A non-driving behavior risk protection method and system based on driving posture recognition

Abstract

The present invention relates to a non-driving behavior risk protection method and system based on driving posture recognition, the method comprising: receiving driver video data, performing joint point recognition, and obtaining the driver's driving posture at each moment; comparing the driver's driving posture at each moment with the standard driving posture for similarity, calculating the degree of dislocation, and thereby performing a safety score for the degree of dislocation; determining the duration of the non-driving behavior according to a posture similarity threshold, and determining the corresponding dislocation time according to the time reserved for safety measures, and performing a safety score for the dislocation time by comparing it with the human body's reaction time; using a matrix scoring method to comprehensively consider the safety score results of the degree of dislocation and the safety score results of the corresponding dislocation time, determining the risk level score result, and thereby executing the corresponding safety protection strategy. Compared with the prior art, the present invention can greatly improve the experience and efficiency of human-computer interaction in intelligent driving vehicles, reduce false alarms, and improve vehicle safety.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to a method and system for preventing non-driving behavior risks based on driving posture recognition. Background Art

[0002] With the increasing popularity of intelligent vehicles, drivers are engaging in many non-driving activities while driving. Current distraction detection systems on the market suffer from frequent false positives, resulting in a poor user experience. Many consumers even disable these notifications, posing a significant safety risk. Furthermore, current products fail to tailor safety measures to specific non-driving behaviors, and limited research is conducted on the risk levels of these behaviors.

[0003] The main reason is that the definition and identification of risk levels requires real-time and accurate identification of the current driver's limb position, and quantitative research on driving posture. This involves real-time quantitative analysis of the degree of deviation and deviation time of human joints in driving posture, which is very difficult. Summary of the Invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and provide a non-driving behavior risk protection method and system based on driving posture recognition, so as to realize the non-driving behavior risk level identification and safety protection strategy generation based on driving posture recognition.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A non-driving behavior risk protection method based on driving posture recognition includes the following steps:

[0007] receiving driver video data;

[0008] Performing joint recognition on the driver video data through a vision-based human posture recognition system to obtain the driver's driving posture at each moment;

[0009] Compare the similarity between the driver's driving posture at each moment and the preset standard driving posture, calculate the degree of deviation of the driving posture at each moment, and assign a safety score to the degree of deviation of non-driving behavior based on the degree of deviation;

[0010] The similarity comparison result is compared with the preset posture similarity threshold to determine the duration of the non-driving behavior. Based on the preset safety measures, the corresponding time to leave the position is determined. The time to leave the position is compared with the human body's reaction time to provide a safety score for the time to leave the position for non-driving behavior.

[0011] A matrix scoring method is used to comprehensively consider the safety scoring results of the degree of departure and the corresponding safety scoring results of the departure time to determine the risk level scoring results of non-driving behavior, so as to implement the corresponding safety protection strategy.

[0012] Furthermore, the degree of displacement is calculated based on the head deflection ratio and joint deviation ratio in the similarity comparison result, and the corresponding calculation expression is:

[0013] Deviation= +

[0014] Where Deviation is the overall driving posture deviation, is the weight coefficient for calculating the head deflection ratio to the overall driving posture deviation, is the head deflection ratio, is the calculation weight coefficient of the proportion of each joint deviation to the overall driving posture deviation, is the joint deviation ratio.

[0015] Furthermore, the scoring process of the overall driving posture out-of-position degree is specifically as follows:

[0016] Pre-divide the range of separation corresponding to different safety levels and set the corresponding score for each range;

[0017] According to the interval to which the calculated degree of deviation belongs, the corresponding score value is obtained.

[0018] Furthermore, the calculation process of the off-position time is specifically as follows:

[0019] The similarity comparison result is compared with the preset posture similarity threshold. If the posture similarity threshold is exceeded, the timer starts to determine the duration of the non-driving behavior. The corresponding time to leave the position is calculated based on the current driving scenario and the time reserved for safety measures. The corresponding calculation expression is:

[0020] T=

[0021] Where T is the time of leaving the position, is the duration of non-driving behavior, is the scene coefficient, Allow time for safety measures.

[0022] Furthermore, the scoring process of the off-site time is specifically as follows:

[0023] Pre-divide the departure time intervals corresponding to different safety levels and set the corresponding scores for each interval;

[0024] According to the off-location time interval to which the calculated off-location time belongs, a corresponding score value is obtained.

[0025] Furthermore, the process of determining the non-driving behavior risk level scoring result by the matrix scoring method is specifically as follows:

[0026] The safety score result of the degree of departure and the safety score result of the departure time are multiplied together to obtain the non-driving behavior risk level score result.

[0027] Furthermore, the execution judgment process of the security protection strategy includes:

[0028] Pre-dividing the non-driving behavior risk level scoring results into intervals corresponding to different risk levels, wherein the risk levels include high risk, relatively high risk, medium risk and low risk;

[0029] If the risk is high, the corresponding response measure is vehicle takeover;

[0030] If the risk is high, the corresponding response is to request vehicle takeover;

[0031] If the risk is medium, the corresponding response measure is to warn of the risk;

[0032] If the risk is low, the corresponding response measure is no prompt.

[0033] Furthermore, the risk prompting method includes voice prompting through a speaker and vibration prompting through a seat drive structure.

[0034] The present invention also provides a system for implementing the above-mentioned non-driving behavior risk protection method based on driving posture recognition, comprising:

[0035] A camera module for acquiring driver video data;

[0036] A computing module, configured to execute the above-mentioned non-driving behavior risk protection method based on driving posture recognition;

[0037] The execution mechanism is used to perform actions according to the security protection strategy output by the calculation module.

[0038] Furthermore, the system further comprises:

[0039] The display module is used to display the data processing results of the calculation module.

[0040] Compared with the prior art, the present invention has the following advantages:

[0041] (1) The present invention first determines the driving posture by identifying the joint points of the driver's video data, and compares the similarity with the standard driving posture to calculate the degree of dislocation. On the one hand, the safety score of the degree of dislocation of non-driving behavior is scored based on the degree of dislocation, and on the other hand, the safety score of the time of dislocation of non-driving behavior is scored based on the time of dislocation. Finally, a matrix scoring method is used to comprehensively consider the safety score results of the degree of dislocation and the safety score results of the corresponding time of dislocation to determine the risk level score result of the non-driving behavior, so as to implement the corresponding safety protection strategy.

[0042] The present invention defines the risk level of non-driving behavior, provides a feasible solution for hierarchical safety protection based on risk level, and can provide personalized safety protection strategies. In actual vehicle applications, it can greatly enhance the driver's human-computer interaction experience, reduce false alarms, improve safety protection effects, enable the driver to be better protected during a collision, and greatly reduce the degree of personal injury. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A flowchart of a non-driving behavior risk protection method based on driving posture recognition provided in an embodiment of the present invention;

[0044] Figure 2 This is a schematic structural diagram of a non-driving behavior risk protection system based on driving posture recognition provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0046] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0047] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0048] Example 1

[0049] like Figure 1As shown, this embodiment provides a non-driving behavior risk protection method based on driving posture recognition, including the following steps:

[0050] S1: receiving driver video data;

[0051] S2: Use a vision-based human posture recognition system to identify joint points in the driver's video data and obtain the driver's driving posture at each moment;

[0052] S3: Compare the similarity between the driver's driving posture at each moment and the preset standard driving posture, calculate the degree of deviation of the driving posture at each moment, and assign a safety score to the degree of deviation of non-driving behavior based on the degree of deviation;

[0053] S4: Determine the duration of the non-driving behavior based on the similarity comparison result and the preset posture similarity threshold. Determine the corresponding time to leave the position based on the preset safety measures. Compare the time to leave the position with the human body's reaction time to provide a safety score for the time to leave the position for the non-driving behavior.

[0054] S5: A matrix scoring method is used to comprehensively consider the safety scoring results of the degree of departure and the corresponding safety scoring results of the departure time to determine the non-driving behavior risk level scoring results, so as to implement the corresponding safety protection strategy.

[0055] In other words, this solution takes into account the two dimensions of driving posture, namely, the degree of dislocation and the time of dislocation, and uses the product of the fitting scores of the two dimensions (ranging from 0 to 5 points) as the risk level score, ranging from 0 to 25 points.

[0056] First, the current driver's driving posture is obtained through a vision-based human posture recognition system. Then, the similarity comparison algorithm between the current driving posture and the standard driving posture is used to calculate the deviation degree of the current driving posture. The deviation degree calculation formula is as follows:

[0057] Deviation= +

[0058] Where Deviation is the overall driving posture deviation, is the weight coefficient for calculating the head deflection ratio to the overall driving posture deviation, is the head deflection ratio, is the calculation weight coefficient of the proportion of each joint deviation to the overall driving posture deviation, is the joint deviation ratio.

[0059] The scoring process for the overall driving posture out-of-position degree is as follows:

[0060] Pre-divide the range of separation corresponding to different safety levels and set the corresponding score for each range;

[0061] According to the interval to which the calculated degree of deviation belongs, the corresponding score value is obtained.

[0062] In this embodiment, the out-of-position score is fitted into a score range of 0 to 5, as shown in Table 1.

[0063] Table 1 Overall driving posture out-of-position scoring table

[0064]

[0065] By citing different human reaction times in the literature and reserving the AEB (automatic emergency braking system) reaction time, the timing starts from the time when it is determined that the non-driving behavior is engaged and ends when the behavior ends.

[0066] The calculation process of the off-position time is as follows:

[0067] The similarity comparison result is compared with the preset posture similarity threshold. If the posture similarity threshold is exceeded, the timer starts to determine the duration of the non-driving behavior. The corresponding time to leave the position is calculated based on the current driving scenario and the time reserved for safety measures. The corresponding calculation expression is:

[0068] T=

[0069] Where T is the effective off-position time, is the duration of non-driving behavior, is the scene coefficient, representing the weight coefficient of different scenes, Time is reserved for safety measures, such as AEB intervention time, and is temporarily set to a constant of 1.6s.

[0070] The specific scoring process for departure time is as follows:

[0071] Pre-divide the departure time intervals corresponding to different safety levels and set the corresponding scores for each interval;

[0072] According to the off-location time interval to which the calculated off-location time belongs, a corresponding score value is obtained.

[0073] In this embodiment, different off-position times are fitted into scores of 0 to 5, as shown in Table 2.

[0074] Table 2 Non-driving behavior away time score

[0075]

[0076] Disengagement time risk levels are categorized based on the maximum reaction times required by different parts of the human body. For example, it takes at least 0.3 seconds for the eyes to return to the front after straying, while the minimum reaction time for the human nervous system is 0.5 seconds. Complex tasks can lead to cognitive distractions, requiring a full reaction time of at least 3 seconds. Based on these parameters, the short disengagement time period of 0-0.3 seconds, which requires only visual response, is classified as low-risk. As the disengagement time increases, the threshold of 0.8 seconds (0.3 seconds plus 0.5 seconds), which requires both visual and neural responses, is used as the classification boundary, placing the 0.3-0.8 second range in the medium-risk range. Complex tasks, requiring cognitive analysis and longer reaction times, place the 0.8-3 second range in the high-risk range. When the disengagement time exceeds 3 seconds, in an emergency requiring the driver to take over, the human body requires at least 3 seconds or even longer to effectively take over the vehicle, making an accident highly likely to occur. Therefore, disengagement times exceeding 3 seconds carry an extremely high risk.

[0077] Using a matrix-based scoring system, we can determine the risk level of different driving postures at the current moment, and these scores will change dynamically over time. For example, for a behavior of the same risk level, its risk level will increase as its duration increases; whereas, a high-risk driving posture with a short duration will not receive a high risk score.

[0078] The matrix scoring method can be:

[0079] The safety score result of the degree of departure is multiplied by the safety score result of the time of departure to obtain the non-driving behavior risk level score result and the risk R.

[0080] In this embodiment, the matrix risk scoring table is shown in Table 3.

[0081] Table 3 Matrix risk scoring

[0082]

[0083] The execution and judgment process of the security protection strategy includes:

[0084] The risk level of non-driving behavior is pre-divided into intervals corresponding to different risk levels, including high risk, relatively high risk, medium risk and low risk;

[0085] If the risk is high, the corresponding response measure is vehicle takeover;

[0086] If the risk is high, the corresponding response is to request vehicle takeover;

[0087] If the risk is medium, the corresponding response measure is to warn of the risk;

[0088] If the risk is low, the corresponding response measure is no prompt.

[0089] In this embodiment, different targeted security protection strategies are defined as shown in Table 4.

[0090] Table 4 Strategies for dealing with risky driving postures

[0091]

[0092] Example 2

[0093] like Figure 2 As shown, this embodiment further provides a system for implementing a non-driving behavior risk protection method based on driving posture recognition as in embodiment 1, comprising:

[0094] A camera module for acquiring driver video data;

[0095] A computing module, configured to execute the non-driving behavior risk protection method based on driving posture recognition as described in Example 1 above;

[0096] The actuator is used to perform actions according to the safety protection strategy output by the computing module, including the horn, seat and steering wheel.

[0097] Preferably, the system further includes: a display module for displaying the data processing results of the calculation module.

[0098] After the system starts running, it first obtains driver video data through the camera, and then enters the computing module through data transmission. The deployed and trained matrix non-driving behavior risk assessment algorithm can derive the corresponding risk level score and corresponding response strategy. The corresponding posture and risk parameters will be displayed on the display module, and the actuator will perform corresponding execution actions according to the signals given by the response strategy.

[0099] The software interface layout of this system can be: the upper left corner is the image display area, the right area is the human posture related parameter display area, the lower left area is the high-risk driving posture screenshot and its parameters, as well as the response strategy based on the risk level, etc., and the bottom area is the operation tool menu bar.

[0100] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A non-driving behavior risk protection method based on driving posture recognition, characterized in that: The following steps are involved: receiving driver video data; Performing joint recognition on the driver video data through a vision-based human posture recognition system to obtain the driver's driving posture at each moment; Compare the similarity between the driver's driving posture at each moment and the preset standard driving posture, calculate the degree of deviation of the driving posture at each moment, and assign a safety score to the degree of deviation of non-driving behavior based on the degree of deviation; The similarity comparison result is compared with the preset posture similarity threshold to determine the duration of the non-driving behavior. Based on the preset safety measures, the corresponding time to leave the position is determined. The time to leave the position is compared with the human body's reaction time to provide a safety score for the time to leave the position for non-driving behavior. A matrix scoring method is used to comprehensively consider the safety scores of the degree of dislocation and the corresponding safety scores of the dislocation time to determine the risk level score of non-driving behavior, so as to implement the corresponding safety protection strategy; The degree of displacement is calculated based on the head deflection ratio and joint deviation ratio in the similarity comparison result, and the corresponding calculation expression is: Deviation= + Where Deviation is the overall driving posture deviation, is the weight coefficient for calculating the head deflection ratio to the overall driving posture deviation, is the head deflection ratio, is the calculation weight coefficient of the proportion of each joint deviation to the overall driving posture deviation, is the joint deviation ratio; The calculation process of the off-position time is specifically as follows: The similarity comparison result is compared with the preset posture similarity threshold. If the posture similarity threshold is exceeded, the timer starts to determine the duration of the non-driving behavior. The corresponding time to leave the position is calculated based on the current driving scenario and the time reserved for safety measures. The corresponding calculation expression is: T= Where T is the time of leaving the position, is the duration of non-driving behavior, is the scene coefficient, Allow time for safety measures.

2. The non-driving behavior risk protection method based on driving posture recognition according to claim 1 is characterized in that: The scoring process of the overall driving posture out-of-position degree is specifically as follows: Pre-divide the range of separation corresponding to different safety levels and set the corresponding score for each range; According to the interval to which the calculated degree of deviation belongs, the corresponding score value is obtained.

3. The non-driving behavior risk protection method based on driving posture recognition according to claim 1 is characterized in that: The scoring process of the off-site time is specifically as follows: Pre-divide the departure time intervals corresponding to different safety levels and set the corresponding scores for each interval; According to the off-location time interval to which the calculated off-location time belongs, a corresponding score value is obtained.

4. The non-driving behavior risk protection method based on driving posture recognition according to claim 1 is characterized in that: The process of determining the non-driving behavior risk level scoring result by the matrix scoring method is specifically as follows: The safety score result of the degree of departure and the safety score result of the departure time are multiplied together to obtain the non-driving behavior risk level score result.

5. The non-driving behavior risk protection method based on driving posture recognition according to claim 1 is characterized in that: The execution and judgment process of the security protection strategy includes: Pre-dividing the non-driving behavior risk level scoring results into intervals corresponding to different risk levels, wherein the risk levels include high risk, relatively high risk, medium risk and low risk; If the risk is high, the corresponding response measure is vehicle takeover; If the risk is high, the corresponding response is to request vehicle takeover; If the risk is medium, the corresponding response measure is to prompt the risk; If the risk is low, the corresponding response measure is no prompt.

6. The non-driving behavior risk protection method based on driving posture recognition according to claim 5 is characterized in that: The risk warning methods include voice warnings through speakers and vibration warnings through seat drive structures.

7. A system for implementing a non-driving behavior risk protection method based on driving posture recognition as described in any one of claims 1 to 6, characterized in that: include: A camera module for acquiring driver video data; A computing module, configured to execute a non-driving behavior risk protection method based on driving posture recognition as described in any one of claims 1 to 6; The execution mechanism is used to perform actions according to the security protection strategy output by the calculation module.

8. The system according to claim 7, characterized in that The system further comprises: The display module is used to display the data processing results of the calculation module.

Images