A voice recognition emergency braking system and method for intelligent driving vehicles

Abstract

The invention discloses a voice recognition emergency braking system and method for an intelligent driving vehicle, relates to the technical field of voice recognition emergency braking, and solves the technical problems that, in performing emergency braking through voice recognition, different characteristics of the volume in a vehicle for different users in different time periods are ignored, and emergency braking is performed according to different characteristics, and it is difficult to perform self-optimization in the division of time periods. The invention obtains sound data when a vehicle is driving; dynamically plans time periods of each day, analyzes historical volume in the vehicle to obtain characteristic volume in each time period inside the vehicle, obtains discrimination factors based on characteristic volume, volume in the vehicle, human tone and sound data outside the vehicle; obtains control instructions based on human voice text and the discrimination factors; and brakes the vehicle based on the control instructions. The invention can improve the self-adaptability and accuracy of the voice recognition emergency braking system for an intelligent driving vehicle.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent driving, and relates to a voice recognition emergency braking technology, in particular to a voice recognition emergency braking system and method for an intelligent driving vehicle. Background Art

[0002] As road traffic becomes increasingly complex, drivers have higher and higher requirements for the convenience and safety of vehicles. Especially in urban traffic, drivers' attention may be distracted, resulting in failure to take timely action in emergency situations. Traditional emergency braking systems usually rely on physical buttons or touch screen operations, which may cause operation delays in certain emergency situations. With the development of intelligent driving technology, vehicles need to have more intelligent and efficient emergency response capabilities during driving to cope with various emergencies. The voice recognition emergency braking system can quickly initiate emergency braking through voice information when the driver cannot react in time, reducing the risk of accidents.

[0003] At present, most smart driving car voice recognition emergency braking systems ignore the fact that different users have different volume characteristics in the car during different time periods during voice recognition emergency braking, and perform emergency braking based on different characteristics. Ignoring this situation may cause the system to mistakenly identify non-emergency voice information as emergency voice information, resulting in unnecessary emergency braking, increasing the error recognition rate and affecting driving safety. At the same time, most smart driving car voice recognition emergency braking systems, when planning different time periods for different users, find it difficult to self-optimize the time periods during division, which reduces the system's self-adaptability and intelligence in time period division.

[0004] Therefore, the present invention discloses a voice recognition emergency braking system and method for an intelligent driving vehicle, which are used to solve the above technical problems. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes a voice recognition emergency braking system and method for an intelligent driving vehicle, which is used to solve the technical problems that, in emergency braking through voice recognition, different characteristics of the volume in the car for different users in different time periods are ignored, and emergency braking is performed according to different characteristics, and it is difficult to make the time period self-optimized in the division. The present invention obtains sound data when the vehicle is driving; dynamically plans the time period of each day, analyzes the historical volume in the car to obtain the characteristic volume in each time period inside the vehicle, obtains a discrimination factor based on the characteristic volume, the volume in the car, the tone of the person and the sound data outside the car; obtains a control instruction based on the person's voice text and the discrimination factor; and brakes the vehicle based on the control instruction to solve the above problems.

[0006] To achieve the above-mentioned object, the first aspect of the present invention provides a voice recognition emergency braking system for an intelligent driving vehicle, comprising: a voice recognition module, and a data collection module, a vehicle control module and a database connected thereto;

[0007] The data collection module is used to obtain the sound data of the vehicle when it is driving; the sound data includes the sound data inside the vehicle and the sound data outside the vehicle; the sound data inside the vehicle includes the volume inside the vehicle, the tone of the personnel and the voice text of the personnel; the sound data outside the vehicle includes the volume of pedestrians, the frequency of honking of other vehicles and the maximum duration of a single honking;

[0008] The speech recognition module is used to dynamically plan the time periods of each day, analyze the historical in-vehicle volume to obtain the characteristic volume of each time period inside the vehicle, obtain the discrimination factor 1 based on the characteristic volume, the in-vehicle volume and the tone of the personnel, and analyze the sound data outside the vehicle to obtain the discrimination factor 2;

[0009] The vehicle control module is used to obtain control instruction one according to the voice and text of the personnel, and to obtain control instruction two according to discrimination factor one and discrimination factor two; and to brake the vehicle based on control instruction one and control instruction two.

[0010] Preferably, the step of obtaining the sound data of the vehicle while it is traveling comprises:

[0011] Based on user authorization, the real-time volume inside the car is obtained through the volume sensor, and the real-time volume is marked as the volume inside the car; the tone and voice of the person are obtained through the on-board microphone, and the voice of the person is converted into text using natural language recognition technology, and the text is marked as the person's voice text; the volume of pedestrians outside the car, the frequency of honking of other vehicles, and the maximum duration of a single honking are obtained through a number of intelligent acoustic sensors.

[0012] Preferably, the dynamic planning of the time periods of each day includes:

[0013] A1: At a specific time on the day, the position of the next day in the week is obtained, and several days in the historical data with the same day position as the next day are marked as reference days. The time points at which the vehicle speed is not 0 in each reference day are marked as target time points. The continuous target time points are integrated into target time periods; the continuous non-target time points are integrated into non-target time periods, and the non-target time periods are planned into several time periods according to fixed time lengths; wherein the fixed time length is obtained by manual setting;

[0014] A2: Get the number of times each target time point exists in the history, mark the target time point corresponding to the number of times equal to the threshold as the boundary point, divide each target period into several sub-periods based on the boundary points, and determine whether the duration of each sub-period is less than the fixed duration. If yes, jump to A3; if no, based on the formula HS= ZS / GS The number of divisions HS is obtained, and the sub-period is evenly divided into several time periods based on the number of divisions HS; wherein the number threshold is obtained by manual setting; GS is a fixed duration, and ZS is the duration of the sub-period. is the rounding symbol;

[0015] A3: Mark the sub-periods that are less than the fixed duration and greater than the fixed duration multiplied by the proportional coefficient BL as time periods; combine the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are continuous into several time periods from front to back in time, and the duration of the combined several time periods is not greater than the value of the fixed duration multiplied by the proportional coefficient BL, and mark the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are separate as time periods; where the value range of the proportional coefficient BL is (0,1).

[0016] It is worth noting that different users have different driving times. Some users only drive when commuting, while some users need to drive all day. If the same time period is used for analysis of all users, the designed time period will not be able to meet the needs of all users. The present invention obtains the personalized driving time of each user by analyzing the historical data of each user. When dividing the time period for each user according to the personalized time, it can make a more detailed division within the time range of the user's frequent driving, so that the data analysis of the time range of the user's frequent driving is more accurate; the time range of the user's infrequent driving is divided into standard time lengths to save the system's computing resources; at the same time, the present invention is dynamic in obtaining time periods for different users, and can adaptively optimize the time periods of each day according to the user's personal data, so that the time period division of the system has a dynamic correction process, which increases the scientific nature of time division.

[0017] Preferably, the step of analyzing the historical in-vehicle volume to obtain characteristic volume of the vehicle in each time period includes:

[0018] Based on the time range of each time period, the historical in-car volume of the corresponding time period is extracted from the database, and the historical in-car volume is divided into several volume groups based on the time period; the variance of each volume group is extracted in turn, and it is judged whether the variance is less than the variance judgment threshold; if yes, the data of the volume group is retained; if no, the volume with the largest difference from the mode of the volume in the volume group is removed, and the variance is judged again until the variance of the volume group is less than the variance judgment threshold, and the remaining data of the volume group is retained; wherein the variance judgment threshold is obtained through experience;

[0019] The maximum value DZ, average value PJ and mode ZS of the retained data in each volume group are extracted, and the characteristic volume TL of the corresponding time period is obtained based on the formula TL=α1×DZ+α2×PJ+α3×ZS; wherein α1, α2 and α3 are all proportional adjustment coefficients greater than 0, and the value ranges of α1, α2 and α3 are all [0,1], α1<α2<α3.

[0020] It is worth noting that due to different driving habits of different users or different driving habits of the same user at different times, the characteristics of the volume in the car of different users at different times may be different; for example: User 1 likes to listen to music when driving, and User 2 likes a quiet environment when driving, which will result in different characteristics of the volume in the car of User 1 and User 2; User 3 has to drive to pick up children at a fixed time on each weekday, and User 3 likes a quiet environment when driving by himself, and likes to talk to the children after getting in the car, which will result in different characteristics of the volume in the car of the same user at different times; if the characteristic value of the volume in the car in a certain time period is high, it indicates that the user's living habits make the car have a high volume in the current time period. If the volume determination threshold is not dynamically set in this time period, it will lead to frequent erroneous braking. Therefore, the present invention extracts characteristic values ​​for each time period, and uses the characteristic values ​​of each time period as the volume determination threshold to analyze whether the current volume in the car is reasonable, thereby avoiding the situation where the system sets fixed thresholds for different users in different time periods, which easily leads to erroneous braking, and improves the stability of the system and the self-adaptability to different users.

[0021] Preferably, the first discrimination factor is obtained based on the characteristic volume, the volume in the car and the tone of the person, including:

[0022] B1: extract the characteristic volume TL of the time period of the current time and the in-car volume CL of the current time, and determine whether the in-car volume CL is less than the characteristic volume TL; if yes, do nothing; if no, jump to B2;

[0023] B2: Extract the in-car volume CL and personnel voice pitch RD at the current time, and obtain the discrimination factor BY1 based on the formula BY1=δ×exp(β1×(CL-BCL) / BCL+β2×(RD-BRD) / BRD); wherein BCL is the manually set standard in-car volume, and BRD is the manually set standard personnel voice pitch; δ is the amplitude adjustment factor of the exp() function, and the value range of δ is (0,2); β1 and β2 are both proportional adjustment coefficients greater than 0, and β1+β2=1.

[0024] Preferably, the analysis of the external sound data to obtain the second discriminant factor includes:

[0025] C1: Extract the pedestrian volume RY at the current time, and determine whether the placement position of the intelligent acoustic sensor that first receives the pedestrian volume RY is located in front of or behind the vehicle; if yes, the pedestrian volume factor RZ is obtained based on the formula RZ=μ1×(ln((RY-BRY) / BRY+1)+1); if no, the pedestrian volume factor RZ is obtained based on the formula RZ=μ2×(ln((RY-BRY) / BRY+1)+1); where BRY is the manually set standard pedestrian volume, μ1 and μ2 are the amplitude adjustment coefficients of the ln() function, and the value range of μ1 and μ2 are both (0,2), μ1>μ2;

[0026] The horn frequency DP and the maximum duration of a single horn DC of other vehicles at the current time and within n seconds before the current time are extracted, and the discrimination factor BY2 is obtained based on the formula BY2=RZ+ρ×exp(σ1×DP / BDP+σ2×DC / BDC); wherein n is obtained by manual setting, ρ is the amplitude adjustment factor of the exp() function, and the value range of ρ is (0,2); σ1 and σ2 are both proportional adjustment coefficients greater than 0, and σ1+σ2=1, σ1>σ2; BDP is the manually set standard horn frequency, and BDC is the manually set standard maximum duration of a single horn.

[0027] It is worth noting that the pedestrian volume coming from the front or rear of the car proves that the pedestrian is most likely located in front of or behind the car. At this time, the risk of pedestrians being hit will be greater than that on both sides of the car. Therefore, the pedestrian volume factor used by the present invention when obtaining the second discrimination factor is first distinguished by the position of the pedestrian, and then calculated. This is conducive to obtaining different danger values ​​for pedestrians in different positions, which can optimize the resource allocation of the emergency braking system, focus more attention and warning resources on areas with higher risks, and improve the safety of pedestrians and drivers.

[0028] Preferably, the step of obtaining a control instruction according to the voice and text of a person includes:

[0029] The keywords set by the user according to his / her own idiomatic expressions are marked as keyword library 1, the keywords of the in-car voices during the car accidents in history are counted to obtain keyword library 2, and keyword library 1 and keyword library 2 are integrated into an emergency word library;

[0030] Extract the personnel's voice and text, and determine whether the personnel's voice and text are keywords in the emergency vocabulary; if yes, issue an operation instruction 1; if no, do nothing.

[0031] Preferably, obtaining control instruction 2 according to discriminant factor 1 and discriminant factor 2 includes:

[0032] Extract the discriminant factor BY1, and determine whether the discriminant factor BY1 exceeds the factor threshold value 1; if yes, issue the control instruction 2; if no, do nothing;

[0033] Extract the second discriminant factor BY2, and determine whether the second discriminant factor BY2 exceeds the factor threshold value 2; if yes, issue the second control instruction; if no, do nothing; wherein the mean of the factor threshold value 1 and the factor threshold value 2 is obtained by manual setting.

[0034] Preferably, the braking of the vehicle based on the control instruction 1 and the control instruction 2 includes:

[0035] D1: Get the current vehicle speed and determine whether the vehicle speed exceeds the low-speed judgment threshold; if yes, do not perform any operation; if no, jump to D2; wherein the low-speed judgment threshold is obtained by manual setting;

[0036] D2: When control instruction 1 or control instruction 2 is received, the current vehicle is decelerated according to a preset speed until the vehicle stops; wherein the preset speed is obtained according to the current vehicle speed.

[0037] A second aspect of the present invention provides a voice recognition emergency braking method for an intelligent driving vehicle, comprising the following steps:

[0038] S1: Acquire the sound data of the vehicle when it is driving;

[0039] S2: Dynamically plan the time periods of each day, analyze the historical in-car volume to obtain the characteristic volume of each time period inside the vehicle, obtain the discriminant factor 1 based on the characteristic volume, the in-car volume and the voice tone of the personnel, and analyze the sound data outside the vehicle to obtain the discriminant factor 2;

[0040] S3: Obtain control instruction one according to the voice and text of the person, and obtain control instruction two according to discrimination factor one and discrimination factor two; brake the vehicle based on control instruction one and control instruction two.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] 1. The present invention obtains the sound data of the vehicle when it is driving; dynamically plans the time periods of each day, analyzes the historical in-vehicle volume to obtain the characteristic volume of each time period inside the vehicle, obtains the discrimination factor based on the characteristic volume, the in-vehicle volume, the tone of the personnel and the sound data outside the vehicle; obtains the control instruction according to the personnel voice text and the discrimination factor; brakes the vehicle based on the control instruction, thereby solving the technical problems of ignoring the different characteristics of the in-vehicle volume of different users in different time periods in emergency braking through voice recognition, performing emergency braking according to different characteristics, and making it difficult to self-optimize the time period in the division; the present invention can improve the self-adaptability and accuracy of the voice recognition emergency braking system of the intelligent driving vehicle.

[0043] 2. The present invention accurately identifies the personalized driving time of each user by deeply analyzing the historical data of each user. When dividing the time period according to the personalized time, the system can make a more detailed division within the time range when the user frequently drives, thereby ensuring that the data analysis results within the time period are more accurate and reliable. At the same time, within the time range when the user drives less frequently, the system uses standard time lengths for division, which effectively saves computing resources and improves overall efficiency; in addition, the present invention has the ability to adjust dynamic time periods, and can adaptively optimize the daily time period according to the user's personal data. This dynamic adjustment process not only ensures the flexibility and accuracy of time period division, but also further enhances the scientific nature of time management.

[0044] 3. The present invention extracts characteristic values ​​for each time period, and uses the characteristic values ​​for each time period as a volume determination threshold to analyze whether the current volume in the car is reasonable. This avoids the situation where the system sets fixed thresholds for different users in different time periods, which easily leads to erroneous braking, and improves the stability of the system and its self-adaptability to different users. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] 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.

[0046] Figure 1 It is a schematic diagram of the operation steps of the present invention;

[0047] Figure 2 It is a schematic diagram of the system module of the present invention;

[0048] Figure 3 It is a schematic diagram of the operation steps of dynamically planning time periods of the present invention. DETAILED DESCRIPTION

[0049] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] See also Figure 1-Figure 2, the first aspect of the present invention provides a voice recognition emergency braking system for an intelligent driving vehicle, comprising: a voice recognition module, and a data collection module, a vehicle control module and a database connected thereto;

[0051] Data collection module: used to obtain sound data when the vehicle is driving; the sound data includes in-vehicle sound data and out-vehicle sound data; in-vehicle sound data includes in-vehicle volume, voice tone and voice text of personnel; out-vehicle sound data includes pedestrian volume, honking frequency of other vehicles and maximum duration of a single honking;

[0052] Speech recognition module: used to dynamically plan the time periods of each day, analyze the historical in-car volume to obtain the characteristic volume of each time period inside the vehicle, obtain the discrimination factor 1 based on the characteristic volume, the in-car volume and the tone of the personnel, and analyze the sound data outside the vehicle to obtain the discrimination factor 2;

[0053] Vehicle control module: used to obtain control instruction one according to the voice and text of the personnel, and obtain control instruction two according to discrimination factor one and discrimination factor two; brake the vehicle based on control instruction one and control instruction two.

[0054] It should be noted that the pitch of a person's voice will become higher when encountering an emergency or being frightened.

[0055] It should be noted that the historical in-car volume refers to the in-car volume in historical data.

[0056] In this application, the sound data of the vehicle driving is obtained, including:

[0057] Based on user authorization, the real-time volume inside the car is obtained through the volume sensor, and the real-time volume is marked as the volume inside the car; the tone and voice of the person are obtained through the on-board microphone, and the voice of the person is converted into text using natural language recognition technology, and the text is marked as the person's voice text; the volume of pedestrians outside the car, the frequency of honking of other vehicles, and the maximum duration of a single honking are obtained through a number of intelligent acoustic sensors.

[0058] See also Figure 3 , the time periods for each day are dynamically planned in this application, including:

[0059] A1: At a specific time on the day, the position of the next day in the week is obtained, and several days in the historical data with the same day position as the next day are marked as reference days. The time points at which the vehicle speed is not 0 in each reference day are marked as target time points. The continuous target time points are integrated into target time periods; the continuous non-target time points are integrated into non-target time periods, and the non-target time periods are planned into several time periods according to fixed time lengths; wherein the fixed time length is obtained by manual setting;

[0060] A2: Get the number of times each target time point exists in the history, mark the target time point corresponding to the number of times equal to the threshold as the boundary point, divide each target period into several sub-periods based on the boundary points, and determine whether the duration of each sub-period is less than the fixed duration. If yes, jump to A3; if no, based on the formula HS= ZS / GS The number of divisions HS is obtained, and the sub-period is evenly divided into several time periods based on the number of divisions HS; wherein the number threshold is obtained by manual setting; GS is a fixed duration, and ZS is the duration of the sub-period. is the rounding symbol;

[0061] A3: Mark the sub-periods that are less than the fixed duration and greater than the fixed duration multiplied by the proportional coefficient BL as time periods; combine the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are continuous into several time periods from front to back in time, and the duration of the combined several time periods is not greater than the value of the fixed duration multiplied by the proportional coefficient BL, and mark the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are separate as time periods; where the value range of the proportional coefficient BL is (0,1).

[0062] It should be noted that the division of time periods for the next day and the acquisition of characteristic volume for each time period for the next day are both performed at a specific time of the day.

[0063] It should be noted that the week in which the next day is obtained is the week in which the next day belongs.

[0064] It should be noted that the time period for each day is planned at a specific time the day before each day.

[0065] It should be noted that the day positions are specifically: Monday is at 1, Tuesday is at 2, Wednesday is at 3, Thursday is at 4, Friday is at 5, Saturday is at 6, and Sunday is at 7.

[0066] It should be noted that the unit of time point is second.

[0067] It should be noted that the time points in each reference day where the vehicle speed is not 0 are marked as target time points, and the number of times each target time point exists in the history is obtained. An example can be given as follows: if there are 6 reference days, and a certain time point is 18:00:00, the vehicle speed at 18:00:00 is not 0 on the 1st reference day and the 6th reference day, so the time point 18:00:00 can be marked as the target time point, and the number of times the target time point 18:00:00 exists in the history is 2; if a certain time point is 19:00:00, the vehicle speed at 19:00:00 is 0 on the 6 reference days, so the time point 19:00:00 is marked as a non-target time point.

[0068] It should be noted that continuous non-target time points are integrated into non-target time periods, and the non-target time periods are planned into several time periods according to a fixed duration. If a non-target time period cannot be completely planned into several time periods according to a fixed duration, the incomplete time period after the last complete time period is marked as one time period; for example: a non-target time period is 14:00-15:00, and the fixed duration is 9 minutes, then the time period planning is: 14:00-14:09, 14:09-14:18, 14:18-14:27, 14:27-14:36, 14:36-14:45, 14:45-14:54, 14:54-15:00.

[0069] It should be noted that the non-target period is planned into several time periods according to a fixed duration. If the duration of the non-target period is less than the fixed duration, the non-target period is marked as one time period.

[0070] It should be noted that each target time period is divided into a number of sub-time periods based on the definition point, and the length of the divided sub-time periods is random, and the length of the sub-time periods may be a few seconds or several hours.

[0071] It should be noted that the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are continuous are combined into several time periods from front to back in time, and the durations of the combined several time periods are not greater than the fixed duration multiplied by the proportional coefficient BL. This sentence can be illustrated as follows:

[0072] If the fixed duration is 10 minutes and the proportional coefficient BL = 0.6, then the fixed duration multiplied by the proportional coefficient BL is 6 minutes;

[0073] When the duration of a sub-period is between 6 and 10 minutes, the current sub-period is marked as a time period;

[0074] When the length of a sub-period is less than 6 minutes, if there are consecutive sub-periods of 17:30:05-17:31:00, 17:31:00-17:32:00, 17:32:00-17:34:00, 17:34:00-17:39:30, 17:39:30-17:41:30, then these consecutive sub-periods are combined into several time periods from front to back according to time: 17:30:05-17:34:00, 17:34:00-17:39:30, 17:39:30-17:41:30, so as to ensure that the length of the combined several time periods is no more than 6 minutes.

[0075] It should be noted that marking a sub-period that is less than the fixed duration multiplied by the proportional coefficient BL as a time period can be illustrated as follows:

[0076] If the fixed duration is 10 minutes and the proportional coefficient BL = 0.6, then the fixed duration multiplied by the proportional coefficient BL is 6 minutes;

[0077] When the duration of a sub-period is less than 6 minutes, if there are separate sub-periods of 17:16:05-17:18:00, 17:19:00-17:30:00, then the separate sub-periods of less than 6 minutes are marked as time periods: 17:16:05-17:18:00, 17:19:00-17:30:00;

[0078] Since the sub-period of 17:18:00-17:19:00 has been evenly divided into several time periods based on the division number HS, there is no need to divide the time periods again.

[0079] In this application, the historical in-vehicle volume is analyzed to obtain the characteristic volume of the vehicle in each time period, including:

[0080] Based on the time range of each time period, the historical in-car volume of the corresponding time period is extracted from the database, and the historical in-car volume is divided into several volume groups based on the time period; the variance of each volume group is extracted in turn, and it is judged whether the variance is less than the variance judgment threshold; if yes, the data of the volume group is retained; if no, the volume with the largest difference from the mode of the volume in the volume group is removed, and the variance is judged again until the variance of the volume group is less than the variance judgment threshold, and the remaining data of the volume group is retained; wherein the variance judgment threshold is obtained through experience;

[0081] The maximum value DZ, average value PJ and mode ZS of the retained data in each volume group are extracted, and the characteristic volume TL of the corresponding time period is obtained based on the formula TL=α1×DZ+α2×PJ+α3×ZS; wherein α1, α2 and α3 are all proportional adjustment coefficients greater than 0, and the value ranges of α1, α2 and α3 are all [0,1], α1<α2<α3.

[0082] It should be noted that since people have a natural aversion to loud volumes, the volume for listening to music and talking will be controlled within a reasonable range. The characteristic value of the volume in the car is obtained based on this reasonable range, so the characteristic value obtained is a value lower than the abnormally high volume.

[0083] In this embodiment, if user 1 retains the maximum value DZ=60dB, the average value PJ=35dB, and the mode ZS=39dB of the data in the volume group corresponding to the time period of 15:00:00-15:06:00, then based on the formula TL=α1×DZ+α2×PJ+α3×ZS=0.2×60+0.3×35+0.5×39=42dB, the characteristic volume TL=42dB for the corresponding time period is obtained; wherein, in this embodiment, the value of the proportional adjustment coefficient α1 is 0.2, the value of the proportional adjustment coefficient α2 is 0.3, and the value of the proportional adjustment coefficient α3 is 0.5.

[0084] In this application, a discrimination factor 1 is obtained based on the characteristic volume, the volume in the car and the tone of the personnel, including:

[0085] B1: Extract the characteristic volume TL of the time period of the current time and the in-car volume CL of the current time, and determine whether the in-car volume CL is less than the characteristic volume TL; if yes, do nothing; if no, jump to B2;

[0086] B2: Extract the in-car volume CL and personnel voice pitch RD at the current time, and obtain the discrimination factor BY1 based on the formula BY1=δ×exp(β1×(CL-BCL) / BCL+β2×(RD-BRD) / BRD); wherein BCL is the manually set standard in-car volume, and BRD is the manually set standard personnel voice pitch; δ is the amplitude adjustment factor of the exp() function, and the value range of δ is (0,2); β1 and β2 are both proportional adjustment coefficients greater than 0, and β1+β2=1.

[0087] In this embodiment, if the characteristic volume TL of the time period at the current time is 42dB, and the in-car volume CL at the current time is 45dB, the personnel tone RD at the current time is extracted as 100Hz, and based on the formula BY1=δ×exp(β1×(CL-BCL) / BCL+β2×(RD-BRD) / BRD)=0.9×exp(0.6×(45-50) / 50+0.4×(100-150) / 150)=0.742, the discrimination factor BY1=0.742 is obtained; wherein, in this embodiment, the value of the amplitude adjustment factor δ is 0.9, the value of the proportional adjustment coefficient β1 is 0.6, the value of the proportional adjustment coefficient β2 is 0.4, the value of the standard in-car volume BCL is 50dB, and the value of the standard personnel tone BRD is 150Hz.

[0088] In this application, the external sound data is analyzed to obtain the second discriminant factor, including:

[0089] C1: Extract the pedestrian volume RY at the current time, and determine whether the placement position of the intelligent acoustic sensor that first receives the pedestrian volume RY is located in front of or behind the vehicle; if yes, the pedestrian volume factor RZ is obtained based on the formula RZ=μ1×(ln((RY-BRY) / BRY+1)+1); if no, the pedestrian volume factor RZ is obtained based on the formula RZ=μ2×(ln((RY-BRY) / BRY+1)+1); where BRY is the manually set standard pedestrian volume, μ1 and μ2 are the amplitude adjustment coefficients of the ln() function, and the value range of μ1 and μ2 are both (0,2), μ1>μ2;

[0090] The horn frequency DP and the maximum duration of a single horn DC of other vehicles at the current time and within n seconds before the current time are extracted, and the discrimination factor BY2 is obtained based on the formula BY2=RZ+ρ×exp(σ1×DP / BDP+σ2×DC / BDC); wherein n is obtained by manual setting, ρ is the amplitude adjustment factor of the exp() function, and the value range of ρ is (0,2); σ1 and σ2 are both proportional adjustment coefficients greater than 0, and σ1+σ2=1, σ1>σ2; BDP is the manually set standard horn frequency, and BDC is the manually set standard maximum duration of a single horn.

[0091] It should be noted that the louder the pedestrian's voice is, the closer he is to the car, and the greater the probability that the pedestrian will be hit by the current vehicle.

[0092] It should be noted that the greater the frequency of other vehicles honking their horns or the maximum duration of a single honking, the more it indicates that other drivers are reminding the current driver to pay attention to the road.

[0093] In another embodiment, the pedestrian volume RY=35dB at the current time is extracted. Since the placement position of the intelligent acoustic sensor that first receives the pedestrian volume RY is located in front of the vehicle, based on the formula RZ=μ1×(ln((RY-BRY) / BRY+1)+1)=1×(ln((35-30) / 30+1)+1)=1.15, the pedestrian volume factor RZ=1.15 is obtained; wherein, in this embodiment, the value of the amplitude adjustment coefficient μ1 is 1, and the value of the standard pedestrian volume BRY is 30dB;

[0094] The honking frequency DP=5 times / second and the maximum duration of a single honking DC=1.06 seconds of other vehicles at the current time and within 3 seconds before the current time are extracted, and based on the formula BY2=RZ+ρ×exp(σ1×DP / BDP+σ2×DC / BDC)=1.15+0.2×exp(0.6×5 / 3+0.4×1.06 / 1)=1.98, the discrimination factor BY2=1.98 is obtained; wherein, in this embodiment, the value of the amplitude adjustment factor ρ is 0.2, the value of the proportional adjustment coefficient σ1 is 0.6, the value of the proportional adjustment coefficient σ2 is 0.4, the value of the standard honking frequency BDP is 3 times / second, and the value of the standard maximum duration of a single honking BDC is 1 second.

[0095] In this application, a control instruction 1 is obtained according to the voice and text of the personnel, including:

[0096] The keywords set by the user according to his / her own idiomatic expressions are marked as keyword library 1, the keywords of the in-car voices during the car accidents in history are counted to obtain keyword library 2, and keyword library 1 and keyword library 2 are integrated into an emergency word library;

[0097] Extract the personnel's voice and text, and determine whether the personnel's voice and text contain keywords in the emergency vocabulary; if yes, issue an operation instruction 1; if no, do nothing.

[0098] In this application, the control instruction 2 is obtained according to the discriminant factor 1 and the discriminant factor 2, including:

[0099] Extract the discriminant factor BY1, and determine whether the discriminant factor BY1 exceeds the factor threshold value 1; if yes, issue the control instruction 2; if no, do not perform any operation;

[0100] Extract the discriminant factor BY2, and determine whether the discriminant factor BY2 exceeds the factor threshold value 2; if yes, issue the control instruction 2; if no, do nothing; wherein the mean of the factor threshold value 1 and the factor threshold value 2 is obtained by manual setting.

[0101] In the present application, braking the vehicle based on control instruction 1 and control instruction 2 includes:

[0102] D1: Get the current vehicle speed and determine whether the vehicle speed exceeds the low-speed judgment threshold; if yes, do not perform any operation; if no, jump to D2; the low-speed judgment threshold is set manually;

[0103] D2: When control instruction 1 or control instruction 2 is received, the current vehicle is decelerated according to a preset speed until the vehicle stops; wherein the preset speed is obtained according to the current vehicle speed.

[0104] It should be noted that the low-speed judgment threshold is a judgment threshold used to analyze whether the current vehicle speed will cause danger. This low-speed judgment threshold is obtained based on manual settings and can be set to 0 or a very small speed. If the current vehicle speed does not exceed the low-speed judgment threshold, it proves that the current vehicle is stationary or traveling at a very small speed. At this time, the driver can easily stop the vehicle. Therefore, there is no need to perform system braking at this stage, reducing the system's computational complexity. If the current vehicle speed exceeds the low-speed judgment threshold, it is necessary to add system braking on the basis of the driver's braking. This can provide a guarantee for the driver and the user, and increase the safety of the user's life and property.

[0105] It should be noted that the preset speed is obtained based on the current vehicle speed. The specific description is as follows: The preset speed is the deceleration per second corresponding to the manually set different vehicle speeds:

[0106] The vehicle can be decelerated according to the set ratio of the vehicle speed. For example, when the vehicle speed is 20 km / h, the set ratio is 50%, and the proportional deceleration is 50% of the current speed; when the vehicle speed is 30 km / h, the set ratio is 30%, and the proportional deceleration is 30% of the current speed.

[0107] Different decelerations per second can also be set for different vehicle speeds, for example: when the vehicle speed is 20 km / h, the deceleration per second is 10 km / h; when the vehicle speed is 30 km / h, the deceleration per second is 9 km / h.

[0108] A second aspect of the present invention provides a method for emergency braking of an intelligent driving vehicle by voice recognition, comprising the following steps:

[0109] S1: Acquire the sound data of the vehicle when it is driving;

[0110] S2: Dynamically plan the time periods of each day, analyze the historical in-car volume to obtain the characteristic volume of each time period inside the vehicle, obtain the discriminant factor 1 based on the characteristic volume, the in-car volume and the voice tone of the personnel, and analyze the sound data outside the vehicle to obtain the discriminant factor 2;

[0111] S3: Obtain control instruction one according to the voice and text of the person, and obtain control instruction two according to discrimination factor one and discrimination factor two; brake the vehicle based on control instruction one and control instruction two.

[0112] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0113] Working principle of the present invention:

[0114] Acquire the sound data of the vehicle when it is driving; dynamically plan the time periods of each day. This step can adaptively optimize the time periods of each day according to the user's personal data, so that the time period division of this system has a dynamic correction process, which increases the scientific nature of time division; analyze the historical volume inside the vehicle to obtain the characteristic volume of each time period inside the vehicle. This step avoids the situation where the system sets fixed thresholds for different users and different time periods, which easily leads to erroneous braking, and improves the stability of the system and its self-adaptability to different users; obtain discrimination factor one based on the characteristic volume, the volume inside the vehicle and the tone of the person, and obtain discrimination factor two by analyzing the sound data outside the vehicle; obtain control instruction one according to the person's voice and text, and obtain control instruction two according to discrimination factor one and discrimination factor two; brake the vehicle based on control instruction one and control instruction two.

[0115] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A voice recognition emergency braking system for an intelligent driving vehicle, characterized in that: include: A speech recognition module, and a data collection module, a vehicle control module and a database connected thereto; The data collection module is used to obtain the sound data of the vehicle when it is driving; the sound data includes the sound data inside the vehicle and the sound data outside the vehicle; the sound data inside the vehicle includes the volume inside the vehicle, the tone of the personnel and the voice text of the personnel; the sound data outside the vehicle includes the volume of pedestrians, the frequency of honking of other vehicles and the maximum duration of a single honking; The speech recognition module is used to dynamically plan the time periods of each day, analyze the historical in-vehicle volume to obtain the characteristic volume of each time period inside the vehicle, obtain the discrimination factor 1 based on the characteristic volume, the in-vehicle volume and the tone of the personnel, and analyze the sound data outside the vehicle to obtain the discrimination factor 2; The vehicle control module is used to obtain control instruction one according to the voice and text of the personnel, and to obtain control instruction two according to discrimination factor one and discrimination factor two; and to brake the vehicle based on control instruction one and control instruction two.

2. The intelligent driving vehicle voice recognition emergency braking system according to claim 1, characterized in that: The step of obtaining the sound data of the vehicle while it is traveling comprises: Based on user authorization, the real-time volume inside the car is obtained through the volume sensor, and the real-time volume is marked as the volume inside the car; the tone and voice of the person are obtained through the on-board microphone, and the voice of the person is converted into text using natural language recognition technology, and the text is marked as the person's voice text; the volume of pedestrians outside the car, the frequency of honking of other vehicles, and the maximum duration of a single honking are obtained through a number of intelligent acoustic sensors.

3. The intelligent driving vehicle voice recognition emergency braking system according to claim 1, characterized in that: The dynamic planning of the time periods of each day includes: A1: At a specific time on the day, obtain the position of the next day in the week, mark several days in the historical data with the same position as the next day as reference days, mark the time points where the vehicle speed is not 0 in each reference day as target time points, integrate continuous target time points into target time periods; integrate continuous non-target time points into non-target time periods, and plan the non-target time periods into several time periods according to fixed durations; A2: Get the number of times each target time point exists in the history, mark the target time point corresponding to the number of times equal to the threshold as the boundary point, divide each target period into several sub-periods based on the boundary points, and determine whether the duration of each sub-period is less than the fixed duration. If yes, jump to A3; if no, based on the formula HS= ZS / GS The number of divisions HS is obtained, and the sub-period is evenly divided into several time periods based on the number of divisions HS; wherein GS is a fixed duration, ZS is the duration of the sub-period, is the rounding symbol; A3: Mark the sub-periods that are less than the fixed duration and greater than the fixed duration multiplied by the proportional coefficient BL as time periods; combine the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are continuous into several time periods from front to back in time, and the duration of the combined several time periods is not greater than the value of the fixed duration multiplied by the proportional coefficient BL, and mark the sub-periods that are less than the fixed duration multiplied by the proportional coefficient BL and are separate as time periods; where the value range of the proportional coefficient BL is (0,1).

4. The intelligent driving vehicle voice recognition emergency braking system according to claim 1, characterized in that: The analysis of the historical in-vehicle volume to obtain characteristic volume of the vehicle in each time period includes: Based on the time range of each time period, the historical in-car volume of the corresponding time period is extracted from the database, and the historical in-car volume is divided into several volume groups based on the time period; the variance of each volume group is extracted in turn, and it is judged whether the variance is less than the variance judgment threshold; if yes, the data of the volume group is retained; if no, the volume with the largest difference from the mode of the volume in the volume group is removed, and the variance is judged again until the variance of the volume group is less than the variance judgment threshold, and the remaining data of the volume group is retained; The maximum value DZ, average value PJ and mode ZS of the retained data in each volume group are extracted, and the characteristic volume TL of the corresponding time period is obtained based on the formula TL=α1×DZ+α2×PJ+α3×ZS; wherein α1, α2 and α3 are all proportional adjustment coefficients greater than 0, and the value ranges of α1, α2 and α3 are all [0,1], α1<α2<α3.

5. The intelligent driving vehicle voice recognition emergency braking system according to claim 4, characterized in that: The first discrimination factor is obtained based on the characteristic volume, the volume in the car and the tone of the personnel, including: B1: extract the characteristic volume TL of the time period of the current time and the in-car volume CL of the current time, and determine whether the in-car volume CL is less than the characteristic volume TL; if yes, do nothing; if no, jump to B2; B2: Extract the in-car volume CL and the voice tone RD of the person at the current time, and obtain the discrimination factor BY1 based on the formula BY1=δ×exp(β1×(CL-BCL) / BCL+β2×(RD-BRD) / BRD); where BCL is the standard in-car volume and BRD is the standard voice tone; δ is the amplitude adjustment factor of the exp() function, and the value range of δ is (0,2); β1 and β2 are both proportional adjustment coefficients greater than 0, and β1+β2=1.

6. The intelligent driving vehicle voice recognition emergency braking system according to claim 5, characterized in that: The second discriminant factor is obtained by analyzing the external sound data, including: C1: Extract the pedestrian volume RY at the current time, and determine whether the placement position of the intelligent acoustic sensor that first receives the pedestrian volume RY is located in front of or behind the vehicle; if yes, the pedestrian volume factor RZ is obtained based on the formula RZ=μ1×(ln((RY-BRY) / BRY+1)+1); if no, the pedestrian volume factor RZ is obtained based on the formula RZ=μ2×(ln((RY-BRY) / BRY+1)+1); where BRY is the standard pedestrian volume, μ1 and μ2 are the amplitude adjustment coefficients of the ln() function, and the value range of μ1 and μ2 are both (0,2), μ1>μ2; Extract the horn frequency DP and the maximum duration of a single horn DC of other vehicles at the current time and within n seconds before the current time, and obtain the second discrimination factor BY2 based on the formula BY2=RZ+ρ×exp(σ1×DP / BDP+σ2×DC / BDC); where ρ is the amplitude adjustment factor of the exp() function, and the value range of ρ is (0,2); σ1 and σ2 are both proportional adjustment coefficients greater than 0, and σ1+σ2=1, σ1>σ2; BDP is the standard horn frequency, and BDC is the standard maximum duration of a single horn.

7. The intelligent driving vehicle voice recognition emergency braking system according to claim 1, characterized in that: The step of obtaining a control instruction according to the voice and text of the personnel includes: The keywords set by the user according to his / her own idiomatic expressions are marked as keyword library 1, the keywords of the in-car voices during the car accidents in history are counted to obtain keyword library 2, and keyword library 1 and keyword library 2 are integrated into an emergency word library; Extract the personnel's voice and text, and determine whether the personnel's voice and text are keywords in the emergency vocabulary; if yes, issue an operation instruction 1; if no, do nothing.

8. The intelligent driving vehicle voice recognition emergency braking system according to claim 6, characterized in that: The step of obtaining control instruction 2 according to discriminant factor 1 and discriminant factor 2 includes: Extract the discriminant factor BY1, and determine whether the discriminant factor BY1 exceeds the factor threshold value 1; if yes, issue the control instruction 2; if no, do nothing; Extract the second discriminant factor BY2, and determine whether the second discriminant factor BY2 exceeds the factor threshold value 2; if yes, issue the second control instruction; if no, do nothing.

9. The intelligent driving vehicle voice recognition emergency braking system according to claim 1, characterized in that: The braking of the vehicle based on the control instruction 1 and the control instruction 2 includes: D1: Get the current vehicle speed and determine whether the vehicle speed exceeds the low speed determination threshold; if yes, do not perform any operation; if no, jump to D2; D2: When control instruction 1 or control instruction 2 is received, the current vehicle is decelerated according to a preset speed until the vehicle stops; wherein the preset speed is obtained according to the current vehicle speed.

10. A method for emergency braking of an intelligent driving vehicle by voice recognition, based on the operation of an intelligent driving vehicle voice recognition emergency braking system according to any one of claims 1 to 9, characterized in that: S1: Acquire the sound data of the vehicle when it is driving; S2: Dynamically plan the time periods of each day, analyze the historical in-car volume to obtain the characteristic volume of each time period inside the vehicle, obtain the discriminant factor 1 based on the characteristic volume, the in-car volume and the voice tone of the personnel, and analyze the sound data outside the vehicle to obtain the discriminant factor 2; S3: Obtain control instruction one according to the voice and text of the person, and obtain control instruction two according to discrimination factor one and discrimination factor two; brake the vehicle based on control instruction one and control instruction two.

Images