Method for predicting head injury of balance car driver in different postures in collision accident
Through a combination of simulation and real experiments, a real-time risk assessment framework based on the decision tree model was established, which solved the safety problems of electric balanced vehicle drivers in different postures, achieved accurate prediction and risk assessment of driver head injuries, and improved the accuracy of driver safety prediction.
Patent Information
- Application Number
- CN202510276490.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-10
- Publication Date
- 2025-06-27
AI Technical Summary
The safety situation of drivers of electric balance bikes is not optimistic. The head injury in the accident varies greatly due to different postures. It is difficult for the existing technology to effectively predict and evaluate the driver's risks in different postures.
Using a combination of simulation experiments and real experiments, through video data sorting, indoor and outdoor experiments, sensors and experimental platforms are used to obtain body posture data of electric balance bike drivers under different postures, and a real-time risk assessment framework based on decision tree models is established to predict the possibility of head injury in different postures.
Real-time risk assessment of electric balanced vehicle drivers in different postures is realized, and a framework for real-time risk assessment of vehicles in future autonomous driving scenarios is provided, new ideas for vehicle appearance design, and improved the accuracy of driver safety prediction.
Smart Images
Figure CN120217660A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of the safety of electric scooter drivers, and proposes a method for predicting the safety of electric scooter drivers, and particularly establishes a framework for real-time risk assessment of vehicles in future autonomous driving scenarios. Background Art
[0002] With the continuous development of the global economy, the automotive field, as one of the most important industries in modern technology, has been expanding in scale. This has made long-distance travel more convenient for people, but at the same time has brought problems such as the increasingly severe road safety situation and traffic congestion. Therefore, as a short-distance means of transportation, electric scooters have gradually come into people's view and have gradually started to be popularized. This has also made the safety issues of electric scooters more severe and requires in-depth research.
[0003] An electric scooter is a new type of short-distance transportation vehicle, and its safety issues have not been perfectly solved. In recent years, researchers have carried out a series of studies on the safety issues of electric scooter drivers. However, at present, the safety situation of electric scooter drivers in China is still not optimistic, and traffic accidents involving electric scooter drivers are still on the rise year by year. Through investigation, it is found that the physical injuries of electric scooter drivers during accidents are closely related to their posture characteristics, and the head injuries caused by electric scooter drivers in different postures are very different after accidents. The present invention uses a method combining simulation experiments and real experiments to specifically establish a real-time risk assessment framework for vulnerable road users in different postures, provides a new method for predicting and analyzing the physical injuries of electric scooter drivers in the future, and provides a new idea for the vehicle shape design considering the safety issues of electric scooter drivers. The present invention extracts past video data and conducts indoor and outdoor experiments. On the premise of ensuring the safety of volunteers, various sensors and experimental platforms are used to restore the body posture data of electric scooter drivers as realistically as possible, and then the head injury data of electric scooter drivers in different postures are obtained through simulation experiments, and then a real-time risk assessment framework based on a decision tree model is established. The present invention uses a method combining real experiments and simulation experiments to accurately grasp the classic posture data of electric scooter drivers. Summary of the Invention
[0004] In order to achieve the above object, the present invention provides a possible real-time risk assessment framework for vehicles in future autonomous driving scenarios, which is used to judge the influence of different postures of electric scooter drivers on their head injuries.
[0005] To solve the above problems and achieve the corresponding object, the present invention adopts a combination of multiple methods for research, mainly including the following steps:
[0006] S1: Preparation before experiment: Organize video data and summarize a series of classic postures that may occur when the driver of the electric unicycle travels at a constant speed, accelerates, or takes emergency evasive actions.
[0007] S2: Conduct a road experiment to extract the body posture characteristics of the driver of the electric unicycle in classic postures. Then, conduct an indoor experiment with volunteers, asking them to simulate the classic postures of the driver of the electric unicycle. Next, use a motion capture system or platform to capture their actions.
[0008] S3: Process the data extracted from the experiment and output the key joint angle parameters. Then, process the data to provide a basis for subsequent research.
[0009] S4: Establishment and verification of a multi-rigid body model. Use various software and platforms to establish a multi-rigid body model of the driver of the electric unicycle and the vehicle.
[0010] S5: Input the processed data from the experiment into the multi-rigid body model for posture reconstruction, and then conduct multiple simulation analyses.
[0011] S6: Based on the simulation results and data in S5, obtain the head kinematic response and injury degree of the driver of the electric unicycle in different postures.
[0012] S7: Establish a decision tree model for different postures based on the sorted data.
[0013] Specifically, in S1, when organizing video data, the postures of the driver of the electric unicycle collected should be typical and extensive to ensure the accuracy of the subsequent model establishment.
[0014] Specifically, in S2, when conducting the indoor experiment, various devices or platforms are needed to capture the posture characteristics of the volunteers. And compare the posture characteristics and data extracted from the road experiment and indoor experiment with those obtained from the video data organization of the postures. The two sets of posture data must be basically the same to ensure the universality of the extracted postures.
[0015] Specifically, in S3, when measuring the key parameters of the volunteers in each posture, multiple motion capture devices need to be installed at the corresponding parts of the volunteers' bodies to ensure the accuracy of the measurement.
[0016] Specifically, in S4: All the models used in the research are multi-rigid body models, mainly to explore the kinematic response of the head of the driver of the electric unicycle.
[0017] Specifically, S5 is as follows: When performing multi-rigid-body simulation, it is necessary to input the data obtained in the experiment into the dummy model. After performing multiple groups of simulation analyses, head injury indices and kinematic response data are output to explore the influence of posture and its related parameters on head injury.
[0018] Specifically, S6 is as follows: Based on the simulation results, the head kinematic responses and injuries in different postures are sorted out to obtain the relationship between the posture of the electric unicycle rider and their head injury.
[0019] Specifically, S7 is as follows: On the basis of S6 above, a decision tree model in different postures is established with the driving characteristic parameters of the electric unicycle as the independent variable and the degree of its head injury as the dependent variable.
[0020] Preferably, conducting real-person experiments is to better simulate the real situation and better ensure the accuracy of experimental data.
[0021] Preferably, the purpose of conducting indoor experiments is to ensure the safety of volunteers and obtain better experimental results.
[0022] Preferably, conducting real-vehicle experiment acquisition is to ensure the real reliability of the experiment. However, problems such as signal interference may occur during the experiment, resulting in abnormal experimental data. Therefore, it is necessary to carry out multiple groups of simulation experiment acquisitions on the premise of ensuring that the actual posture of the electric unicycle rider is consistent with the simulated posture to obtain more experimental data.
[0023] Preferably, conducting experiments based on volunteers with different physical signs parameters is to obtain more reasonable posture data, and then verify the universality of the obtained classic postures.
[0024] Preferably, in the three postures of uniform speed driving posture, variable speed posture, and emergency avoidance posture, the differences between the left and right legs and some angles are not considered because in these three postures, the volunteers can be considered to be symmetric along the human body axis and the relevant angles can be ignored.
[0025] Preferably, the ankle swing angle and knee swing angle are the most obvious in both single and different driving postures. This may be caused by the habit differences of volunteers and the goal of lowering their own center of gravity. The overall changes of the other four angles are relatively weak, which may be affected by the working mode and range of the electric unicycle rider.
[0026] Preferably, in the three driving postures of uniform speed driving posture, variable speed posture, and competitive posture, the volunteers can be considered to be symmetric along the human body axis and the relevant angles can be ignored; in the emergency avoidance driving posture, the asymmetry of the left and right sides of the human body can be clearly observed and the relevant angles are obviously not negligible.
[0027] Preferably, after building the model, it is necessary to compare the model with the simulation animation motion mode to verify the effectiveness of the model.
[0028] Preferably, when exploring the specific situation of head injuries, the research uses the head injury criterion HIC as the evaluation standard.
[0029] Preferably, under different working conditions, the maximum value of HIC may appear in different collision stages.
[0030] Preferably, during the human-vehicle collision process, the head of the electric scooter driver is mainly subjected to two impacts, namely head-vehicle and head-ground collisions. Therefore, the collision process is divided into a primary collision and a secondary collision. This study mainly focuses on the primary collision to analyze the influencing factors of head injuries of electric scooter drivers in different postures.
[0031] Preferably, the present invention provides a real-time risk assessment framework based on a highly realistic collision scenario that conforms to biomechanical characteristics. The method includes the following steps:
[0032] First, extract the posture parameters through road experiments and indoor experiments, and conduct analysis.
[0033] Secondly, select a typical vehicle-electric scooter driver collision accident scenario, and based on this, establish a multi-rigid body simplified model of the vehicle. Combine it with the model of the electric scooter driver to verify the multi-rigid body collision model of the vehicle-electric scooter driver.
[0034] Finally, based on the established multi-rigid body collision model, conduct multi-rigid body simulation analysis, and output the head kinematic response and head injury criterion to provide data support for subsequent research.
[0035] As a study considering the influence of the driver's posture on their head injuries, the present invention has the following
[0036] Beneficial effects:
[0037] Through tests by volunteers, use posture acquisition equipment combined with multi-rigid body simulation analysis software to conduct posture modeling, and establish a highly realistic collision scenario that conforms to biomechanical characteristics;
[0038] Consider the different driving postures of the driver, conduct a hierarchical prediction of the head injuries of electric scooter drivers in collision accidents, and refine the key injury-causing factors, providing a new idea for the vehicle shape design considering the safety of vulnerable road users;
[0039] Study the head injury mechanism of the driver in the collision scenario, conduct a quantitative study on the head injuries of electric scooter drivers considering different driving postures in collision accidents, and clarify the internal relationship between the posture parameters and head injuries.
[0040] In addition to the purposes, features and advantages described above, the present invention has other purposes, features and advantages. The present invention will be further described in detail below with reference to the accompanying drawings. Description of the Drawings
[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following briefly introduces the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.
[0042] Figure 1 It is a schematic flowchart of the overall research idea and method of this patent;
[0043] Figure 2 It is a schematic diagram of the process of collecting the typical postures of electric unicycle drivers;
[0044] Figure 3 It is a schematic diagram of the key limb angles of the typical postures of electric unicycle drivers;
[0045] In the figure, 1 - hip flexion angle; 2 - knee flexion angle; 3 - ankle flexion angle; 4 - hip abduction angle; 5 - knee swing angle; 6 - ankle swing angle. Detailed Embodiments
[0046] The following describes the specific embodiments of the present invention. Those skilled in the art can intuitively understand other advantages and effects of the present invention from the content described in the specification. The present invention can also be implemented or applied through other different specific embodiments. The details of this specification can also be modified and supplemented from different perspectives without violating the logic of the present invention. It should be noted that, where feasible, the features in the embodiments can be combined with each other.
[0047] Embodiment: The present invention provides a method for predicting head injuries of drivers in electric unicycles with different postures. The operation process is as Figure 1 shown, and the specific implementation steps are as follows:
[0048] Step 1: As Figure 2 shown, first collect the driving postures of electric unicycle drivers. Conduct on-vehicle experiments to extract the postures of electric unicycle drivers during actual driving and obtain parametric posture data. It is necessary to sort out the video data of the on-vehicle experiments and summarize several common driving postures of electric unicycle drivers, including four typical postures: uniform speed driving posture, variable speed posture, competitive posture, and emergency avoidance posture.
[0049] Step 2: Extract the body features of basic driving posture, such as Figure 3 As shown, the measured body posture characteristics include the posture parameters collected and output by this study, which are hip flexion angle, knee flexion angle, ankle flexion angle, hip spread angle, knee swing angle and ankle swing angle.
[0050] Conduct indoor volunteer experiments, with each person simulating four postures, record videos and keep good records of experimental data. The number of male and female volunteers recruited is relatively balanced, and the range of height, hip height, knee height, shoulder width and hip width should all be within the common and reasonable range. Volunteers with Chinese physical characteristics are recruited. By selecting volunteers with different physical parameters for experiments, more reasonable posture data can be obtained, thereby verifying the universality of classic driving postures.
[0051] To capture the driver's motion, the driver needs to perform the experiment according to the four postures in step one. During the experiment, multiple sensors can be installed on the volunteer to collect the angles of key parts using commercial posture acquisition equipment and software to obtain body angle data in different postures.
[0052] Step 3: Use the platform to process the data and output key joint angle parameters and experimental data. By observing the experimental data, it is found that the ankle swing angle and knee swing angle have the most obvious changes in single and different driving postures, and the other four angles have weaker overall changes.
[0053] Step 4: Establish and verify the multi-rigid body model of vehicle-electric scooter driver. All models in this study are multi-rigid body models, and their kinematic responses are mainly studied. The pedestrian model uses the pedestrian series model that comes with the multi-rigid body software, which has been widely used by researchers. The electric scooter model is a verified multi-rigid body model. Use the software to establish a multi-rigid body vehicle model, and then establish a vehicle-scooter-pedestrian collision model. After the multi-rigid body model is established, it is necessary to simulate the constructed model and obtain the model simulation animation. Then compare the simulation animation with the collected video. When the posture of the driver in the animation and the video is basically the same, the next step of the complete simulation experiment can be carried out.
[0054] Step 5: Conduct simulation experiments under different postures. Input the experimental data obtained in step 3 into three dummy models with different characteristics (05, 50, and 95 percentile dummies), reconstruct the postures respectively, and obtain multiple sets of simulation analysis data. Finally, output the head injury index HIC and kinematic response data to explore the influence of posture and its related parameters on head injury.
[0055] Step 6: After conducting the simulation experiment, proceed with data processing of the simulation experiment. Based on the experimental results, obtain the relationship between the head kinematic response and the injury level of the electric unicycle rider at different postures and the key angular parameters of the rider's body posture, and establish a schematic diagram showing the variation of the head kinematic response and HIC over time. It is observed that the maximum head linear velocity appears in the 05th dummy in the variable-speed posture, and the maximum head angular velocity appears in the 50th dummy in the variable-speed posture. Under different driving postures, the differences in the head kinematic response of the dummy are closely related to its motion mode and the vehicle-human collision position.
[0056] Step 7: Establish the HIC 36 decision tree model for different postures according to the motion mode of the dummy, the head motion response of the dummy, and the head injury condition. Divide the levels of HIC 36 into primary injury, secondary injury, and tertiary injury, and then perform recursive processing on the data. Use the dummy type and the key angles of its limbs as independent variables, and HIC 36 as the dependent variable to establish a decision tree model based on different driving postures, and finally use the confusion matrix for model testing.
[0057] The present invention is a prediction method for the head injury of an electric unicycle rider in different postures based on a decision tree model. In the decision tree model, according to the body posture and its characteristic angles of the electric unicycle rider, it can be used as the first node, and the optimal parameters can be recursively selected downward according to the values and defined as the new root node. Keep looping until the stop condition is met and the leaf node is output, then the head injury of the electric unicycle rider can be predicted more accurately.
[0058] Finally, the accuracy of the decision tree models for the four driving postures obtained has reached an acceptable level. The different prediction accuracies of the models under different postures are caused by the differences in their input parameters, the randomness of the rider's motion in that posture, and the influence of height and body type on the rider's motion mode. Industry insiders can grasp the regularity of the rider's motion when conducting research on similar prediction models, which can effectively improve the prediction accuracy of the models.
[0059] In classic vulnerable road user collision accidents, the severity of head injuries is mainly affected by the speed of motor vehicles. However, no researchers have yet explored the impact of driving postures on the head injuries of scooter riders under collision conditions. In this patent, various different posture types are classified from video materials for driving postures, and real vehicle experiments are carried out to extract physical sign parameters and then multi-rigid body simulation analysis is carried out. The severity of head injuries is analyzed using the injury index HIC. It is found during the research process that in the vast majority of simulations, the maximum head HIC occurs in a primary collision. Therefore, this patent mainly focuses on the head injury situation in a primary collision.
[0060] The above embodiments are only illustrative of the principles and effects of the present invention and are not intended to limit the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the logic and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical ideas disclosed by the present invention should still be covered by the claims of the present invention.
Claims
1. A method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident, characterized in that: The following steps are involved: S1: Preparation before the experiment: Organize the video data and summarize a series of possible classic postures of the electric balance scooter driver, such as moving at a constant speed, accelerating, and avoiding danger in an emergency; S2: Conduct road experiments to extract the body characteristics of the classic posture of electric balance scooter drivers, and then conduct indoor experiments with volunteers to simulate the classic posture of electric balance scooter drivers, and then use a motion capture system or platform to capture their behavior; S3: Process the data extracted from the experiment and output key joint angle parameters, then process the data to provide a basis for subsequent research; S4: Establishment and verification of multi-rigid body model, establish multi-rigid body model of electric balance vehicle driver and vehicle through various software and platforms; S5: Input the processed data from the experiment into the multi-rigid body model to reconstruct the posture, and then perform multiple simulation analyses; S6: Based on the simulation results and data in S5, the head kinematic response and injury degree of the electric balance vehicle driver in different postures are obtained; S7: Establish decision tree models under different postures; Specifically, S1 is to make the collected postures of the electric scooter drivers typical and extensive when collating the video data, so as to ensure the accuracy of the subsequent model establishment; Specifically, S2 is that when conducting indoor experiments, it is necessary to use various devices or platforms to capture the posture features of volunteers; and compare the posture features and data extracted from the road experiment and the indoor test with the posture features and data obtained by video sorting. The two posture data must be basically consistent to ensure that the extracted posture is universal; Specifically, S3 includes installing multiple motion capture devices on corresponding parts of the volunteer's body to ensure the accuracy of the measurement when measuring the key parameters of each posture of the volunteer; The S4 is specifically as follows: the models used in the study are all multi-rigid body models, which are mainly used to explore the kinematic response of the electric balance vehicle driver; Specifically, S5 includes: when performing multi-rigid body simulation, the data obtained in the experiment needs to be input into the dummy model, and after performing multiple groups of simulation analysis, the head injury index and kinematic response data are output to explore the influence of posture and its related parameters on head injury; Specifically, S6 is to sort out the head kinematic responses and injuries under different postures according to the simulation results, and obtain the influence of the posture of the electric balance vehicle driver on his head injury; Specifically, S7 is to establish a decision tree model under different postures based on the above S6, taking the driving characteristic parameters of the electric balance vehicle as independent variables and the degree of head injury as dependent variable.
2. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: In the above step S1, the classic posture of the electric balance vehicle driver collected according to the video data is universal.
3. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: In the above step S2, conducting indoor experiments can ensure the safety of volunteers and obtain better experimental results. The experiments are divided into real vehicle experiments and simulation experiments. The real vehicle experiments are to better ensure the authenticity and reliability of the experiments. In addition, the experiments need to be based on volunteers with different physical parameters, in order to obtain more reasonable posture data, and then verify that the classic posture is universally applicable.
4. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: In the above step S3, when motion capturing the behavior of the volunteer, the motion capture device needs to take characteristic parameters of the body posture.
5. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: In the above step S4, after the multi-rigid body model is established, it is necessary to perform simulation verification on the model, and compare the simulation animation with the motion pattern in the video to verify the validity of the model.
6. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: In the above step S5, when a person collides with a vehicle, the collision of the electric balance vehicle driver is divided into a primary collision and a secondary collision. This patent mainly analyzes the primary collision to determine the degree of head injury of the electric balance vehicle driver in different postures.
7. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: In the above step S6, the data of the dummy's head kinematic response and injury degree in different postures are summarized and sorted, so as to obtain the specific relationship between the head injury and the posture parameters in different postures.
8. The method for predicting head injuries of a self-balancing vehicle driver in different postures in a collision accident according to claim 1, characterized in that: The above step S7 is specifically as follows: when establishing a decision tree model for predicting head injuries of electric balance scooter drivers, starting from the first node, recursively select the optimal parameters downwards, and then define it as a new node. The complete prediction model can be obtained by sequentially performing the above steps. After the model is established, its accuracy needs to be verified.