognition of different touch states.

[0107] Furthermore, the model training and deployment module 300 is used to train the detection model using difference data with classification labels, and deploy the trained detection model to the controller.

[0108] Among them, the detection model is a 5-layer fully connected neural network. The input layer of the detection model includes 3 nodes, which serve as the input of the fully connected neural network and are used to represent the capacitance signal of the object. The detection model includes three hidden layers, and each hidden layer has 16 nodes, and each hidden layer is activated by the Relu activation function. The output layer includes 5 nodes, and the output layer is activated by the Softmax activation function.

[0109] Among them, the Relu activation function is:

[0110] Relu(x)=max(0,x)

[0111] Among them, if the input x is greater than or equal to 0, the output of the Relu activation function is x, if the output x is less than 0, the output of the Relu activation function is 0.

[0112] The Softmax activation function is:

[0113]

[0114] In it, given a vector with multiple categories, the Softmax activation function converts each element x into a value between 0 and 1 to represent the probability of the category.

[0115] Assume that for each sample, the true label is y (0 or 1), the output of the model is x (ranging from 0 to 1), and the loss function of the fully connected neural network is:

[0116]

[0117] Where n is the number of samples, y is the true label of the sample (0 or 1), and x is the model's predicted value for the sample. When the true label is 1, the loss function will punish the model output as 0, so that the model learns in the right direction. When the true label is 0, the loss function will punish the model output as 1.

[0118] The model training and deployment module 300 is also used to train the detection model using a data set with classification labels to minimize the prediction error through repeated iterations; finally, the trained detection model is converted into a format suitable for running on a microcontroller through operations such as compression, quantization or conversion into a format suitable for the target hardware, and the converted detection model is deployed to the target microcontroller.

[0119] Furthermore, the detection module 400 is used to collect current reference capacitance data and current touch capacitance data, and input the difference data into the detection model to obtain a touch gesture classification result, and determine whether the steering wheel is released according to the touch gesture classification result.

[0120] The collection and calculation method of the current reference capacitance data and the current touch capacitance data is as follows:

[0121] When the capacitance sensor is in single CDAC mode (there is only a reference capacitance DAC in the capacitance sensor and the compensation capacitance DAC is turned off), the reference capacitance data and touch capacitance data are calculated according to the following conversion formula:

[0122]

[0123] Among them, Rawdata is the touch data, N sub is the sub-conversion number of the capacitive sensor, TxCLK Div is the Tx clock division frequency of the capacitive sensor, CP is the parasitic capacitance value of the capacitive sensor, and CF is the finger capacitance value collected by the capacitive sensor.

[0124] When the capacitance sensor is in dual CDAC mode (both the reference capacitance DAC and the compensation capacitance DAC are enabled in the capacitance sensor), the reference capacitance data and the touch capacitance data are calculated according to the following conversion formula:

[0125] Among them, Rawdata is the touch data, N sub is the sub-conversion number of the capacitive sensor, TxCLK Div is the Tx clock division frequency of the capacitive sensor, CP is the parasitic capacitance value of the capacitive sensor, CF is the finger capacitance value collected by the capacitive sensor, Ccomp is the compensation capacitance value of the capacitive sensor, CompClkDiv is the compensation CDAD division frequency of the capacitive sensor, and Cref is the reference capacitance value of the capacitive sensor.

[0126] like Figure 4 As shown, the present invention also provides a specific embodiment of a smart car.

[0127] A smart car, reference Figure 4 The smart car includes a memory 10 and a processor 20. The memory 10 stores a computer program. When the computer program is executed by the processor 20, the processor 20 executes the steps of any of the above steering wheel hand-off detection methods. The detailed steps can be found above.

[0128] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0129] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0130] It should be understood that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. For those skilled in the art, the technical solutions described in the above embodiments can be modified, or some of the technical features therein can be replaced by equivalents; and all these modifications and replacements should fall within the protection scope of the claims attached to the present invention.

Claims

1. A method for detecting hands-off steering wheel, characterized in that: The following steps are involved: Using a capacitive sensor to collect reference capacitance data when the steering wheel is not touched and touch capacitance data generated in different touch modes, and subtracting the touch capacitance data from the reference capacitance data to obtain difference data; Clean the acquired difference data and add classification labels to the cleaned difference data; A neural network is used to train the difference data with classification labels to obtain a detection model, and the trained detection model is deployed to the controller; The current reference capacitance data and touch capacitance data are collected to obtain the current difference data, the current difference data is input into the detection model to obtain the touch gesture classification result, and the steering wheel is judged whether the hands are off the steering wheel according to the touch gesture classification result.

2. The method for detecting hands-off steering wheel according to claim 1, characterized in that: The "collecting reference capacitance data when the steering wheel is not touched and touch capacitance data generated under different touch modes by using a capacitance sensor" includes the following steps: Capacitive sensors are provided in multiple areas of the steering wheel; The average capacitance values ​​of the corresponding areas when they are not touched are collected by the capacitance sensors as reference capacitance values; The capacitive sensors are used to collect the touch capacitance values ​​generated by different people touching the corresponding areas in different gestures and directions. The reference capacitance value and the touch capacitance value are converted into reference capacitance data and touch capacitance data respectively according to a preset conversion formula.

3. The method for detecting hands-off steering wheel according to claim 2, characterized in that: The mode of the capacitive sensor is the mutual capacitance mode, and the conversion formula of the capacitive sensor in the single CDAC mode is: The conversion formula of the capacitive sensor in dual CDAC mode is: in, Maxcount=N sub *TxCLK Div Rawdata is the touch data, Maxcount is the maximum touch capacitance data collected by the capacitance sensor; N sub is the sub-conversion number of the capacitive sensor, TxCLK Div is the Tx clock division frequency of the capacitive sensor, CP is the parasitic capacitance value of the capacitive sensor, CF is the finger capacitance value collected by the capacitive sensor, Cref is the reference capacitance value of the capacitive sensor, and CompClkDiv is the compensation CDAD division frequency of the capacitive sensor.

4. The method for detecting hands-off steering wheel according to claim 1, characterized in that: The step of “cleaning the acquired difference data and adding classification labels to the cleaned difference data” includes the following steps: Invalid data and abnormal data in the difference data are filtered out by a digital filter to form a data set; The filtered difference data is classified according to the preset classification rules and corresponding classification labels are added to form a label set.

5. The method for detecting hands-off steering wheel according to claim 4, characterized in that: The “using a neural network to train the difference data with classification labels to obtain a detection model, and deploying the trained detection model to the controller” includes the following steps: Select part of the data from all the difference data as the training set, and another part of the data as the test set; Import the training set and test set in Python, and load the difference data and corresponding label set in the training set; The detection model is trained for multiple rounds using the difference data in the training set and the corresponding label set to generate a trained detection model, which is then deployed to the controller.

6. The method for detecting hands-off steering wheel according to claim 1, characterized in that: The detection model is a 5-layer fully connected neural network, which includes an input layer, 3 hidden layers and an output layer, the input layer includes 3 nodes, each of the hidden layers includes 16 nodes, and the output layer includes 5 nodes, wherein the hidden layer is activated by a Relu activation function, and the output layer is activated by a Softmax activation function.

7. The method for detecting hands-off steering wheel according to claim 6, characterized in that: The Relu activation function is: Relu(x)=max(0,x) The Softmax activation function is: The loss function of the fully connected neural network is:

8. A hands-off-steering-wheel detection system, used to implement the hands-off-steering-wheel detection method according to any one of claims 1 to 7, characterized in that: include: A data acquisition module, used for acquiring reference capacitance data and touch capacitance data; Data processing and annotation module, used to calculate difference data and classify and annotate the difference data; The model training and deployment module is used to train the detection model using difference data with classification labels and deploy the trained detection model to the controller; The detection module is used to collect current touch capacitance data, input the current touch capacitance data into the detection model to obtain the touch gesture classification result, and judge whether the steering wheel is released according to the touch gesture classification result.

9. A computer-readable storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 7.

10. A smart car, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 7.

Images