Post-accident injury assessment method of human body model introducing driver posture characteristics

Abstract

The invention mainly relates to a post-accident injury assessment method for introducing a driver posture feature human body model, and belongs to the field of intelligent automobile safety. The method comprises the steps that a driver measures and fills in posture parameters by himself, and posture characteristic data of the driver are obtained; based on the obtained driver posture characteristic data, constructing a driver personalized human body model in human body model software; driver characteristic joint points and data and vehicle kinematics parameters are obtained through a vehicle-mounted sensor; through information of the key nodes, human body model software is imported to correct the posture of the driver; and importing the data into finite element analysis, and predicting a detailed injury result of the driver after collision through vehicle kinematics parameters based on a personalized human body model. By adopting the method, more accurate and detailed after-accident driver injury information can be output, and more accurate driver posture data and injury data support can be provided for a vehicle safety protection system.

Description

Technical Field

[0001] The invention mainly relates to a post-accident damage assessment method which introduces a human body model of driver posture characteristics, and belongs to the field of intelligent automobile safety. Background Art

[0002] Traditional posture recognition technology relies too much on deep learning methods or feature extraction methods. The entire process is a black-box data-driven process. It fails to consider multiple influencing factors such as the driver's posture and driving status, which affect the driver's driving posture. At the same time, a large amount of reliable prior data is wasted, such as the driver's driving style and posture data.

[0003] Traditional vehicle safety research usually only considers the forms of human injury under standard signs. However, human signs are diverse, and traditional safety research cannot characterize the impact of diverse human signs on injury outcomes.

[0004] Traditional damage prediction results are relatively simple, usually only giving whether damage occurs, that is, binary damage prediction results, which are difficult to meet the data requirements for post-accident rescue, damage assessment, driving strategies, and adaptive restraint system optimization.

[0005] In order to adapt to the development of intelligent vehicles, it is necessary to address the shortcomings of current post-accident damage assessment. Summary of the invention

[0006] In view of this, the present invention provides a post-accident injury assessment method that introduces a human body model of driver posture characteristics, comprehensively considers the driver's posture physical signs, driving style, etc., and finally outputs more accurate and detailed post-accident driver injury information, providing more accurate driver posture data and injury data support for the vehicle safety protection system. In order to better illustrate the content of the invention, the human body model software takes OpenSim as an example, and the finite element analysis takes MADYMO as an example. The actual invention includes but is not limited to these two software.

[0007] In a first aspect, the present invention provides a method for calibrating joint information taking into account driver characteristics, wherein the system characterization equation of the occupant's posture over time t is as follows:

[0008]

[0009] Among them, J is the passenger joint, p (x，y，z) (t) is the position of the joint relative to the global coordinate system, ν (x，y，z) (t) and a (x，y，z) (t) represents the velocity and acceleration of the joint in the three coordinate axes respectively.

[0010] The camera identifies and captures the driver's characteristic joints, mainly obtaining the driver's torso joint angle information C1 (x, y, z), driver's left upper limb information U 1-l (x, y, z), driver's right upper limb information U 1-r (x, y, z) and head joint information H 1 (x,y,z).

[0011] After obtaining and calibrating the driver's characteristic joint information, it is necessary to roughly calibrate the characteristic joint information based on the previously obtained driver's physical data. The calibration reference points are selected according to different body parts, where the driver's torso joints must include the three-dimensional spatial information of at least three joints: chest T1 and left and right shoulder joints. The head joints must include at least three joints in the left and right eyes, nose and mouth positions. The upper limb joints include two joints: left and right shoulder joints.

[0012] Further explanation, taking the left upper limb as an example, based on the collected physical sign data of the driver's left upper limb, the shoulder joint S l To elbow joint E l The distance L SE_l , elbow joint E l To the center of the palm P l The distance L EP_l . Take the shoulder joint S l To calibrate the reference point, calculate the L collected by the image information SE_l ' space length, compared with L SE_l With L SE_l ′. If L SE_l <L SE_l ′, then E l ′ The coordinate information is shrunk along the direction of the big arm to the same position as S l Distance L SE_l If L SE_l >L SE_l ′, then E l ′ Coordinate information is stretched along the direction of the big arm to the same direction as S l Distance L SE_l Place.

[0013] Then the elbow joint E l To calibrate the reference point, calculate the L collected by the image information EP_l ' space length, compared with L EP_l With L EP_l ′. If L EP_l <L EP_l ′, then P l ′ The coordinate information is shrunk along the forearm direction to the same position as E l Distance L EP_l If L EP_l >L EP_l ′, then P l′ Coordinate information is stretched along the forearm direction to the same direction as E l Distance L EP_l .

[0014] In the second aspect, the present invention provides a process for obtaining detailed injury results of a driver after an accident. A user-personalized human body model is constructed in OpenSim and MADYMO / finite element based on the acquired prior data. For example, the OpenSim model is redeveloped to separate the head and neck of the model, and to add pitch, yaw, rotation and other degrees of freedom; the arm length, chest circumference, waist circumference and other parameters of the multi-rigid body model are adjusted in MADYMO; a mapping of the feature points on the finite element model before and after the transformation is established, and then all the nodes of the model are transformed to new positions according to the established mapping by interpolation.

[0015] Using the acquired driver's posture data at the moment of collision, the three-dimensional coordinate space position and angle of the joint of the human body model are reconstructed in OpenSim through inverse dynamics.

[0016] Since OpenSim and MADYMO have differences in defining human body models, the joint output data units and corresponding coordinate systems need to be converted from OpenSim to MADYMO. The specific conversion formula is as follows:

[0017] MADYMO LumbarSpine_jnt(R1 R2 R3)←Opensim

[0018] R1 = (lumbar_rotation / 57.3)

[0019] R2=-(lumbar_extension / 57.3)

[0020] R3=-(lumbar_bending / 57.3)

[0021] MADYMO NeckPivot4_jnt(R1 R2 R3)←Opensim

[0022] R1=(head_bending / 57.3)

[0023] R2=-(head_extension / 57.3)

[0024] R3=(head_rotation / 57.3)

[0025] MADYMO ShoulderL_jnt(R1 R2 R3)←Opensim

[0026] R1=-(arm_flex_l / 57.3)

[0027] R2=-(arm_add_l / 57.3)

[0028] R3=(arm_rot_l / 57.3)

[0029] MADYMO ElbowL_jnt(R1 R2)←Opensim

[0030] R1=(pro_sup_l / 57.3)

[0031] R2=(elbow_flex_l / 57.3)

[0032] The collision waveform (X, Y, Z three-axis acceleration), collision position (offset) and collision angle generated by the car collision (first-order collision) are obtained based on the on-board sensors as the input of the occupant collision (higher-order collision) simulation.

[0033] Based on the simulation results obtained by MADYMO / finite element, the damage evaluation indicators (HIC, BrIC, N ij , CTI, etc.) or Abbreviated Injury Scale (AIS) is used as the driver's injury assessment budget under this collision condition.

[0034] Beneficial effects:

[0035] A calibration method for the joint change data of the driver's posture characteristics in the car is proposed to ensure the accuracy of the human joint information input and realize the accurate acquisition of the driver's posture change process.

[0036] A solution has been designed to obtain detailed injury assessment results for drivers after an accident. By utilizing the fusion mode of "human body model software pre-processing + finite element analysis post-output", the output injury prediction results can be further refined, thereby optimizing the data support for the driver's post-accident danger assessment strategy.

[0037] Combined with the driver's personalized body characteristics, different percentile models in the finite element model library are scaled to improve the coverage and accuracy of the prediction results.

[0038] The neck and head posture greatly affect the driver's injury degree and survival rate after the accident. In order to include the head and neck data in the injury prediction range, the neck degree of freedom was introduced based on the secondary development of the human body model.

[0039] In order to better illustrate the beneficial effects of the present invention, two practical usage scenarios are proposed below.

[0040] Serving the vehicle emergency call system eCall (active safety system optimization)

[0041] When a collision accident occurs, this patented system can calculate the severity of damage to the driver's important parts in the cloud based on the driver's posture information and collision conditions in the pre-collision stage and at the time of collision, and determine whether it is necessary to activate the vehicle emergency call system based on the prediction results (for example, minor injuries can only require notifying the traffic police for handling or vehicle rescue, while serious injuries require immediate medical rescue).

[0042] Serving adaptive restraint systems (passive safety system optimization)

[0043] The existing restraint system parameter configuration is mostly designed based on the 50th percentile standard human body model, lacking consideration and injury protection for non-standard people. Based on this patented system, the posture and injury of the occupant at the moment of collision can be predicted, and the occupant's injury risk can be assessed by combining the parameter allocation of different types of restraint systems. Based on this, the restraint system parameter configuration can be optimized in a targeted manner to achieve the adaptive design of the occupant restraint system. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The invention is an overall technical flow chart of post-accident damage assessment budget which introduces a driver characteristic human body model.

[0045] Figure 2 It is a schematic diagram of driver characteristic joint data calibration.

[0046] Figure 3 This is a schematic diagram of the characteristic joint data calibration taking the left upper limb as an example.

[0047] Figure 4 It is a technical flow chart serving the vehicle emergency call system eCall.

[0048] Figure 5 It is a technical flow chart serving to optimize the adaptive constraint system. DETAILED DESCRIPTION

[0049] The present invention is described in detail below with reference to the accompanying drawings and examples. To facilitate the description of the specific implementation of the present invention, the human body model software takes OpenSim as an example, and the finite element simulation software MADYMO is taken as an example. The embodiments of the present invention are only part of the embodiments of the present invention, not 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 belong to the protection scope of the present invention.

[0050] Basic data: Collect the driver's physical data, including but not limited to height, weight, BMI, seat height, arm length, etc.

[0051] Step 1: Build a personalized human model of the driver based on basic data:

[0052] Step 1.1: Based on the original human body model of OpenSim, secondary development introduces the neck degree of freedom.

[0053] Step 1.2: Based on the standard human body model in the MADYMO / finite element human body library, scale the standard model according to the basic data, mainly considering factors that have a significant impact on collision risk, such as height, seat height, arm length, and weight, to establish a personalized human body model that meets the driver's physical characteristics.

[0054] Step 2: Identify and obtain the posture data of the driver's characteristic joints and vehicle kinematics data during vehicle driving based on sensors:

[0055] Step 2.1: Use the in-vehicle sensor to obtain the posture data of the driver's characteristic joints during the vehicle's driving. Taking the camera as the information input source as an example, the driver's torso joint angle information C 1 (x, y, z), driver's left upper limb information U 1-l (x, y, z), driver's right upper limb information U 1-r (x, y, z) and head joint information H 1 (x,y,z).

[0056] Step 2.2: Use the on-board computer to obtain vehicle driving data information, such as vehicle speed, vehicle acceleration, driving heading angle, occupant restraint system usage status, etc.

[0057] Step 3: Use OpenSim to calibrate the obtained driver posture coordinate information:

[0058] Step 3.1: Rough calibration: Based on the human body sign data obtained from the basic data, the characteristic joint data identified by the sensor is roughly screened, that is, the spatial distance between two characteristic joints should be close to the actual physical sign measurement data.

[0059] Step 3.2: Detailed calibration. Select reference objects for different body parts for further calibration. For example, for the torso, use the left and right shoulder joints and the midpoint of the line connecting the two as calibration reference points, and calibrate the upper limbs and torso parts respectively; the recognition of facial features can be introduced to assist in the calibration of the head.

[0060] Step 4: Output detailed driver damage results:

[0061] Step 4.1: Based on the driver's posture in the pre-crash and collision stages and the collected vehicle kinematic information, the driver's injury status is calculated in the cloud through a personalized human body model, and the injury result output can be customized for different body parts.

Claims

1. A post-accident injury assessment method using a human body model with driver posture characteristics, characterized in that: A method for calibrating joint information taking into account the driver's posture characteristics is provided, and a process for obtaining detailed injury results of the driver after an accident is provided.

2. The method for calibrating joint information considering the driver's posture characteristics according to claim 1, characterized in that: The system characterization equation of the occupant posture over time t is as follows:

3. The system characterization equation of the occupant posture over time t according to claim 2, characterized in that: J is the passenger joint, p (x，y，z) (t) is the position of the joint relative to the global coordinate system, v (x，y，z) (t) and a (x，y，z) (t) represents the velocity and acceleration of the joint in the three coordinate axes respectively.

4. The method for calibrating joint information considering the driver's posture characteristics according to claim 1, characterized in that: The camera recognizes and captures the characteristic joints of the driver's posture, mainly obtaining the driver's torso joint angle information C1 (x, y, z) and the driver's left upper limb information U 1-l (x, y, z), driver's right upper limb information U 1-r (x, y, z) and driver's head joint information H1(x, y, z).

5. The recognition and capture of the characteristic joints of the driver by the camera according to claim 4 is characterized in that: The joints are roughly calibrated based on the driver's body characteristics obtained in the first step, and calibration reference points are selected for different body parts.

6. The method for selecting calibration reference points for different body parts according to claim 5, characterized in that: The driver's torso joints must include at least 3 joints: chest T1 and left and right shoulder joints; the driver's head joints must include at least 3 joints: left and right eyes, nose and mouth; the driver's upper limb joints include 2 joints: left and right shoulder joints.

7. The joint selection according to claim 6, characterized in that: Taking the left upper limb as an example, based on the collected physical sign data of the driver's left upper limb, the shoulder joint S l To elbow joint E l The distance L SE_l , elbow joint E l To the center of the palm P l The distance L EP_l . Take the shoulder joint S l To calibrate the reference point, calculate the L collected by the image information SE_l ' space length, compared with L SE_l With L SE_l ′. If L SE_l <L SE_l ′, then E l ′ The coordinate information is shrunk along the direction of the big arm to the same position as S l Distance L SE_l If L SE_l >L SE_l ′, then E l ′ Coordinate information is stretched along the direction of the big arm to the same direction as S l Distance L SE_l Then at the elbow joint E l To calibrate the reference point, calculate the L collected by the image information EP_l ' space length, compared with L EP_l With L EP_l ′. If L EP_l <L EP_l ′, then P l ′ The coordinate information is shrunk along the forearm direction to the same position as E l Distance L EP_l If L EP_l >L EP_l ′, then P l ′ Coordinate information is stretched along the forearm direction to the same direction as E l Distance L EP_l .

8. The process for obtaining detailed injury results of a driver after an accident according to claim 1 is characterized in that: Based on the acquired prior data, a user-personalized human body model is constructed in the human body model software and the finite element analysis software.

9. The method of constructing a user-personalized human body model in human body model software and finite element analysis software based on acquired prior data according to claim 8, characterized in that: The human body model was redeveloped to separate the head and neck of the model, and to add pitch, yaw, rotation and other degrees of freedom; the arm length, chest circumference, waist circumference and other parameters of the multi-rigid body model were adjusted in the finite element analysis software; The mapping of the feature points on the finite element model before and after the transformation is established, and all the nodes of the model are transformed to the new position according to the established mapping by interpolation. Using the acquired driver's posture data at the moment of collision, the three-dimensional coordinate space position and angle of the joint of the human body model are reconstructed in the human body model software through inverse dynamics.

10. According to claim 1, a method for calibrating joint information considering the posture characteristics of the driver and a process for obtaining detailed injury results of the driver after an accident are provided, characterized in that: The collision waveform (X, Y, Z three-axis acceleration), collision position (offset) and collision angle generated by the car collision (first-order collision) are obtained by using the on-board sensor as the input of the occupant collision (high-order collision) simulation. The simulation results are obtained by using the finite element analysis software, and the injury evaluation indicators (HIC, BrIC, N ij , CTI, etc.) or Abbreviated Injury Scale (AIS) is used as the driver injury assessment method under this collision condition.

Images