Pedestrian collision avoidance active safety control method and device suitable for multiple road conditions

Abstract

The invention belongs to the technical field of road traffic safety, and particularly relates to a pedestrian collision avoidance active safety control method and device suitable for multiple road conditions. Comprising the steps of 1, calculating normalized transverse and longitudinal tire forces through a vehicle dynamics model and a tire model, and obtaining an estimated value of a road adhesion coefficient by taking transverse and longitudinal acceleration and yaw velocity of a vehicle as observation vectors; 2, carrying out coordinate conversion based on a current vehicle coordinate system, and predicting a pedestrian motion track; step 3, determining a collision avoidance control opportunity adopted by pedestrian collision avoidance active safety control; 4, making a pedestrian collision avoidance active safety control decision; and 5, by establishing an inverse dynamic model of a vehicle braking system, according to the obtained expected braking deceleration, brake pressure calculation is carried out, so that braking control over the vehicle is achieved. The method is suitable for different road surface working conditions, and the practicability and robustness of the active safety control method for pedestrian protection are effectively improved.

Description

Technical Field

[0001] The invention belongs to the technical field of road traffic safety, and in particular to a pedestrian collision avoidance active safety control method and device suitable for multiple road conditions. Background Art

[0002] Due to the uncertainty of pedestrian movement trajectories, it is more difficult to predict pedestrian movement trajectories. Pedestrians are a vulnerable group participating in road traffic, and the casualty rate in road traffic accidents is extremely high. In order to reduce the occurrence of traffic accidents, reduce the casualty rate of traffic accidents, and improve automobile safety, pedestrian active safety collision avoidance system is an important research direction in the automotive industry.

[0003] Active safety systems at this stage still have great limitations. Most active safety systems only consider the situation where the road conditions are good. Usually, the road adhesion coefficient is set to a constant, which makes the active safety system only applicable to certain good roads, and cannot achieve good collision avoidance effects on medium and low adhesion roads. When emergency braking on low adhesion roads, the vehicle dynamic response is significantly different, and it is easy to cause side slip or tail swing due to wheel locking, affecting handling stability and braking performance. Therefore, accurate estimation of the road adhesion coefficient of the current road surface and development of corresponding control methods for the vehicle active safety system can better perform collision avoidance control.

[0004] In summary, pedestrian road safety protection has always been an important topic in automotive active safety technology. However, considering the different braking performance of vehicles under different road conditions, it is necessary to accurately estimate the adhesion coefficient of the current road surface, and conduct collision risk level assessment and corresponding control logic decision research based on the road adhesion coefficient. Summary of the invention

[0005] The present invention provides a pedestrian collision avoidance active safety control method and device that are adaptable to multiple road conditions, which are applicable to different road conditions, effectively improve the practicality and robustness of the active safety control method for pedestrian protection, and overcome the influence of the uncertainty of pedestrian movement and multiple road conditions on the active safety control of pedestrian collision avoidance.

[0006] The technical solution of the present invention is described as follows in conjunction with the accompanying drawings: In a first aspect, the present invention provides a pedestrian collision avoidance active safety control method adapted to multiple road conditions, comprising the following steps: Step 1: Calculate the normalized lateral and longitudinal tire forces through the vehicle dynamics model and the tire model, and then use the lateral and longitudinal accelerations and yaw rate of the vehicle as observation vectors to obtain an estimated value of the road adhesion coefficient; Step 2: Perform coordinate transformation based on the current vehicle coordinate system, assuming that the pedestrian moves in a straight line at a uniform speed, and predict the pedestrian's movement trajectory; Step 3: Determine the collision avoidance control timing adopted by the pedestrian collision avoidance active safety control; Step 4: Consider the vehicle dynamics, potential collision risk of pedestrians, and road traffic safety factors of pedestrians and drivers to make active safety control decisions for pedestrian collision avoidance; Step 5: By establishing an inverse dynamics model of the vehicle braking system, the braking pressure is calculated according to the desired braking deceleration to achieve braking control of the vehicle.

[0007] Furthermore, the specific method of step one is as follows: 11) Establish the vehicle dynamics equation as follows: , , , In the formula, is the total mass of the vehicle; is the vehicle longitudinal acceleration; is the longitudinal force of the left front wheel; is the longitudinal force of the right front wheel; is the lateral force of the left front wheel; is the lateral force of the right front wheel; is the front wheel turning angle; is the longitudinal force of the left rear wheel; is the longitudinal force of the right rear wheel; is the lateral force of the left rear wheel; is the lateral force of the right rear wheel; is the lateral acceleration of the vehicle; is the moment of inertia of the vehicle around the z-axis; is the yaw angular acceleration; is the distance from the center of mass to the front axle; is the distance from the center of mass to the rear axle; is the front wheel track; is the rear wheel track; 12) Establish a tire model as follows: Considering the axle load transfer during vehicle driving, the vertical load formula of each wheel of the vehicle is: , In the formula, is the vertical load on the left front wheel; is the vertical load on the right front wheel; is the vertical load on the left rear wheel; is the vertical load on the right rear wheel; is the distance from the front axle to the rear axle; is the vehicle center of mass height.

[0008] For single wheel force analysis, the formulas for tire longitudinal force and lateral force in the tire coordinate system are: , , In the formula, is the longitudinal force in the wheel coordinate system; is the lateral force in the wheel coordinate system; is the road adhesion coefficient; is the normalized tire longitudinal force of each tire; is the normalized tire lateral force of each tire; is the vertical load of the tire; is the tire longitudinal slip coefficient; is the tire cornering stiffness; is the tire slip rate; is the tire slip angle; is the boundary correction value; is the boundary correction function of the tire force.

[0009] 13) The extended Kalman filter is used to estimate the road adhesion coefficient of each wheel, and the vehicle dynamics system is modeled as follows: , , In the formula, and is the state variable; is the observed variable; is the input variable, is the nonlinear state transfer equation; is the nonlinear observation equation; is the process noise; To measure noise.

[0010] Among them, the state variable , input variable , observed variables ; The calculation formula of the observation equation is: , In the formula, is the road adhesion coefficient of the left front wheel; is the normalized longitudinal force of the left front wheel; is the normalized lateral force of the left front wheel; is the road adhesion coefficient of the right front wheel; is the normalized longitudinal force of the right front wheel; is the normalized lateral force of the right front wheel; is the road adhesion coefficient of the left rear wheel; is the normalized longitudinal force of the left rear wheel; is the normalized lateral force of the left rear wheel; is the road adhesion coefficient of the right rear wheel; is the normalized longitudinal force of the right rear wheel; is the normalized lateral force of the right rear wheel; The Jacobian matrix of the partial derivative of the state vector is solved for the observation equation, and the observation equation of the road adhesion coefficient estimation algorithm is obtained as follows: , In the formula, is the influence coefficient of the left front wheel on the yaw acceleration; is the influence coefficient of the right front wheel on the yaw acceleration; is the influence coefficient of the left rear wheel on the yaw angular acceleration; is the influence coefficient of the right rear wheel on the yaw acceleration; is the road adhesion coefficient of the left front wheel at time step k; is the road adhesion coefficient of the right front wheel at time step k; is the road adhesion coefficient of the left rear wheel at time step k; is the road adhesion coefficient of the right rear wheel at time step k; in: , , , .

[0011] Furthermore, in step 13), the specific method of the extended Kalman filter is as follows: 131) Initialize state variables and set state variables and the covariance matrix ; Define the noise covariance matrices Q and R to represent process noise and measurement noise respectively; 132) Time update: Generate the Sigma point of the state variable through unscented transformation as follows: , In the formula, is the Sigma point matrix; is the state mean estimate; is the state variable dimension; is the scaling parameter; is the state covariance matrix; Substitute each Sigma point into the nonlinear state transfer equation , as shown below: , In the formula, is the Sigma point after nonlinear transformation, that is, the state prediction value at time k+1; Calculate the predicted state mean based on the propagated Sigma point and the predicted covariance , as shown below: , , In the formula, is the state prediction value at time k+1; is the prediction covariance; is the weight of the mean; is the weight of the covariance; is the process noise covariance matrix; 133) Measurement update, through the observation equation , calculate the observed value of each Sigma point as follows: , In the formula, is the observed value corresponding to the Sigma point; is the input variable; Calculate the predicted observation mean as follows: , In the formula, is the mean of the predicted observations; is the mean weight; The observation covariance is calculated as follows: , In the formula, is the observation covariance matrix; is the measurement noise covariance matrix; The cross-covariance of states and observations is calculated as follows: , In the formula, is the cross-covariance matrix between states and observations; Update the state estimate using the Kalman gain Update the state mean and covariance as follows: , , , In the formula, is the Kalman gain matrix; is the inverse matrix of the observation covariance; is the updated state mean; is the actual observed value; is the updated state covariance matrix; is the prediction covariance matrix.

[0012] Furthermore, the specific method of step 2 is as follows: 21) During the vehicle motion, assuming that the vehicle speed and yaw rate remain unchanged, the vehicle motion model is a uniform angular velocity and uniform speed model, as shown below: , , In the formula, is the longitudinal displacement of the vehicle; is the lateral displacement of the vehicle; is the vehicle speed; is the vehicle yaw angular velocity; is the vehicle yaw angular velocity; is the time step; 22) The pedestrian and vehicle motions are decoupled into two motion modes: translation and rotation, as follows: The coordinate mapping of the pedestrian motion state vector of the vehicle translation motion is as follows: , In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current coordinate system; is the pedestrian's horizontal coordinate at the previous moment in the current coordinate system; is the pedestrian's longitudinal speed at the previous moment in the current coordinate system; is the pedestrian's lateral speed at the previous moment in the current coordinate system; is the pedestrian's longitudinal coordinate in the coordinate system of the previous moment; is the lateral coordinate of the pedestrian in the coordinate system of the previous moment; At the current time in the vehicle coordinate system, the state vector mapping result of the pedestrian at the previous moment is : In the formula, is the motion state vector of the pedestrian; is the longitudinal coordinate of the pedestrian in the current vehicle coordinate system; is the lateral coordinate of the pedestrian; is the pedestrian longitudinal speed; is the pedestrian lateral speed; The mapping coordinates of the pedestrian motion state vector after the vehicle rotation are updated as follows: , In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current coordinate system; is the horizontal coordinate of the pedestrian at the previous moment; The pedestrian motion state vector at the previous moment in the current vehicle coordinate system is: , In the formula, is the longitudinal coordinate of the pedestrian at the previous moment in the current vehicle coordinate system; is the lateral coordinate of the pedestrian at the previous moment in the current vehicle coordinate system; is the pedestrian's longitudinal speed at the previous moment in the current vehicle coordinate system; is the pedestrian's lateral speed at the previous moment in the current vehicle coordinate system; Since the motion state of pedestrians changes slowly relative to the motion state of vehicles, it is assumed that pedestrians move at a uniform speed, that is, the pedestrian motion model is a uniform motion model. Therefore, the motion state of pedestrians at the current moment is: ; 23) According to Taylor's formula, the trigonometric function is approximated as: , The calculation formula for the pedestrian motion state vector at the current moment is: , 24) Use the recursive least squares identification method to predict the pedestrian's motion state, that is, the pedestrian's motion trajectory: , in, The matrices are: , , In the formula, is the horizontal and vertical coordinates of the pedestrian, that is ; is the pedestrian motion state vector at the previous moment, that is .

[0013] Furthermore, in step 24), the specific method of using the recursive least squares identification method to predict the pedestrian motion state is as follows: 241) Calculate the error between the actual value of the pedestrian's motion state and the predicted value at the previous moment : , 242) According to the error value , calculate the gain matrix , and introduce the forgetting factor , is the covariance matrix obtained in the previous calculation; therefore, the gain matrix calculation The calculation formula is as follows: , In the formula, is the gain matrix; is the covariance matrix of the previous moment; is the state transition matrix; For the forgetting factor; is the identity matrix; 243) According to the gain matrix , and the covariance matrix Update to prepare for the next trajectory prediction algorithm calculation; among them, the covariance matrix The calculation formula is: , In the formula, is the updated covariance matrix; 244) According to the gain matrix , update the pedestrian state vector prediction value: , Furthermore, the specific method of step three is as follows: 31) Establish a safe distance model; The emergency braking process is divided into four stages, namely, the driver's reaction to danger, braking reaction, continuous braking, and releasing the brake; among them, the critical braking distance is expressed as: , in, is the current speed; is the maximum braking deceleration; for driver reaction time; is the braking reaction time; The influencing factors of road adhesion coefficient introduced into the safety distance model , as shown below: , In the formula, is the current road adhesion coefficient; is the maximum adhesion coefficient; is the minimum adhesion coefficient; The safety distance model considering the road adhesion coefficient is as follows: , In the formula, is the minimum safe distance; 32) Establish a collision risk assessment model; The collision risk is assessed based on the safety distance and collision time threshold. The collision time threshold is dynamically calculated based on the motion state of the vehicle and pedestrian and the road adhesion coefficient to quantify the collision risk. The danger zone is defined as the overlapping part of the vehicle trajectory and the pedestrian trajectory. The time it takes for a vehicle to pass through the danger zone is expressed as: , , In the formula, The time when the vehicle enters the danger zone; The time it takes for the vehicle to leave the danger zone; is the longitudinal displacement of the vehicle at the current moment; The lateral safety distance for pedestrians; is the longitudinal speed of the vehicle relative to the pedestrian at the current moment; is the vehicle length; The time it takes for a pedestrian to pass through the danger zone is expressed as: , , In the formula, The starting time for pedestrians to leave the danger zone; Time for pedestrians to leave the danger zone; is the lateral displacement of the vehicle at the current moment; is the lateral safety distance of the vehicle; is the vehicle width; is the lateral speed of the pedestrian relative to the vehicle at the current moment; 33) By comparing the start and end points of the time period when vehicles and pedestrians pass through the dangerous area, the collision risk level at the current moment is determined and the collision time is obtained. ; 34) According to the collision time Determine the risk level and take appropriate measures based on the risk level; For pedestrian crossings, the risk level is divided into a warning zone and an automatic braking zone; among them, the warning zone reminds the driver to pay attention to the danger, and the automatic braking zone adopts automatic emergency braking control to avoid collision.

[0014] The TTC thresholds for the warning area and automatic braking area are defined as follows: , , In the formula, is the warning area threshold; is the automatic braking area threshold; is the TTC threshold coefficient.

[0015] Furthermore, the specific method of step 4 is as follows: 41) Get TTC error value , as follows: , In the formula, is the TTC critical value; 42) According to different situations and road adhesion conditions, the collision time error and road adhesion coefficient are selected as the system input, and the expected deceleration is selected as the system output; the Gaussian function is selected as the membership function, and the fuzzy reasoning model is used for decision making.

[0016] Furthermore, the specific method of step five is as follows: 51) In the feedforward stage, the inverse dynamics model of the braking system is established as follows: , In the formula, is the braking deceleration; For braking force; is the air resistance coefficient; is the frontal area of ​​the vehicle; is the air density; is the rolling resistance coefficient; 52) Calculate the brake pressure as follows: , In the formula, is the proportional coefficient of braking force and braking force pressure; 53) In the feedback stage, the PID algorithm is used for feedback control, as follows: , , In the formula, For target pressure; is the deceleration error of the low-level controller; is the target deceleration; is the actual deceleration; is the proportionality coefficient; is the integration coefficient, is the differential coefficient.

[0017] In a second aspect, the present invention further provides a pedestrian collision avoidance active safety control device adapted to multiple road conditions, which is used to implement a pedestrian collision avoidance active safety control method adapted to multiple road conditions, comprising: The road adhesion coefficient estimation module 110 is used to calculate the normalized lateral and longitudinal tire forces through the vehicle dynamics model and the tire model, and then use the lateral and longitudinal accelerations and yaw rate of the vehicle as observation vectors to obtain an estimated value of the road adhesion coefficient; A pedestrian trajectory prediction module 120 is used to perform coordinate transformation based on the current vehicle coordinate system, assuming that the pedestrian moves in a straight line at a uniform speed, and predict the pedestrian's movement trajectory; A collision risk assessment module 130, used to determine the collision avoidance control timing adopted by the pedestrian collision avoidance active safety control; The active control upper layer module 140 is used to make active safety control decisions for pedestrian collision avoidance by considering vehicle dynamics characteristics, potential collision risks of pedestrians, and road traffic safety factors of pedestrians and drivers; The active control lower layer module 150 is used to establish an inverse dynamics model of the vehicle braking system and calculate the braking pressure according to the obtained expected braking deceleration to achieve braking control of the vehicle.

[0018] The beneficial effects of the present invention are: 1) The present invention improves the adaptability of active braking control strategy under different road conditions, especially the braking effect under low-adhesion road conditions, by accurately estimating the road adhesion coefficient, thereby improving driving safety; 2) The pedestrian trajectory prediction algorithm in the present invention, combined with the recursive least squares method, reduces the impact of historical data on the algorithm, making the pedestrian trajectory prediction more accurate and better matching the actual intention of pedestrians to cross the street; 3) The present invention introduces fuzzy control theory to make braking decisions, taking the road adhesion coefficient and collision time error as input, thus improving the flexibility and robustness of the control algorithm under complex working conditions; 4) The combination of the modules proposed in the present invention enables the effectiveness of the active braking control strategy under different road conditions, ensuring the feasibility and reliability of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0020] Figure 1 A schematic diagram of an active safety control device for avoiding collisions with pedestrians applicable to multiple road conditions according to the present invention; Figure 2 Schematic diagram of vehicle coordinate transformation; Figure 3It is a schematic diagram of the vehicle braking process; Figure 4 This is a schematic diagram of the dangerous collision area; Figure 5 Schematic diagram of vehicle-pedestrian interaction in scenario 1; Figure 6 Schematic diagram of vehicle-pedestrian interaction in scenario 2; Figure 7 Schematic diagram of vehicle-pedestrian interaction in scenario 3; Figure 8 Schematic diagram of vehicle-pedestrian interaction in scenario 4; Fig. 9 Schematic diagram of vehicle-pedestrian interaction in scenario 5; Fig.10 Schematic diagram of vehicle-pedestrian interaction conditions in scenario 6; Fig.11 It is a schematic diagram of collision risk level area; Fig.12 Schematic diagram of the collision risk assessment decision process. DETAILED DESCRIPTION

[0021] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0022] Example 1: See Figure 1-Figure 12 This embodiment provides a pedestrian collision avoidance active safety control method suitable for multiple road conditions, and the specific method is as follows: Step 1: Calculate the normalized lateral and longitudinal tire forces through the vehicle dynamics model and tire model. Then, use the lateral and longitudinal accelerations and yaw rate of the vehicle as observation vectors to obtain an estimated value of the road adhesion coefficient, as follows: 11) Establish the vehicle dynamics equation as follows: , , , In the formula, is the total mass of the vehicle; is the vehicle longitudinal acceleration; is the longitudinal force of the left front wheel; is the longitudinal force of the right front wheel; is the lateral force of the left front wheel; is the lateral force of the right front wheel; is the front wheel turning angle; is the longitudinal force of the left rear wheel; is the longitudinal force of the right rear wheel; is the lateral force of the left rear wheel; is the lateral force of the right rear wheel; is the lateral acceleration of the vehicle; is the moment of inertia of the vehicle around the z-axis; is the yaw angular acceleration; is the distance from the center of mass to the front axle; is the distance from the center of mass to the rear axle; is the front wheel track; Rear wheel track.

[0023] 12) Obtain the normalized lateral and longitudinal tire forces as follows: Considering the axle load transfer during vehicle driving, the vertical load formula of each wheel of the vehicle is: , In the formula, is the vertical load on the left front wheel; is the vertical load on the right front wheel; is the vertical load on the left rear wheel; is the vertical load on the right rear wheel; is the distance from the front axle to the rear axle; is the vehicle center of mass height.

[0024] For single wheel force analysis, the formulas for tire longitudinal force and lateral force in the tire coordinate system are: , , In the formula, is the longitudinal force in the wheel coordinate system; is the lateral force in the wheel coordinate system; is the road adhesion coefficient; is the normalized tire longitudinal force of each tire; is the normalized tire lateral force of each tire; is the vertical load of the tire; is the tire longitudinal slip coefficient; is the tire cornering stiffness; is the tire slip rate; is the tire slip angle; is the boundary correction value; is the boundary correction function of the tire force.

[0025] 13) The extended Kalman filter is used to estimate the road adhesion coefficient of each wheel, and the vehicle dynamics system is modeled as follows: , , In the formula, and is the state variable; is the observed variable; is the input variable, is the nonlinear state transfer equation; is the nonlinear observation equation; is the process noise; To measure noise; Among them, the state variable , input variable , observed variables ; The calculation formula of the observation equation is: , In the formula, is the road adhesion coefficient of the left front wheel; is the normalized longitudinal force of the left front wheel; is the normalized lateral force of the left front wheel; is the road adhesion coefficient of the right front wheel; is the normalized longitudinal force of the right front wheel; is the normalized lateral force of the right front wheel; is the road adhesion coefficient of the left rear wheel; is the normalized longitudinal force of the left rear wheel; is the normalized lateral force of the left rear wheel; is the road adhesion coefficient of the right rear wheel; is the normalized longitudinal force of the right rear wheel; is the normalized lateral force on the right rear wheel.

[0026] The Jacobian matrix of the partial derivative of the state vector is solved for the observation equation, and the observation equation of the road adhesion coefficient estimation algorithm is obtained as follows: , In the formula, is the influence coefficient of the left front wheel on the yaw acceleration; is the influence coefficient of the right front wheel on the yaw acceleration; is the influence coefficient of the left rear wheel on the yaw angular acceleration; is the influence coefficient of the right rear wheel on the yaw acceleration; is the road adhesion coefficient of the left front wheel at time step k; is the road adhesion coefficient of the right front wheel at time step k; is the road adhesion coefficient of the left rear wheel at time step k; is the road adhesion coefficient of the right rear wheel at time step k.

[0027] in: , , , .

[0028] Wherein, in step 13), the specific method of the extended Kalman filter is as follows: 131) Initialize state variables and set state variables and the covariance matrix ; Define the noise covariance matrices Q and R to represent process noise and measurement noise respectively; 132) Time update: Generate the Sigma point of the state variable through unscented transformation as follows: , In the formula, is the Sigma point matrix; is the state mean estimate; is the state variable dimension; is the scaling parameter; is the state covariance matrix.

[0029] Substitute each Sigma point into the nonlinear state transfer equation , as shown below: , In the formula, is the Sigma point after nonlinear transformation, that is, the state prediction value at time k+1.

[0030] Calculate the predicted state mean based on the propagated Sigma point and the predicted covariance , as shown below: , , In the formula, is the state prediction value at time k+1; is the prediction covariance; is the weight of the mean; is the weight of the covariance; is the process noise covariance matrix.

[0031] 133) Measurement update, through the observation equation , calculate the observed value of each Sigma point as follows: , In the formula, is the observed value corresponding to the Sigma point; is the input variable; Calculate the predicted observation mean as follows: , In the formula, is the mean of the predicted observations; is the mean weight; The observation covariance is calculated as follows: , In the formula, is the observation covariance matrix; is the measurement noise covariance matrix; The cross-covariance of states and observations is calculated as follows: , In the formula, is the cross-covariance matrix between states and observations; , , , In the formula, is the Kalman gain matrix; is the inverse matrix of the observation covariance; is the updated state mean; is the actual observed value; is the updated state covariance matrix; is the prediction covariance matrix.

[0032] Based on the iteration of 131)-133), the estimation of the road adhesion coefficient can be realized, and this method can be configured in the vehicle controller.

[0033] Step 2: Perform coordinate transformation based on the current vehicle coordinate system, assuming that the pedestrian moves in a straight line at a uniform speed, and predict the pedestrian's motion trajectory, as follows: See also Figure 2 , map the state vectors of the pedestrian and the main vehicle at the previous moment to the vehicle coordinate system at the current moment.

[0034] 21) During the vehicle motion, assuming that the vehicle speed and yaw rate remain unchanged, the vehicle motion model is a uniform angular velocity and uniform speed model, as shown below: , , In the formula, is the longitudinal displacement of the vehicle; is the lateral displacement of the vehicle; is the vehicle speed; is the vehicle yaw angular velocity; is the vehicle yaw angular velocity; is the time step.

[0035] 22) To simplify the calculation, the pedestrian and vehicle motions are decoupled into two motion modes: translation and rotation, as follows: The coordinate mapping of the pedestrian motion state vector of the vehicle translation motion is as follows: , In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current coordinate system; is the pedestrian's horizontal coordinate at the previous moment in the current coordinate system; is the pedestrian's longitudinal speed at the previous moment in the current coordinate system; is the pedestrian's lateral speed at the previous moment in the current coordinate system; is the pedestrian's longitudinal coordinate in the coordinate system of the previous moment; is the lateral coordinate of the pedestrian in the coordinate system of the previous moment; At the current time in the vehicle coordinate system, the state vector mapping result of the pedestrian at the previous moment is : In the formula, is the motion state vector of the pedestrian; is the longitudinal coordinate of the pedestrian in the current vehicle coordinate system; is the pedestrian's lateral coordinate; is the pedestrian longitudinal speed; is the lateral speed of the pedestrian.

[0036] The mapping coordinates of the pedestrian motion state vector after the vehicle rotation are updated as follows: , In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current coordinate system; is the horizontal coordinate of the pedestrian at the previous moment.

[0037] From the above formula, we can know that the pedestrian motion state vector at the previous moment in the current vehicle coordinate system is: , In the formula, is the longitudinal coordinate of the pedestrian at the previous moment in the current vehicle coordinate system; is the lateral coordinate of the pedestrian at the previous moment in the current vehicle coordinate system; is the pedestrian's longitudinal speed at the previous moment in the current vehicle coordinate system; It is the pedestrian's lateral speed at the previous moment in the current vehicle coordinate system.

[0038] Since the motion state of pedestrians changes slowly relative to the motion state of vehicles, it is assumed that pedestrians move at a uniform speed, that is, the pedestrian motion model is a uniform motion model. Therefore, the motion state of pedestrians at the current moment is: , 23) Since the operating cycle of vehicle components such as on-board sensors and controllers is generally 10 , according to Taylor's formula, the trigonometric function is approximated as: , The calculation formula for the pedestrian motion state vector at the current moment is: , 24) Use recursive least squares identification method to predict pedestrian motion state: , in, The matrices are: ， , In the formula, is the horizontal and vertical coordinates of the pedestrian, that is ; is the pedestrian motion state vector at the previous moment, that is .

[0039] Among them, in step 24), the specific method of using the recursive least squares identification method to predict the pedestrian motion state is as follows: 241) Calculate the error between the actual value of the pedestrian's motion state and the predicted value at the previous moment : , 242) According to the error value , calculate the gain matrix , and introduce the forgetting factor , is the covariance matrix obtained in the previous calculation; therefore, the gain matrix calculation The calculation formula is as follows: , In the formula, is the gain matrix; is the covariance matrix of the previous moment; is the state transition matrix; For the forgetting factor; is the identity matrix.

[0040] 243) According to the gain matrix , and the covariance matrix Update to prepare for the next trajectory prediction algorithm calculation; among them, the covariance matrix The calculation formula is: , In the formula, is the updated covariance matrix.

[0041] 244) According to the gain matrix , update the pedestrian state vector prediction value: .

[0042] Based on the iteration of steps 241)-244), the prediction of the pedestrian trajectory can be achieved, and the algorithm can be configured in the vehicle controller; Step 3: Determine the collision avoidance control timing adopted by the pedestrian collision avoidance active safety control, as follows: 31) Establish a safe distance model; The emergency braking process is divided into four stages, namely, the driver's reaction after encountering danger, braking reaction, continuous braking and releasing the brake; the braking process is shown in 3; Among them, the critical braking distance is expressed as: , in, is the current speed; is the maximum braking deceleration; for driver reaction time; is the braking reaction time; Considering the influencing factors of the road adhesion coefficient, the safety distance model introduces the influencing factors of the road adhesion coefficient , as shown below: , In the formula, is the current road adhesion coefficient; is the maximum adhesion coefficient; is the minimum adhesion coefficient; The safety distance model considering the road adhesion coefficient is as follows: , In the formula, is the minimum safe distance; 32) Establish a collision risk assessment model; The collision time threshold is dynamically calculated based on the motion state of the vehicle and pedestrian and the road adhesion coefficient to quantify the collision risk. The danger zone can be defined as the overlap between the vehicle trajectory and the pedestrian trajectory. Figure 4 shown.

[0043] The time it takes for a vehicle to pass through the danger zone is expressed as: , , In the formula, The time when the vehicle enters the danger zone; The time it takes for the vehicle to leave the danger zone; is the longitudinal displacement of the vehicle at the current moment; The lateral safety distance for pedestrians; is the longitudinal speed of the vehicle relative to the pedestrian at the current moment; is the vehicle length; The time it takes for a pedestrian to pass through the danger zone is expressed as: , , In the formula, The starting time for pedestrians to leave the danger zone; Time for pedestrians to leave the danger zone; is the lateral displacement of the vehicle at the current moment; is the lateral safety distance of the vehicle; is the vehicle width; is the lateral speed of the pedestrian relative to the vehicle at the current moment; 33) By comparing the starting and ending points of the time periods when vehicles and pedestrians pass through the dangerous area, they can be divided into six situations, such as Figure 5-Figure 10 shown.

[0044] According to the above situation, the collision risk level at the current moment is determined and the collision time is obtained. , as shown in the following table.

[0045]

[0046] 34) According to the collision time Determine the risk level and take appropriate measures based on the risk level; For pedestrian crossings, the risk level is divided into warning zone and automatic braking zone, such as Fig.11 As shown; among them, the warning area reminds the driver to pay attention to the danger, and the automatic braking area adopts automatic emergency braking control to avoid collision.

[0047] The TTC thresholds for the warning area and automatic braking area are defined as follows: , , In the formula, is the warning area threshold; is the automatic braking area threshold; is the TTC threshold coefficient. The decision logic of risk assessment is as follows: Fig.12 The collision risk assessment algorithm is mainly based on the collision time (TTC) judgment, in which the collision time at the current moment is mainly calculated by comparing the time nodes when the vehicle and pedestrian pass through the dangerous collision area, and compared with the TTC threshold to obtain the collision risk level at the current moment.

[0048] Step 4: Consider the vehicle dynamics, potential collision risk of pedestrians, and road traffic safety factors of pedestrians and drivers, and make active safety control decisions for pedestrian collision avoidance, as follows: 41) When there is a risk of collision with a pedestrian in front, if the driver or AEB-P intervenes, the TTC value will first decrease slowly, then gradually increase to infinity, otherwise it will gradually decrease to zero. Get TTC error value , as follows: , In the formula, is the TTC critical value; 42) According to different situations and road adhesion conditions, the collision time error and road adhesion coefficient are selected as the system input, and the expected deceleration is selected as the system output; the Gaussian function is selected as the membership function, and the fuzzy reasoning model is used for decision making.

[0049] Step 5: By establishing an inverse dynamics model of the vehicle braking system, according to the desired braking deceleration obtained, since the control quantity of the vehicle's braking system is generally the braking pressure, it is necessary to convert the desired deceleration obtained by the decision algorithm into the braking pressure; therefore, the braking pressure is calculated to achieve braking control of the vehicle, as follows: 51) In the feedforward stage, the inverse dynamics model of the braking system is established as follows: , In the formula, is the braking deceleration; For braking force; is the air resistance coefficient; is the frontal area of ​​the vehicle; is the air density; is the rolling resistance coefficient; 52) When the braking force does not exceed the maximum braking force limited by road conditions, the braking force and braking pressure are generally linearly related, so the vehicle braking pressure The calculation of is as follows: , In the formula, is the proportional coefficient of braking force and braking force pressure; 53) In the feedback stage, the PID algorithm is used for feedback control, as follows: , , In the formula, For target pressure; is the deceleration error of the low-level controller; is the target deceleration; is the actual deceleration; is the proportionality coefficient; is the integration coefficient, is the differential coefficient.

[0050] Example 2: See Figure 1 This embodiment provides a pedestrian collision avoidance active safety control device adapted to multiple road conditions, which is used to implement a pedestrian collision avoidance active safety control method adapted to multiple road conditions described in Embodiment 1, including: The road adhesion coefficient estimation module 110 is used to calculate the normalized lateral and longitudinal tire forces through the vehicle dynamics model and the tire model, and then use the lateral and longitudinal accelerations and yaw rate of the vehicle as observation vectors to obtain an estimated value of the road adhesion coefficient; A pedestrian trajectory prediction module 120 is used to perform coordinate transformation based on the current vehicle coordinate system, assuming that the pedestrian moves in a straight line at a uniform speed, and predict the pedestrian's movement trajectory; A collision risk assessment module 130, used to determine the collision avoidance control timing adopted by the pedestrian collision avoidance active safety control; The active control upper layer module 140 is used to make active safety control decisions for pedestrian collision avoidance by considering vehicle dynamics characteristics, potential collision risks of pedestrians, and road traffic safety factors of pedestrians and drivers; The active control lower layer module 150 is used to establish an inverse dynamics model of the vehicle braking system and calculate the braking pressure according to the obtained expected braking deceleration to achieve braking control of the vehicle.

[0051] Embodiment 3: This embodiment provides a terminal, including: one or more processors; a memory for storing said one or more processor executable instructions; Wherein, the one or more processors are configured to: An active safety control method for pedestrian collision avoidance adapted to multiple road conditions as described in Embodiment 1 is implemented.

[0052] Embodiment 4: This embodiment provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by the processor of the terminal, the terminal can execute an active safety control method for pedestrian collision avoidance that is suitable for multiple road conditions as described in Embodiment 1.

[0053] In summary, the present invention can be applied to different road conditions, and effectively improves the practicality and robustness of the active safety control method for pedestrian protection.

[0054] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A pedestrian collision avoidance active safety control method adapted to multiple road conditions, characterized in that: The following steps are involved: Step 1: Calculate the normalized lateral and longitudinal tire forces through the vehicle dynamics model and the tire model, and then use the lateral and longitudinal accelerations and yaw rate of the vehicle as observation vectors to obtain an estimated value of the road adhesion coefficient; Step 2: Perform coordinate transformation based on the current vehicle coordinate system, assuming that the pedestrian moves in a straight line at a uniform speed, and predict the pedestrian's movement trajectory; Step 3: Determine the collision avoidance control timing adopted by the pedestrian collision avoidance active safety control; Step 4: Consider the vehicle dynamics, potential collision risk of pedestrians, and road traffic safety factors of pedestrians and drivers to make active safety control decisions for pedestrian collision avoidance; Step 5: By establishing an inverse dynamics model of the vehicle braking system, the braking pressure is calculated according to the desired braking deceleration to achieve braking control of the vehicle.

2. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 1, characterized in that: The specific method of step one is as follows: 11) Establish the vehicle dynamics equation as follows: ， ， ， In the formula, is the total mass of the vehicle; is the vehicle longitudinal acceleration; is the longitudinal force of the left front wheel; is the longitudinal force of the right front wheel; is the lateral force of the left front wheel; is the lateral force of the right front wheel; is the front wheel turning angle; is the longitudinal force of the left rear wheel; is the longitudinal force of the right rear wheel; is the lateral force of the left rear wheel; is the lateral force of the right rear wheel; is the lateral acceleration of the vehicle; is the moment of inertia of the vehicle around the z-axis; is the yaw angular acceleration; is the distance from the center of mass to the front axle; is the distance from the center of mass to the rear axle; is the front wheel track; is the rear wheel track; 12) Establish a tire model as follows: Considering the axle load transfer during vehicle driving, the vertical load formula of each wheel of the vehicle is: ， In the formula, is the vertical load on the left front wheel; is the vertical load on the right front wheel; is the vertical load on the left rear wheel; is the vertical load on the right rear wheel; is the distance from the front axle to the rear axle; is the height of the vehicle's center of mass; For single wheel force analysis, the formulas for tire longitudinal force and lateral force in the tire coordinate system are: ， ， In the formula, is the longitudinal force in the wheel coordinate system; is the lateral force in the wheel coordinate system; is the road adhesion coefficient; is the normalized tire longitudinal force of each tire; is the normalized tire lateral force of each tire; is the vertical load of the tire; is the tire longitudinal slip coefficient; is the tire cornering stiffness; is the tire slip rate; is the tire slip angle; is the boundary correction value; is the boundary correction function of the tire force; 13) The extended Kalman filter is used to estimate the road adhesion coefficient of each wheel, and the vehicle dynamics system is modeled as follows: ， ， In the formula, and is the state variable; is the observed variable; is the input variable, is the nonlinear state transfer equation; is the nonlinear observation equation; is the process noise; To measure noise; Among them, the state variable , input variable , observed variables ; The calculation formula of the observation equation is: ， In the formula, is the road adhesion coefficient of the left front wheel; is the normalized longitudinal force of the left front wheel; is the normalized lateral force of the left front wheel; is the road adhesion coefficient of the right front wheel; is the normalized longitudinal force of the right front wheel; is the normalized lateral force of the right front wheel; is the road adhesion coefficient of the left rear wheel; is the normalized longitudinal force of the left rear wheel; is the normalized lateral force of the left rear wheel; is the road adhesion coefficient of the right rear wheel; is the normalized longitudinal force of the right rear wheel; is the normalized lateral force of the right rear wheel; The Jacobian matrix of the partial derivative of the state vector is solved for the observation equation, and the observation equation of the road adhesion coefficient estimation algorithm is obtained as follows: ， In the formula, is the influence coefficient of the left front wheel on the yaw acceleration; is the influence coefficient of the right front wheel on the yaw acceleration; is the influence coefficient of the left rear wheel on the yaw angular acceleration; is the influence coefficient of the right rear wheel on the yaw acceleration; is the road adhesion coefficient of the left front wheel at time step k; is the road adhesion coefficient of the right front wheel at time step k; is the road adhesion coefficient of the left rear wheel at time step k; is the road adhesion coefficient of the right rear wheel at time step k; in: ， ， ， 。 3. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 2, characterized in that: In step 13), the specific method of the extended Kalman filter is as follows: 131) Initialize state variables and set state variables and the covariance matrix ; The noise covariance matrices Q and R are defined to represent process noise and measurement noise respectively; 132) Time update: Generate the Sigma point of the state variable through unscented transformation as follows: ， In the formula, is the Sigma point matrix; is the state mean estimate; is the state variable dimension; is the scaling parameter; is the state covariance matrix; Substitute each Sigma point into the nonlinear state transfer equation , as shown below: ， In the formula, is the Sigma point after nonlinear transformation, that is, the state prediction value at time k+1; Calculate the predicted state mean based on the propagated Sigma point and the predicted covariance , as shown below: ， ， In the formula, is the state prediction value at time k+1; is the prediction covariance; is the weight of the mean; is the weight of the covariance; is the process noise covariance matrix; 133) Measurement update, through the observation equation , calculate the observed value of each Sigma point as follows: ， In the formula, is the observed value corresponding to the Sigma point; is the input variable; Calculate the predicted observation mean as follows: ， In the formula, is the mean of the predicted observations; is the mean weight; The observation covariance is calculated as follows: ， In the formula, is the observation covariance matrix; is the measurement noise covariance matrix; The cross-covariance of states and observations is calculated as follows: ， In the formula, is the cross-covariance matrix between states and observations; Update the state estimate using the Kalman gain Update the state mean and covariance as follows: ， ， ， In the formula, is the Kalman gain matrix; is the inverse matrix of the observation covariance; is the updated state mean; is the actual observed value; is the updated state covariance matrix; is the prediction covariance matrix.

4. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 1, characterized in that: The specific method of step 2 is as follows: 21) During the vehicle motion, assuming that the vehicle speed and yaw rate remain unchanged, the vehicle motion model is a uniform angular velocity and uniform speed model, as shown below: ， ， In the formula, is the longitudinal displacement of the vehicle; is the lateral displacement of the vehicle; is the vehicle speed; is the vehicle yaw angular velocity; is the vehicle yaw angular velocity; is the time step; 22) The pedestrian and vehicle motions are decoupled into two motion modes: translation and rotation, as follows: The coordinate mapping of the pedestrian motion state vector of the vehicle translation motion is as follows: ， In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current coordinate system; is the pedestrian's horizontal coordinate at the previous moment in the current coordinate system; is the pedestrian's longitudinal speed at the previous moment in the current coordinate system; is the pedestrian's lateral speed at the previous moment in the current coordinate system; is the pedestrian's longitudinal coordinate in the coordinate system of the previous moment; is the lateral coordinate of the pedestrian in the coordinate system of the previous moment; At the current time in the vehicle coordinate system, the state vector mapping result of the pedestrian at the previous moment is : In the formula, is the motion state vector of the pedestrian; is the longitudinal coordinate of the pedestrian in the current vehicle coordinate system; is the lateral coordinate of the pedestrian; is the pedestrian longitudinal speed; is the pedestrian lateral speed; The mapping coordinates of the pedestrian motion state vector after the vehicle rotation are updated as follows: ， In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current coordinate system; is the horizontal coordinate of the pedestrian at the previous moment; The pedestrian motion state vector at the previous moment in the current vehicle coordinate system is: ， In the formula, is the pedestrian's longitudinal coordinate at the previous moment in the current vehicle coordinate system; is the lateral coordinate of the pedestrian at the previous moment in the current vehicle coordinate system; is the pedestrian's longitudinal speed at the previous moment in the current vehicle coordinate system; is the pedestrian's lateral speed at the previous moment in the current vehicle coordinate system; Since the motion state of pedestrians changes slowly relative to the motion state of vehicles, it is assumed that pedestrians move at a uniform speed, that is, the pedestrian motion model is a uniform motion model. Therefore, the motion state of pedestrians at the current moment is: ； 23) According to Taylor's formula, the trigonometric function is approximated as: ， The calculation formula for the pedestrian motion state vector at the current moment is: ； 24) Use the recursive least squares identification method to predict the pedestrian's motion state, that is, the pedestrian's motion trajectory: ， in, The matrices are: ， ， In the formula, is the horizontal and vertical coordinates of the pedestrian, that is ; is the pedestrian motion state vector at the previous moment, that is .

5. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 4, characterized in that: In step 24), the specific method of using the recursive least squares identification method to predict the pedestrian motion state is as follows: 241) Calculate the error between the actual value of the pedestrian's motion state and the predicted value at the previous moment : ， 242) According to the error value , calculate the gain matrix , and introduce the forgetting factor , is the covariance matrix obtained in the previous calculation; therefore, the gain matrix calculation The calculation formula is as follows: ， In the formula, is the gain matrix; is the covariance matrix of the previous moment; is the state transition matrix; For the forgetting factor; is the identity matrix; 243) According to the gain matrix , and the covariance matrix Update to prepare for the next trajectory prediction algorithm calculation; among them, the covariance matrix The calculation formula is: ， In the formula, is the updated covariance matrix; 244) According to the gain matrix , update the pedestrian state vector prediction value: 。 6. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 1, characterized in that: The specific method of step three is as follows: 31) Establish a safe distance model; The emergency braking process is divided into four stages, namely, the driver's reaction to danger, braking reaction, continuous braking, and releasing the brake; among them, the critical braking distance is expressed as: ， in, is the current speed; is the maximum braking deceleration; for driver reaction time; is the braking reaction time; The influencing factors of road adhesion coefficient introduced into the safety distance model , as shown below: ， In the formula, is the current road adhesion coefficient; is the maximum adhesion coefficient; is the minimum adhesion coefficient; The safety distance model considering the road adhesion coefficient is as follows: ， In the formula, is the minimum safe distance; 32) Establish a collision risk assessment model; The collision risk is assessed based on the safety distance and collision time threshold. The collision time threshold is dynamically calculated based on the motion state of the vehicle and pedestrian and the road adhesion coefficient to quantify the collision risk. The danger zone is defined as the overlapping part of the vehicle trajectory and the pedestrian trajectory. The time it takes for a vehicle to pass through the danger zone is expressed as: ， ， In the formula, The time when the vehicle enters the danger zone; The time it takes for the vehicle to leave the danger zone; is the longitudinal displacement of the vehicle at the current moment; The lateral safety distance for pedestrians; is the longitudinal speed of the vehicle relative to the pedestrian at the current moment; is the vehicle length; The time it takes for a pedestrian to pass through the danger zone is expressed as: ， ， In the formula, The starting time for pedestrians to leave the danger zone; Time for pedestrians to leave the danger zone; is the lateral displacement of the vehicle at the current moment; is the lateral safety distance of the vehicle; is the vehicle width; is the lateral speed of the pedestrian relative to the vehicle at the current moment; 33) By comparing the start and end points of the time period when vehicles and pedestrians pass through the dangerous area, the collision risk level at the current moment is determined and the collision time is obtained. ; 34) According to the collision time Determine the risk level and take appropriate measures based on the risk level; For pedestrian crossings, the risk level is divided into a warning zone and an automatic braking zone; the warning zone alerts the driver to the danger, and the automatic braking zone takes automatic emergency braking control to avoid collision; The TTC thresholds for the warning area and automatic braking area are defined as follows: ， ， In the formula, is the warning area threshold; is the automatic braking area threshold; is the TTC threshold coefficient.

7. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 1, characterized in that: The specific method of step 4 is as follows: 41) Get TTC error value , as follows: ， In the formula, is the TTC critical value; 42) According to different situations and road adhesion conditions, the collision time error and road adhesion coefficient are selected as the system input, and the expected deceleration is selected as the system output; the Gaussian function is selected as the membership function, and the fuzzy reasoning model is used for decision making.

8. The active safety control method for pedestrian collision avoidance adapted to multiple road conditions according to claim 1, characterized in that: The specific method of step five is as follows: 51) In the feedforward stage, the inverse dynamics model of the braking system is established as follows: ， In the formula, is the braking deceleration; For braking force; is the air resistance coefficient; is the frontal area of ​​the vehicle; is the air density; is the rolling resistance coefficient; 52) Calculate the brake pressure as follows: ， In the formula, is the proportional coefficient of braking force and braking force pressure; 53) In the feedback stage, the PID algorithm is used for feedback control, as follows: ， ， In the formula, For target pressure; is the deceleration error of the low-level controller; is the target deceleration; is the actual deceleration; is the proportionality coefficient; is the integration coefficient, is the differential coefficient.

9. An active safety control device for pedestrian collision avoidance adapted to multiple road conditions, used to implement an active safety control method for pedestrian collision avoidance adapted to multiple road conditions as claimed in any one of claims 1 to 8, characterized in that: include: A road adhesion coefficient estimation module (110) is used to calculate normalized lateral and longitudinal tire forces through a vehicle dynamics model and a tire model, and then use the lateral and longitudinal accelerations and yaw angular velocity of the vehicle as observation vectors to obtain an estimated value of the road adhesion coefficient; A pedestrian trajectory prediction module (120) is used to perform coordinate conversion based on the current vehicle coordinate system, assuming that the pedestrian moves in a straight line at a uniform speed, and predict the pedestrian's movement trajectory; A collision risk assessment module (130), used to determine the collision avoidance control timing adopted by the pedestrian collision avoidance active safety control; An active control upper layer module (140) is used to make active safety control decisions for pedestrian collision avoidance by considering vehicle dynamics characteristics, potential collision risks of pedestrians, and road traffic safety factors of pedestrians and drivers; The active control lower layer module (150) is used to establish an inverse dynamics model of the vehicle braking system and calculate the braking pressure according to the obtained expected braking deceleration, so as to realize braking control of the vehicle.

Images