Voice braking method

Abstract

The invention relates to the technical field of artificial intelligence, in particular to a voice braking method, which allows a user to customize a braking triggering sound according to personal preference and driving habits, the voice can be a clear voice instruction or a specific voice feature, such as scream or shouting, so that the flexibility and applicability of the method are improved; sound features in emergency situations such as screaming and shouting are deeply analyzed, and a more accurate sound feature model is established; therefore, the method can quickly respond when a driver or a passenger makes an emergency sound characteristic, the emergency braking response time is shortened, the probability and severity of accidents are reduced, and the problem that in the prior art, an emergency braking mode mainly depends on a physical sensor and the operation behavior of the driver, and the safety is poor can be effectively solved. The technical problems that the potential information source, namely the sound characteristics of people in the vehicle, is not fully utilized for braking, so that the information source is limited, and the emergency braking response is lagged are solved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a voice braking method. Background Art

[0002] In the current car driving environment, driving safety is the primary consideration. The root cause of some traffic accidents is the driver's limited reaction speed or decision-making ability in emergency situations. Especially when encountering sudden danger, the driver may be unable to perform emergency braking quickly and accurately due to tension, panic or misjudgment.

[0003] The existing emergency braking method monitors the obstacles around the vehicle and the driver's operating behavior, and uses sensor fusion technology and algorithms to determine whether emergency braking needs to be triggered.

[0004] However, existing emergency braking methods mainly rely on physical sensors and the driver's operating behavior, and do not fully utilize the voice characteristics of people in the car as a potential source of information for braking, resulting in limitations in information sources and delayed emergency braking response. Summary of the invention

[0005] The purpose of the present invention is to provide a method for voice braking, aiming to solve the technical problem that the emergency braking method in the prior art mainly relies on physical sensors and the driver's operating behavior, and does not fully utilize the voice characteristics of people in the car as a potential information source for braking, resulting in limitations in the information source and a relatively delayed emergency braking response.

[0006] To achieve the above object, the present invention adopts a voice braking method, comprising the following steps:

[0007] Step 1: Use the vehicle's built-in data acquisition device to collect the voice signals of the people in the vehicle in real time;

[0008] Step 2: preprocessing the collected sound signal using a preprocessing module;

[0009] Step 3: extracting feature vectors from the preprocessed sound signal;

[0010] Step 4: matching the extracted feature vector with a preset braking sound or specific sound feature model to identify whether a braking sound or specific sound feature is emitted;

[0011] Step 5. When the braking sound or specific sound characteristics are recognized, the emergency braking signal is sent to the IVI (In-Vehicle Infotainment) system through the vehicle's internal communication network. After receiving the emergency braking signal, the IVI system sends an emergency braking command to the BCM, and the BCM controls the vehicle's braking system to perform an emergency braking operation.

[0012] Among them, in step one, the data acquisition device is a high-sensitivity microphone array installed in the vehicle, which is used to collect the sound signals of people in the vehicle in real time, and the data acquisition device uses noise reduction technology for noise reduction. These devices can accurately capture and transmit sound signals, and at the same time transmit the sound signals after noise reduction to the subsequent sound signal preprocessing.

[0013] Among them, the method of preprocessing the collected sound signal includes denoising, filtering, framing and windowing operations, and the specific operations are as follows:

[0014] Adaptive filtering algorithm is used to reduce the effect of noise by estimating and subtracting the noise signal;

[0015] A bandpass filter is used to allow signals within a specific frequency range to pass;

[0016] After filtering is completed, the sound signal is framed using sliding window technology;

[0017] The Hamming window, Hanning window or rectangular window technology is used to smooth the sound signal after frame processing to reduce spectrum leakage and distortion at the frame boundary.

[0018] The specific method of extracting the feature vector from the preprocessed sound signal is as follows:

[0019] Use speech recognition algorithm to extract features from preprocessed sound signals;

[0020] The extracted feature vector is a feature vector with discrimination, which specifically includes Mel frequency cepstral coefficients, linear prediction coding, spectrum centroid, spectrum flatness, and roll-off point.

[0021] Among them, the speech recognition algorithm used for feature extraction of the preprocessed sound signal is the Long Short-Term Memory Network (LSTM) and / or the Convolutional Recurrent Neural Network (CRNN).

[0022] In step 4, the extracted feature vector is matched with a preset braking sound or specific sound feature model to identify whether the braking sound or specific sound feature is emitted. The specific method is as follows:

[0023] Using a multi-feature fusion algorithm, a similarity calculation method is used to calculate the similarity between the extracted feature vector and the feature vector in the template library;

[0024] Set a similarity threshold. When the calculated similarity exceeds the threshold, the match is successful.

[0025] Among them, the similarity calculation method used for calculating the similarity between the extracted feature vector and the feature vector in the template library adopts cosine similarity, Euclidean distance, dynamic time warping or hidden Markov model, the braking sound includes "brake, stop", etc., and the specific sound features include "screaming, shouting", etc.

[0026] Among them, in step five, BCM controls the vehicle's braking system to perform emergency braking operations, monitors the vehicle status in real time to ensure a smooth and safe braking process, and the linkage functions of the IVI vehicle system and other vehicle safety systems, such as ESP (electronic stability program) and ABS (anti-lock braking system), provide more comprehensive safety protection. In addition, during the execution of emergency braking operations, the safety verification and feedback module monitors the status of the braking system in real time and provides prompts and feedback on emergency braking operations to the driver.

[0027] Among them, in step five, the vehicle's internal communication network adopts a redundant setting to ensure that the system can still work normally when a single component fails, thereby improving the stability and reliability of the system.

[0028] Among them, in the transmission process of each step, the collected sound signal and the transmitted data are encrypted and transmitted using end-to-end encryption. The specific implementation method is as follows:

[0029] Establish an encrypted connection between the sender and the receiver, ensuring that the data is encrypted during transmission.

[0030] The present invention provides a voice braking method, which is used together with the existing braking method. The present invention allows the user to customize the sound that triggers the brakes according to personal preferences and driving habits. It can be either a clear voice command or a specific sound feature, such as screaming or shouting, thereby improving the flexibility and applicability of the method; in addition, the introduction of speech synthesis technology allows the user to record his or her own voice command as a trigger condition, further enhancing the personalization of the method, deeply analyzing the sound features in emergency situations such as screaming and shouting, and establishing a more accurate sound feature model; at the same time, the introduction of emotion recognition technology identifies emergency braking needs by analyzing the emotional information in the sound, thereby improving the intelligence and accuracy of the system, so that the method can respond quickly when the driver or passenger issues an emergency sound feature, shortening the emergency braking response time, reducing the probability and severity of accidents, and ensuring that the system can still work normally when a single component fails by adopting a redundant setting for the vehicle's internal communication network, thereby improving the system The stability and reliability of the system are improved by encrypting the collected sound signals and transmitted data with end-to-end encryption during the transmission process of each step to ensure user privacy and data security. When the BCM controls the vehicle's braking system to perform emergency braking operations, it monitors the vehicle status in real time to ensure a smooth and safe braking process. The IVI car system is linked with other on-board safety systems, such as ESP (electronic stability program) and ABS (anti-lock braking system), to provide more comprehensive safety protection. During the emergency braking operation, the safety verification and feedback module monitors the status of the braking system in real time and provides the driver with prompts and feedback on the emergency braking operation. In this way, the technical problem that the emergency braking method in the existing technology mainly relies on physical sensors and the driver's operating behavior, and does not fully utilize the sound characteristics of the people in the car as a potential information source for braking, resulting in limited information sources and delayed emergency braking response can be effectively solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0032] Figure 1 It is a flow chart of the steps of the voice braking method of the present invention. DETAILED DESCRIPTION

[0033] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0034] See also Figure 1 , Figure 1 It is a flow chart of the steps of the voice braking method of the present invention.

[0035] The present invention provides a method for voice braking, comprising the following steps:

[0036] S101, collecting voice signals of people in the vehicle in real time through a data collection device built into the vehicle;

[0037] According to this specific implementation, the data acquisition device is a high-sensitivity microphone array installed in the vehicle, which is used to collect the sound signals of the people in the vehicle in real time, and the data acquisition device uses noise reduction technology to reduce noise.

[0038] S102, preprocessing the collected sound signal using a preprocessing module;

[0039] According to this specific implementation, the method of preprocessing the collected sound signal includes denoising, filtering, framing and windowing operations, and the specific operations are as follows:

[0040] Adaptive filtering algorithm is used to reduce the effect of noise by estimating and subtracting the noise signal;

[0041] A bandpass filter is used to allow signals within a specific frequency range to pass;

[0042] After filtering is completed, the sound signal is framed using sliding window technology;

[0043] The Hamming window, Hanning window or rectangular window technology is used to smooth the sound signal after frame processing to reduce spectrum leakage and distortion at the frame boundary.

[0044] Through the above preprocessing operations, the purity and quality of the sound signal can be significantly improved, providing a reliable basis for subsequent sound feature extraction and pattern matching. At the same time, these preprocessing operations can also be flexibly adjusted and optimized according to the characteristics of the sound signal and the needs of subsequent processing to adapt to different application scenarios and sound characteristics.

[0045] S103, the specific method of extracting the feature vector from the pre-processed sound signal is as follows:

[0046] Use speech recognition algorithm to extract features from preprocessed sound signals;

[0047] The extracted feature vector is a feature vector with discrimination, which specifically includes Mel frequency cepstral coefficients, linear prediction coding, spectrum centroid, spectrum flatness, and roll-off point.

[0048] Among them, the speech recognition algorithm used for feature extraction of the preprocessed sound signal is the Long Short-Term Memory Network (LSTM) and / or the Convolutional Recurrent Neural Network (CRNN).

[0049] S104, matching the extracted feature vector with a preset braking sound or specific sound feature model to identify whether a braking sound or specific sound feature is emitted;

[0050] For this specific implementation, the extracted feature vector is matched with a preset brake sound or specific sound feature model to identify whether a brake sound or specific sound feature is emitted. The specific method is as follows:

[0051] Using a multi-feature fusion algorithm, a similarity calculation method is used to calculate the similarity between the extracted feature vector and the feature vector in the template library;

[0052] Set a similarity threshold. When the calculated similarity exceeds the threshold, the match is successful.

[0053] Among them, the similarity calculation method used for calculating the similarity between the extracted feature vector and the feature vector in the template library adopts cosine similarity, Euclidean distance, dynamic time warping or hidden Markov model, the braking sound includes "brake, stop", etc., and the specific sound features include "screaming, shouting", etc.

[0054] S105. When the braking sound or specific sound features are recognized, an emergency braking signal is sent to the IVI vehicle system through the vehicle's internal communication network. After receiving the emergency braking signal, the IVI vehicle system sends an emergency braking command to the BCM, and the BCM controls the vehicle's braking system to perform an emergency braking operation.

[0055] For this specific implementation, the BCM controls the vehicle's braking system to perform emergency braking operations, monitors the vehicle status in real time, ensures a smooth and safe braking process, and provides a more comprehensive safety guarantee by linking the IVI vehicle system with other vehicle safety systems, such as ESP (electronic stability program) and ABS (anti-lock braking system);

[0056] The vehicle's internal communication network adopts a redundant setting to ensure that the system can still work normally when a single component fails, improving the stability and reliability of the system.

[0057] In the transmission process of each step, the collected sound signal and the transmitted data are encrypted and transmitted using end-to-end encryption. The specific implementation method is as follows:

[0058] Establish an encrypted connection between the sender and the receiver, ensuring that the data is encrypted during transmission.

[0059] At the same time, in actual use, visual assisted verification is used to improve the reliability of emergency braking instructions. The specific methods are as follows:

[0060] The driver's facial expression and gesture images are captured in real time through the on-board camera; the images are pre-processed, including denoising, enhancement, cropping, etc., to improve the accuracy and efficiency of subsequent processing.

[0061] Based on deep learning algorithms, such as convolutional neural networks (CNN), facial expression recognition is performed on the pre-processed images; the driver's facial expression characteristics, such as tension, panic, fatigue, etc., are identified, which may be related to the issuance of emergency braking commands.

[0062] At the same time, gesture detection is performed on the preprocessed image, and the algorithm is used to determine the action and position of the gesture.

[0063] Identify the driver's gesture information, such as shape, direction, speed, etc.

[0064] At this time, if the braking sound or specific sound features are recognized, but the image processing module does not detect the corresponding facial expression or gesture features, the system will first issue a confirmation prompt to the driver. If the driver does not respond within the set threshold, emergency braking will be performed.

[0065] The voice braking method of the present invention requires all users to perform identity verification before use, for example, by entering a preset password, PIN code, biometric identification (such as fingerprint, facial recognition), etc.

[0066] For the owner or primary user of the vehicle, a higher level of authentication, such as two-factor authentication or multi-factor authentication, can be set up to ensure the security of their access rights.

[0067] In summary, by strictly managing the permissions used in the voice braking method, adopting identity authentication and authorization mechanisms, system access control, data encryption and secure storage, software updates and security maintenance, as well as training and publicity measures, the security and reliability of the voice braking method can be ensured and unauthorized access and operation can be prevented.

[0068] The present invention can continuously learn and optimize, adapt to the voice characteristics and driving habits of different users, improve the accuracy and reliability of recognition, and regularly update the software version of the system to fix known security vulnerabilities and defects. However, during the update process, the stability and compatibility of the system are ensured to avoid affecting the normal use of users.

[0069] A voice braking method using the present embodiment is used together with the existing braking method. The present invention allows the user to customize the sound that triggers the brakes according to personal preferences and driving habits. It can be either a clear voice command or a specific sound feature, such as screaming or shouting, thereby improving the flexibility and applicability of the method; in addition, the introduction of speech synthesis technology allows the user to record his or her own voice command as a trigger condition, further enhancing the personalization of the method, deeply analyzing the sound features of emergency situations such as screaming and shouting, and establishing a more accurate sound feature model; at the same time, the introduction of emotion recognition technology identifies emergency braking needs by analyzing the emotional information in the sound, thereby improving the intelligence and accuracy of the system, so that the method can respond quickly when the driver or passenger issues an emergency sound feature, shortening the emergency braking response time, reducing the probability and severity of accidents, and using a redundant setting for the vehicle's internal communication network to ensure that the system can still work normally when a single component fails, thereby improving High system stability and reliability. In the transmission process of each step, the collected sound signals and transmitted data are encrypted and transmitted end-to-end to ensure user privacy and data security. During the emergency braking operation, the BCM controls the vehicle's braking system to monitor the vehicle status in real time to ensure a smooth and safe braking process. The IVI car system is linked with other on-board safety systems, such as ESP (electronic stability program) and ABS (anti-lock braking system), to provide more comprehensive safety protection. In the emergency braking operation, the safety verification and feedback module monitors the status of the braking system in real time and provides the driver with prompts and feedback on the emergency braking operation. In this way, the problem that the emergency braking method in the existing technology mainly relies on physical sensors and the driver's operating behavior and does not fully utilize the potential information source of the sound characteristics of the people in the car for braking, resulting in limited information sources and delayed emergency braking response can be effectively solved.

[0070] What is disclosed above is only a preferred embodiment of the present invention, and it certainly cannot be used to limit the scope of rights of the present invention. Ordinary technicians in this field can understand that all or part of the processes of the above embodiment and equivalent changes made according to the claims of the present invention still fall within the scope of the invention.

Claims

1. A method of voice braking, It is characterized in that The following steps are involved: Step 1: Use the vehicle's built-in data acquisition device to collect the voice signals of the people in the vehicle in real time; Step 2: preprocessing the collected sound signal using a preprocessing module; Step 3: extracting feature vectors from the preprocessed sound signal; Step 4: matching the extracted feature vector with a preset braking sound or specific sound feature model to identify whether a braking sound or specific sound feature is emitted; Step 5. When the braking sound or specific sound characteristics are recognized, the emergency braking signal is sent to the IVI vehicle system through the vehicle's internal communication network. After receiving the emergency braking signal, the IVI vehicle system sends an emergency braking command to the BCM, and the BCM controls the vehicle's braking system to perform emergency braking operations.

2. The method for voice braking according to claim 1, characterized in that: In step one, the data acquisition device is a high-sensitivity microphone array installed in the vehicle, which is used to collect the sound signals of people in the vehicle in real time, and the data acquisition device uses noise reduction technology to reduce noise.

3. The method of voice braking as claimed in claim 2, characterized in that: The methods of preprocessing the collected sound signals include denoising, filtering, framing and windowing operations. The specific operations are as follows: Adaptive filtering algorithm is used to reduce the effect of noise by estimating and subtracting the noise signal; A bandpass filter is used to allow signals within a specific frequency range to pass; After filtering is completed, the sound signal is framed using sliding window technology; The Hamming window, Hanning window or rectangular window technology is used to smooth the sound signal after frame processing to reduce spectrum leakage and distortion at the frame boundary.

4. The method for voice braking as claimed in claim 3, characterized in that: The specific method of extracting the feature vector from the preprocessed sound signal is as follows: Use speech recognition algorithm to extract features from preprocessed sound signals; The extracted feature vector is a feature vector with discrimination, which specifically includes Mel frequency cepstral coefficients, linear prediction coding, spectrum centroid, spectrum flatness, and roll-off point.

5. The method for voice braking as claimed in claim 4, characterized in that: The speech recognition algorithm used for feature extraction of preprocessed sound signals is a long short-term memory network or a convolutional recurrent neural network.

6. The method for voice braking as claimed in claim 5, characterized in that: In step 4, the extracted feature vector is matched with a preset braking sound or specific sound feature model to identify whether a braking sound or specific sound feature is emitted. The specific method is as follows: Using a multi-feature fusion algorithm, a similarity calculation method is used to calculate the similarity between the extracted feature vector and the feature vector in the template library; Set a similarity threshold. When the calculated similarity exceeds the threshold, the match is successful.

7. The method for voice braking as claimed in claim 6, characterized in that: The similarity calculation method used for calculating the similarity between the extracted feature vector and the feature vector in the template library adopts cosine similarity, Euclidean distance, dynamic time warping or hidden Markov model.

8. The method for voice braking as claimed in claim 7, characterized in that: In step five, the BCM controls the vehicle's braking system to perform emergency braking operations, and monitors the vehicle status in real time to ensure a smooth and safe braking process.

9. The method for voice braking as claimed in claim 8, characterized in that: In step five, a redundant configuration is adopted for the vehicle's internal communication network.

10. The method for voice braking as claimed in claim 9, characterized in that In the transmission process of each step, the collected sound signal and the transmitted data are encrypted and transmitted using end-to-end encryption. The specific implementation method is as follows: Establish an encrypted connection between the sender and the receiver, ensuring that the data is encrypted during transmission.

Images