CN114371707A - Pedestrian trajectory prediction and active collision avoidance method and system considering human-vehicle interaction - Google Patents

Abstract

The invention discloses a pedestrian track prediction and active collision avoidance method and system considering human-vehicle interaction, which are used for identifying the face orientation of a pedestrian in an image in front of a vehicle to judge whether the pedestrian notices the coming vehicle or not, and integrating pedestrian motion state information, the face orientation of the pedestrian and vehicle motion state information to judge the intention of the pedestrian; predicting the pedestrian movement without noticing the vehicle by utilizing a Markov pedestrian model; simultaneously introducing a social force model and a Markov pedestrian model to predict the movement of a pedestrian who pays attention to the vehicle and continues to walk, and performing weighted fusion on the prediction results of the two models to obtain a corrected pedestrian position so as to obtain a track curve within a preset time length; judging the safety state of the pedestrian track, and deciding a corresponding collision avoidance strategy; meanwhile, a pedestrian track prediction and active collision avoidance system considering human-vehicle interaction is constructed; the method and the system designed by the invention can improve the safety and the stability of the intelligent automobile, and ensure that the whole longitudinal and transverse collision avoidance decision system is more perfect and effective.

Description

Pedestrian trajectory prediction and active collision avoidance method and system considering human-vehicle interaction

Technical Field

The invention belongs to the technical field of intelligent driving safety, and particularly relates to a pedestrian trajectory prediction and active collision avoidance method and system considering human-vehicle interaction.

Background

With the continuous progress and development of global industry, the automobile holding amount is in a continuous increasing trend, so that the road traffic situation is more severe. In an urban road scene, vulnerable pedestrians are used as individuals moving independently, and when crossing a road, the vulnerable pedestrians are likely to collide or collide with vehicles coming and going on the road, so that traffic accidents are frequent.

The existing research on a pedestrian collision avoidance method can be roughly divided into four aspects of detection and identification, prediction, risk assessment, collision avoidance and the like of pedestrians. In the aspect of pedestrian trajectory prediction, the main problems of the existing methods are that complicated and changeable motion characteristics of pedestrians cannot be considered, and the influence of the surrounding environment on the motion state of the pedestrians is ignored, so that the predicted pedestrian trajectory is greatly different from the real pedestrian trajectory. In the aspect of collision avoidance methods, most of the existing methods adopt independent longitudinal collision avoidance or transverse collision avoidance, the two modes have respective adaptive working conditions, and the variable behaviors of pedestrians can cause interference on collision avoidance decisions of intelligent automobiles.

Disclosure of Invention

Aiming at the defects in the prior art, the invention provides a pedestrian track prediction and collision avoidance control method and system considering human-vehicle interaction; the future movement track of the pedestrian can be reasonably predicted, and a collision avoidance decision is made for the dangerous state, so that stable and safe driving of the vehicle can be guaranteed, the pedestrian can be protected, and road traffic accidents are reduced.

In order to solve the problems, the invention adopts the following technical proposal,

a pedestrian track prediction and vehicle active collision avoidance method considering human-vehicle interaction comprises the following steps:

and S1, acquiring the driving state information of the vehicle, the motion state information of the pedestrian and the front image of the vehicle.

S2, extracting a head region of the pedestrian and identifying the face direction of the pedestrian according to the acquired image in front of the vehicle, and judging whether the pedestrian notices the coming vehicle according to the face direction of the pedestrian; and on the basis of the logistic regression model, fusing the pedestrian motion state information, the pedestrian face orientation and the vehicle motion state information to judge the intention of the pedestrian.

S3, according to the face orientation of the pedestrian, if the pedestrian is judged not to notice the vehicle, predicting the track of the pedestrian in a future preset time length by using a Markov pedestrian model;

s4, according to the face orientation and intention judgment result of the pedestrian, if the pedestrian is judged to notice the vehicle and continue to walk, introducing a social force model to predict the motion of the pedestrian, and introducing a Markov pedestrian model to predict in the motion process of the pedestrian; carrying out weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain a corrected pedestrian position; and obtaining a track curve within a preset time length according to the set time step length.

And S5, judging the safety state according to the predicted pedestrian track, and deciding a proper collision avoidance strategy of the vehicle under the condition of ensuring the safety.

Further, in S2, the step of identifying the face orientation of the pedestrian includes:

(1) and detecting the head area of the pedestrian by adopting a Yolo algorithm according to the image in front of the vehicle to obtain a head image.

(2) And (3) adopting a convolutional neural network to build a forward/lateral/backward classifier of the face of the pedestrian, and determining the face Orientation of the pedestrian.

Further, in the above S2, the logistic regression model for pedestrian intention determination includes the steps of:

(1) at pedestrian speed v_pedOrientation of pedestrian face, vehicle speed v_vehicleThe Distance between the pedestrian and the vehicle is an independent variable, the intention of the pedestrian to select walking or stopping facing the coming vehicle is a dependent variable in the human-vehicle interaction process, and a function expression in a logistic regression model for judging the intention of the pedestrian is as follows:

wherein the independent variable x is [ c, v ]_ped,Orientation,v_vehicle,Distance]^TC isAn arbitrary constant term. Theta is ═ theta₀,θ₁,θ₂,θ₃,θ₄]^TIs a set of coefficients, θ_iIs the coefficient corresponding to the ith argument, i ═ 0, 1, 2, 3, 4; determining the minimum value theta of the cost function by means of a gradient descent method_iIs the coefficient corresponding to the ith argument; the expression formula of the cost function is as follows:

where J (θ) is the cost function, m is the number of samples, H_θTo assume the value, y⁽ⁱ⁾Is the ith true value, x⁽ⁱ⁾Is the ith independent variable, x⁽ⁱ⁾∈x。

(2) Judging the intention of the pedestrian by using the trained logistic regression model, and if H is the intention of the pedestrian, judging the intention of the pedestrian_θ> 0.5 indicates pedestrian walking, H_θ< 0.5 indicates that the pedestrian stops.

Further, in the above S3, the process of predicting the trajectory of the pedestrian in the preset time period in the future by using the markov pedestrian model is as follows:

the pedestrian does not notice the vehicle, the vehicle on the road can be regarded as the pedestrian of noninterference at this moment, the pedestrian crossing freely moves and accords with the Markov process under the condition without external disturbance, the pedestrian future position and speed depend on his present position and speed, therefore can get the state description of the pedestrian:

State_ped＝(x_p(t),y_p(t),v_x-p(t),v_y-p(t))

in the formula, State_pedIs a pedestrian state, v_x-p(t) and v_y-p(t) shows the speed of the pedestrian at time t in the X-direction and Y-direction, respectively, Δ v_x-pAnd Δ v_y-pRepresenting the velocity increase at time t, v, in the X and Y directions, respectively_x-p(t + Δ t) and v_y-p(t + Δ t) represents the speed of the pedestrian at the time t + Δ t in the X-direction and Y-direction, respectively, k_xAnd k_yThe number of the symbols representing the constant number,

and

respectively representing the desired speed, epsilon, of the pedestrian in the X-direction and in the Y-direction_xAnd ε_yRepresenting random disturbances, X, of the pedestrian's velocity increment in the X and Y directions, respectively_p(t) and y_p(t) represents the displacement function of the pedestrian in the X and Y directions with respect to time t, p_mAnd (t) represents the position of the pedestrian at t predicted by the Markov pedestrian model.

Further, in S4, the predicting the motion of the pedestrian using the social force model includes:

the pedestrian moves to the target point and has a driving force, the pedestrian can receive the repulsion force of the vehicle when facing the coming vehicle, the road can also apply a hidden boundary force to the pedestrian, the forces are accumulated to form a resultant force, and the position of the pedestrian is recurred along with the time step under the action of the social force. The expression formula of the resultant force and the position of the pedestrian is as follows:

in the formula, F_ped(t) social force resultant force to which pedestrian is subjected, F_dRepresenting the driving force received by the pedestrian moving towards the target point, F_vpIndicating the repulsive force of the pedestrian facing the incoming vehicle, F_eIndicating that the road is applying a hidden boundary force to the pedestrian. v. of_n(t) represents the speed of the pedestrian at t, p_s(t) the position of the pedestrian at t predicted by the social force model, p_s(t + Δ t) represents the position of the pedestrian at t + Δ t, Δ t represents a time step, and m represents the mass of the pedestrian.

Further, in S4, the predicted pedestrian position is obtained by weighting the pedestrian position predicted by the markov pedestrian model and the pedestrian position predicted by the social force model, taking the motion of the pedestrian crossing the street into consideration as a combination of the free disturbance-free motion and the incoming disturbance motion, and is expressed by the following equation:

p(t)＝τ_m·p_m(t)+τ_s·p_s(t)

wherein p (t) represents the pedestrian position at t obtained by fusing and correcting the Markov pedestrian model and the social force model, and τ_mAnd τ_sRespectively represents p_m(t) and p_s(t) weight coefficient.

Further, in S5, the method for determining the safety state of the pedestrian and the vehicle includes:

(1) dividing the position of the pedestrian on the road into a dangerous area, a high risk area and a safe area;

(2) calculating the longitudinal collision avoidance time of the vehicle, and considering the position fluctuation of the pedestrian position in the longitudinal direction, wherein the obtained longitudinal collision avoidance time of the vehicle is a time region range:

in the formula, t_vIndicating longitudinal time to collision, v, of the vehicle_vehicleRepresenting the vehicle speed, deltas representing the longitudinal relative distance between the vehicle and the pedestrian, epsilon representing the longitudinal offset when the pedestrian crosses the street, and kappa representing the time elasticity factor;

(3) and uniformly sampling the obtained time region segment to obtain a series of time sequence points:

{t-κ,t-κ+Δt’,t-κ+2Δt’,…,t-κ+(n-1)Δt’,t+κ}

wherein Δ t' is the time interval of uniform sampling;

(4) substituting the time series points into the position expression of the prediction track to generate a position series in the time region:

{P_t-κ,P_t-κ+Δt’,P_t-κ+2Δt’,…,P_t+κ-Δt’,P_t+κ}

(5) and deciding a collision avoidance strategy of the vehicle according to the number of the position sequence points in the three divided regions.

Further, the collision avoidance strategy of the vehicle is as follows:

1) if all the sequence points fall in the dangerous area, performing transverse collision avoidance operation;

2) if all the sequence points fall in the high risk area, performing longitudinal collision avoidance operation;

3) and if only one sequence point falls in the high risk area, performing longitudinal collision avoidance operation.

If the transverse collision avoidance operation is determined, the transverse planning layer plans a transverse collision avoidance path according to the state of the pedestrian relative to the vehicle; if the longitudinal collision avoidance operation is determined, an emergency braking or deceleration method is adopted to achieve the purpose of the longitudinal collision avoidance operation.

Further, the specific steps of planning the transverse collision avoidance path by the transverse planning layer are as follows:

(1) combining the predicted pedestrian track of S3 or S4, marking out a transverse collision avoidance path by adopting an artificial potential field rule; constructing an attractive force potential field, a road boundary repulsive force field and an elliptical obstacle repulsive force field:

(2) obtaining negative gradients of the attractive force potential field, the road boundary repulsive force field and the elliptical obstacle repulsive force field to obtain a potential field force corresponding to each potential field, and the method comprises the following steps:

and adding the three potential field forces to obtain a resultant force applied to the vehicle:

F_total＝F_alt+F_road+F_obs

in the formula of U_altRepresenting the gravitational potential field to which the vehicle is subjected, U_roadIndicating the road boundary repulsive potential field, U, experienced by the vehicle_obsRepresenting the repulsive potential field of an obstacle to which the vehicle is subjected, F_altDenotes the gravitational force, F_roadDenotes road boundary repulsion, F_obsDenotes the obstacle repulsion, κ_altRepresenting the gravitational potential field gain coefficient, X representing the real-time coordinates of the vehicle, X_gRepresenting the coordinates of the target point of the vehicle, k_roadRepresenting road boundary constraint coefficients, x representing the coordinates of the vehicle in the x-direction, y representing the coordinates of the vehicle in the y-direction, y_road,iRepresents the ordinate of the ith road side boundary, W represents the vehicle width, (x)_obs,y_obs) Coordinates representing obstacles, σ_xAnd σ_yIndicating the distance-influencing factor of the effect of the obstacle on the vehicle,

for representing gradient calculations;

(3) the position points where the vehicle moves under the action of resultant force can be obtained by the stress balance of the potential field force in the transverse direction, and the transverse planning path of the obstacle avoidance can be obtained by curve fitting the points;

(4) according to the position change of the pedestrian in the street crossing process, carrying out collision risk analysis on the vehicle and the pedestrian in real time, and judging whether a path needs to be re-planned or not so as to obtain a real-time transverse collision avoidance path; (ii) a The analysis of the collision risk of the vehicle and the pedestrian is carried out according to whether a position area where the pedestrian is likely to appear and a position area where the vehicle is likely to appear are overlapped in a planning period, if the position areas are overlapped, the situation that the collision risk exists is determined to require re-planning of a path, and if the position areas are not overlapped, the situation that the collision risk does not exist and the path does not need to be re-planned.

Further, the method for emergency braking or deceleration comprises the following specific steps:

(1) if emergency braking is adopted, a fuzzy controller is established, the relative speed and the relative distance between the vehicle and the pedestrian are used as input, and expected deceleration is output;

(2) if deceleration is adopted to avoid the pedestrian and the pedestrian normally runs after passing through the road, the expected minimum deceleration of the vehicle is obtained by the following formula:

wherein:

in the formula, t_passIndicating the time, L, at which the pedestrian crosses the road_pathIndicating the length of the road, y_pCoordinates in the Y direction, v, representing a pedestrian_p-yIndicating the speed, v, of the pedestrian in the Y direction_vehicleIndicating the speed, v, of the vehicle_pedRepresenting the speed of the pedestrian, deltas the longitudinal relative distance of the pedestrian from the vehicle, and a the desired minimum deceleration.

A pedestrian track prediction and collision avoidance control system considering human-vehicle interaction specifically comprises an environment sensing module, a pedestrian intention judgment module, a pedestrian track prediction module and a longitudinal and transverse collision avoidance decision module.

The environment perception module is used for acquiring the driving state information of the vehicle, the motion state information of pedestrians and the image in front of the vehicle, and is respectively connected with the pedestrian intention judgment module, the pedestrian track prediction module and the longitudinal and transverse collision avoidance decision module;

the pedestrian intention judging module detects the head of a pedestrian in the image in front of the vehicle by using the image processing unit according to the driving state information of the vehicle, the motion state information of the pedestrian and the image in front of the vehicle, which are acquired by the environment sensing module, so as to identify the face orientation of the pedestrian crossing the street; judging the intention of the pedestrian crossing the street based on the identification result of the face orientation of the pedestrian, and sending the intention judgment result to a pedestrian track prediction module;

the pedestrian trajectory prediction module receives the vehicle running state information, the pedestrian movement state information and the judgment result of the pedestrian intention judgment module sent by the environment sensing module, and predicts the trajectory of a pedestrian in a future preset time length by using a Markov pedestrian model for the pedestrian not paying attention to the vehicle; for pedestrians who notice vehicles and continue to walk, the movement of the pedestrian crossing the street is regarded as a combination of free movement and interference movement of the coming vehicles; introducing a social force model into the motion prediction of the pedestrian, and performing weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain a corrected pedestrian position; and obtaining a track curve within a preset time length according to the set time step length, and sending the predicted track curve to a longitudinal and transverse collision avoidance decision module.

And the longitudinal and transverse collision avoidance decision module receives the pedestrian track predicted by the pedestrian track prediction module, evaluates the feasibility of longitudinal collision avoidance and transverse collision avoidance through the safety state analysis of pedestrians and vehicles, and decides a proper collision avoidance strategy of the vehicles under the condition of ensuring the safety.

Further, the environment sensing module comprises a GPS, a speed sensor, a laser radar and a monocular camera which are installed on the vehicle, and is used for acquiring the position information and the speed information of the vehicle and the position information and the speed information of the pedestrian crossing relative to the vehicle in real time; the monocular camera captures images in front of the vehicle.

The invention has the beneficial effects that:

1. the invention adopts a mode of combining a Markov pedestrian model and a social force model, considers the combination of free motion and interference motion of coming vehicles when pedestrians cross the street, predicts the motion track of the pedestrians under the environment without external interference and the environment with human-vehicle interaction interference, improves the precision of track prediction and leads the prediction result to be more close to the real motion track of the pedestrians.

2. The invention designs a longitudinal and transverse collision avoidance selection strategy aiming at the predicted pedestrian track, which can adapt to the pedestrian collision avoidance requirement with behavior randomness, and ensures that a vehicle selects a proper collision avoidance strategy in advance under the safety, thereby reducing the occurrence of human-vehicle collision accidents and effectively improving the road safety.

Drawings

Fig. 1 is a block diagram of a pedestrian trajectory prediction and vehicle active collision avoidance system in consideration of human-vehicle interaction according to an embodiment of the present invention;

FIG. 2 is a flow chart of a pedestrian trajectory prediction and vehicle active collision avoidance method considering human-vehicle interaction according to an embodiment of the invention;

FIG. 3 is a schematic flow chart for identifying the orientation of a face of a pedestrian;

FIG. 4 is a schematic flow chart for pedestrian intent determination;

fig. 5 is a pedestrian risk region division diagram.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

For the analysis object, a straight line which is parallel to the right road direction and is taken as a vehicle center of mass point is taken as an X axis, a straight line which is perpendicular to the road direction and is taken as a Y axis, and the vehicle center of mass point is taken as a coordinate system origin. The positions of the vehicle and the pedestrian are expressed by coordinates within a coordinate system.

As shown in fig. 2, a flow chart of a pedestrian trajectory prediction and vehicle active collision avoidance method considering human-vehicle interaction is shown. The specific process is as follows:

and S1, acquiring the driving state information of the vehicle, the motion state information of the pedestrian and the front image of the vehicle.

And S2, extracting a pedestrian head region by using a Yolo algorithm according to the acquired image in front of the vehicle, establishing a convolutional neural network to identify the face orientation of the pedestrian, and judging whether the pedestrian notices the coming vehicle according to the face orientation of the pedestrian. The pedestrian motion state information, the pedestrian face orientation, and the vehicle motion state information are fused based on the logistic regression model, and the intention of the pedestrian is determined as part of the input of S4.

And S3, according to the face direction of the pedestrian, if the pedestrian is judged not to be noticed by the vehicle, predicting the track of the pedestrian in the future preset time length by using the Markov pedestrian model.

S4, according to the face orientation and intention judgment result of the pedestrian, if the pedestrian is judged to notice the vehicle and continue to walk, considering the movement of the pedestrian crossing the street as the combination of free interference-free movement and interference-free movement of the coming vehicle, introducing a social force model for the movement prediction of the pedestrian, and introducing a Markov pedestrian model for prediction in the movement process of the pedestrian; and carrying out weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain the corrected pedestrian position. And obtaining a track curve within a preset time length according to the set time step length.

And S5, judging the safety state according to the predicted pedestrian track, and deciding a proper collision avoidance strategy of the vehicle under the condition of ensuring the safety.

Further, in the above S3, the specific process of predicting the trajectory of the pedestrian in the future preset time period by using the markov pedestrian model includes:

the pedestrian does not notice the vehicle, the vehicle on the road can be regarded as the pedestrian of noninterference at this moment, the pedestrian crossing freely moves and accords with the Markov process under the condition without external disturbance, the pedestrian future position and speed depend on his present position and speed, therefore can get the state description of the pedestrian:

State_ped＝(x_p(t),y_p(t),v_x-p(t),v_y-p(t))

in the formula, State_pedIs a pedestrian state, v_x-p(t) and v_y-p(t) shows the speed of the pedestrian at time t in the X-direction and Y-direction, respectively, Δ v_x-pAnd Δ v_y-pRepresenting the velocity increase at time t, v, in the X and Y directions, respectively_x-p(t + Δ t) and v_y-p(t + Δ t) represents the speed of the pedestrian at the time t + Δ t in the X-direction and Y-direction, respectively, k_xAnd k_yThe number of the symbols representing the constant number,

and

respectively representing the desired speed, epsilon, of the pedestrian in the X-direction and in the Y-direction_xAnd ε_yRepresenting random disturbances, X, of the pedestrian's velocity increment in the X and Y directions, respectively_p(t) and y_p(t) represents the displacement function of the pedestrian in the X and Y directions with respect to time t, p_mAnd (t) represents the position of the pedestrian at t predicted by the Markov pedestrian model.

Further, in the above S4, the step of introducing the social force model into the motion prediction of the pedestrian when the pedestrian notices the vehicle and intends to walk facing the coming vehicle includes:

the pedestrian moves to the target point and has a driving force, the pedestrian can receive the repulsion force of the vehicle when facing the coming vehicle, the road can also apply a hidden boundary force to the pedestrian, the forces are accumulated to form a resultant force, and the position of the pedestrian is recurred along with the time step under the action of the social force. The expression formula of the resultant force and the position of the pedestrian is as follows:

in the formula, F_ped(t) social force resultant force to which pedestrian is subjected, F_dRepresenting the driving force received by the pedestrian moving towards the target point, F_vpIndicating the repulsive force of the pedestrian facing the incoming vehicle, F_eIndicating that the road is applying a hidden boundary force to the pedestrian. v. of_n(t) represents the speed of the pedestrian at t, p_s(t) the position of the pedestrian at t predicted by the social force model, p_s(t + Δ t) represents the position of the pedestrian at t + Δ t, Δ t represents a time step, and m represents the mass of the pedestrian.

Further, in S4, the predicted pedestrian position is obtained by weighting the pedestrian position predicted by the markov pedestrian model and the pedestrian position predicted by the social force model, taking the motion of the pedestrian crossing the street into consideration as a combination of the free disturbance-free motion and the incoming disturbance motion, and is expressed by the following equation:

p(t)＝τ_m·p_m(t)+τ_s·p_s(t)

wherein p (t) represents the pedestrian position at t obtained by fusing and correcting the Markov pedestrian model and the social force model, and τ_mAnd τ_sRespectively represents p_m(t) and p_s(t) weight coefficient.

In the process of crossing the street, the pedestrian is in an interference-free state when the pedestrian does not notice the vehicle, the change of the movement direction of the pedestrian is small, the social force model considers the external environment to enable the predicted track to have a certain error, and the prediction result of the Markov pedestrian model is better than that of the social force model, so that the Markov pedestrian model is directly adopted to model the movement of the pedestrian at the moment. The pedestrian notices that the motion state of the incoming vehicle changes, the social force model is utilized to consider the influence of the vehicle on the pedestrian, and the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model are added by weight to obtain the corrected pedestrian position. And respectively obtaining the predicted track curve of the pedestrian who does not notice the vehicle and the predicted track curve of the pedestrian who notices the vehicle and continues to walk within the preset time length according to the set time step.

As shown in fig. 3, a flow chart for recognizing the orientation of a face is illustrated. The image in front of the vehicle collected by the monocular camera is used as an input image, and the head area of the pedestrian is detected by using a Yolo algorithm. And constructing a pedestrian face orientation classifier by using a convolutional neural network to identify a face orientation state for the detected head region of the pedestrian. The method comprises the following specific steps:

(1) and (3) setting a Yolo network structure for adjusting training parameters according to the pedestrian image acquired by the monocular camera, and detecting the head area of the pedestrian by adopting a Yolo algorithm to obtain a head image.

(2) In order to detect whether a pedestrian notices an incoming vehicle, it is necessary to recognize the face orientation of the pedestrian. A forward/lateral/backward classifier of the face of the pedestrian is built by adopting a convolutional neural network, and the classifier adopts four convolutional layers and carries out convolution operation with convolution kernels of 2 multiplied by 2, 3 multiplied by 3, 5 multiplied by 5 and 7 multiplied by 7. The Normalization and activation operations were performed using Batch Normalization (Batch Normalization) and the ReLU activation functions. Redundant information in the feature map obtained by convolution is relieved through the pooling operation of the maximum pooling layer with the step size of 3 multiplied by 3 and 2. And (3) replacing a full-connection layer with a 2 x 2 convolutional layer, classifying the output by adopting a Softmax function, calculating the probability of each orientation, and finally determining the face orientation of the pedestrian. The probability expression formula for calculating the class of the image as j is as follows:

where ζ represents a parameter of the network structure, f_iIndicating that the feature of the ith image was learned.

As shown in fig. 4, a flow chart for pedestrian intention judgment is shown. With four factors of pedestrian speed, pedestrian face orientation, car speed, the distance between pedestrian and the vehicle as independent variables, the pedestrian selects whether to stop or not to walk as the dependent variable in the face of the incoming vehicle in the human-vehicle interaction process, utilizes the logistic regression model to identify the behavior of the pedestrian crossing, specifically:

(1) at pedestrian speed v_pedOrientation of pedestrian face, vehicle speed v_vehicleThe Distance between the pedestrian and the vehicle is an independent variable, the intention of the pedestrian to select walking or stopping facing the coming vehicle is a dependent variable in the human-vehicle interaction process, and a function expression in a logistic regression model for judging the intention of the pedestrian is as follows:

wherein the independent variable x is [ c, v ]_ped,Orientation,v_vehicle,Distance]^TAnd c is any constant term. Theta is ═ theta₀,θ₁,θ₂,θ₃,θ₄]^TIs a set of coefficients, θ_iThe coefficient i corresponding to the ith argument is 0, 1, 2, 3, 4. In this embodiment, the minimum value of the cost function is obtained by a gradient descent method to obtain the coefficient set θ. The cost function is expressed as follows:

wherein m is the number of samples, H_θFor assumed values, y is the true value.

(2) Use of a trained logistic regression model for determining the intent of a pedestrian, H_θ> 0.5 indicates pedestrian walking, H_θ< 0.5 indicates that the pedestrian stops.

FIG. 5 is a pedestrian hazard zone division view, the zone from the right side edge of the vehicle to the right side edge of the road being a hazard zone; the area directly in front of the vehicle may be considered a high risk area; other areas on the road may be considered safe areas. And determining that the vehicle selects a proper collision avoidance strategy under the condition of ensuring the safety through the region position of the pedestrian on the road and the safety state judgment between the pedestrian and the vehicle. The designed longitudinal and transverse collision avoidance selection strategy comprises the following steps:

(1) the position of the pedestrian on the road is divided into a dangerous area, a high risk area and a safe area.

(2) And calculating the longitudinal collision avoidance time of the vehicle. Considering the position fluctuation of the pedestrian position in the longitudinal direction, the obtained longitudinal collision avoidance time of the vehicle is a time zone range:

in the formula, t_vIndicating longitudinal time to collision, v, of the vehicle_vehicleRepresenting the vehicle speed, as representing the longitudinal relative distance between the vehicle and the pedestrian, epsilon representing the longitudinal offset of the pedestrian crossing the street, and kappa representing the time elasticity factor.

(3) And uniformly sampling the obtained time region segment to obtain a series of time sequence points:

{t-κ,t-κ+Δt’,t-κ+2Δt’,…,t-κ+(n-1)Δt’,t+κ}

where Δ t' is the time interval of uniform sampling.

(4) Substituting the time series points into the position expression of the prediction track to generate a position series in the time region:

{P_t-κ,P_t-κ+Δt’,P_t-κ+2Δt’,…,P_t+κ-Δt’,P_t+κ}

(5) and deciding a collision avoidance strategy of the vehicle according to the number of the position sequence points in the three divided regions.

1) If all the sequence points fall in the dangerous area, performing transverse collision avoidance operation;

2) if all the sequence points fall in the high risk area, performing longitudinal collision avoidance operation;

3) and if only one sequence point falls in the high risk area, performing longitudinal collision avoidance operation.

If the transverse collision avoidance operation is determined, the transverse planning layer plans a transverse collision avoidance path according to the state of the pedestrian relative to the vehicle; if the longitudinal collision avoidance operation is determined, an emergency braking or deceleration method is adopted to achieve the purpose of the longitudinal collision avoidance operation.

More specifically, the specific steps of the transverse planning layer for planning the transverse collision avoidance path are as follows:

(1) combining the predicted pedestrian track of S3 or S4, marking out a transverse collision avoidance path by adopting an artificial potential field rule; constructing an attractive force potential field, a road boundary repulsive force field and an elliptical obstacle repulsive force field:

(2) obtaining negative gradients of the attractive force potential field, the road boundary repulsive force field and the elliptical obstacle repulsive force field to obtain a potential field force corresponding to each potential field, and the method comprises the following steps:

and adding the three potential field forces to obtain a resultant force applied to the vehicle:

F_total＝F_alt+F_road+F_obs

in the formula of U_altRepresenting the gravitational potential field to which the vehicle is subjected, U_roadIndicating the road boundary repulsive potential field, U, experienced by the vehicle_obsRepresenting the repulsive potential field of an obstacle to which the vehicle is subjected, F_altDenotes the gravitational force, F_roadDenotes road boundary repulsion, F_obsDenotes the obstacle repulsion, κ_altRepresenting the gravitational potential field gain coefficient, X representing the real-time coordinates of the vehicle, X_gRepresenting the coordinates of the target point of the vehicle, k_roadRepresenting road boundary constraint coefficients, x representing the coordinates of the vehicle in the x-direction, y representing the coordinates of the vehicle in the y-direction, y_road,iRepresents the ordinate of the ith road side boundary, W represents the vehicle width, (x)_obs,y_obs) Coordinates representing obstacles, σ_xAnd σ_yIndicating the distance-influencing factor of the effect of the obstacle on the vehicle,

for representing gradient calculations;

(3) the position points where the vehicle moves under the action of resultant force can be obtained by the stress balance of the potential field force in the transverse direction, and the transverse planning path of the obstacle avoidance can be obtained by curve fitting the points;

(4) according to the position change of the pedestrian in the street crossing process, carrying out collision risk analysis on the vehicle and the pedestrian in real time, and judging whether a path needs to be re-planned or not so as to obtain a real-time transverse collision avoidance path; in this embodiment, the analysis of the risk of collision between the vehicle and the pedestrian is performed according to whether the location area where the pedestrian may appear and the location area where the vehicle may appear overlap in the planning period, and if the overlap exists, it is determined that the risk of collision exists and the path needs to be re-planned, otherwise, it is determined that the risk of collision does not exist and the path does not need to be re-planned.

More specifically, the method for emergency braking or deceleration comprises the following specific steps:

(1) if emergency braking is adopted, a fuzzy controller is established, the relative speed and the relative distance between the vehicle and the pedestrian are used as input, and expected deceleration is output;

(2) if deceleration is adopted to avoid the pedestrian and the pedestrian normally runs after passing through the road, the expected minimum deceleration of the vehicle is obtained by the following formula:

wherein:

in the formula, t_passIndicating the time, L, at which the pedestrian crosses the road_pathIndicating the length of the road, y_pCoordinates in the Y direction, v, representing a pedestrian_p-yIndicating the speed, v, of the pedestrian in the Y direction_vehicleIndicating the speed, v, of the vehicle_pedRepresenting the speed of the pedestrian, deltas the longitudinal relative distance of the pedestrian from the vehicle, and a the desired minimum deceleration.

In order to realize the pedestrian track prediction and vehicle active collision avoidance method considering human-vehicle interaction, the application also designs a pedestrian track prediction and vehicle active collision avoidance system considering human-vehicle interaction as shown in fig. 1, and the system specifically comprises an environment sensing module, a pedestrian intention judgment module, a pedestrian track prediction module and a longitudinal and transverse collision avoidance decision module.

The environment perception module is used for acquiring the running state information (vehicle speed v) of the vehicle_vehicleVehicle position), pedestrian movement state information (pedestrian velocity v)_pedPedestrian position) and a vehicle front image, and the environment sensing module is respectively connected with the pedestrian intention judging module, the pedestrian track predicting module and the longitudinal and transverse collision avoidance decision module. The environment sensing module comprises a GPS, a speed sensor, a laser radar and a monocular camera which are arranged on the vehicle to acquire the vehicle in real timeThe pedestrian-vehicle Distance is obtained based on the position information of the vehicle and the position information of the pedestrian crossing; the monocular camera captures images in front of the vehicle.

The pedestrian intention judging module detects the head of a pedestrian in the image in front of the vehicle by using the image processing unit according to the driving state information of the vehicle, the motion state information of the pedestrian and the image in front of the vehicle, which are acquired by the environment sensing module, so as to identify the face orientation of the pedestrian crossing the street, and when the face of the pedestrian faces the coming vehicle, the pedestrian is considered to notice the coming vehicle; otherwise, the pedestrian is considered not to be noticed by the coming vehicle.

The pedestrian intention judging module judges the intention of the pedestrian crossing the street based on the recognition result of the face orientation of the pedestrian, and sends the intention judging result to the pedestrian track predicting module; the specific process of judging the intention of the pedestrian is as follows:

(1) at pedestrian speed v_pedOrientation of pedestrian face, vehicle speed v_vehicleThe Distance between the pedestrian and the vehicle is an independent variable, the intention of the pedestrian to select walking or stopping facing the coming vehicle is a dependent variable in the human-vehicle interaction process, and a function expression in a logistic regression model for judging the intention of the pedestrian is as follows:

wherein the independent variable x is [ c, v ]_ped,Orientation,v_vehicle,Distance]^TAnd c is any constant term. Theta is ═ theta₀,θ₁,θ₂,θ₃,θ₄]^TIs a set of coefficients, θ_iIs the coefficient corresponding to the ith argument, i ═ 0, 1, 2, 3, 4. In this embodiment, the minimum value of the cost function is obtained by a gradient descent method to obtain the coefficient set θ. The cost function is expressed as follows:

where J (θ) is the cost function, m is the number of samples, H_θTo assume the value, y⁽ⁱ⁾Is the ith true value, x⁽ⁱ⁾Is the ith independent variable, x⁽ⁱ⁾∈x。

(2) The actually detected x ═ c, v_ped,Orientation,v_vehicle,Distance]^TInputting the trained logistic regression model to judge the intention of the pedestrian, and if H is the result, judging the intention of the pedestrian_θ> 0.5 indicates pedestrian walking, H_θ< 0.5 indicates that the pedestrian stops.

The pedestrian trajectory prediction module receives the vehicle running state information, the pedestrian movement state information and the judgment result of the pedestrian intention judgment module sent by the environment sensing module, and predicts the trajectory of a pedestrian in a future preset time length by using a Markov pedestrian model for the pedestrian not paying attention to the vehicle; for noticing the vehicle and continuing to walk (i.e. H)_θ> 0.5), the pedestrian's movement across the street is considered a combination of free movement and interfering movement of incoming vehicles. Introducing a social force model into the motion prediction of the pedestrian, and performing weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain a corrected pedestrian position; and obtaining a track curve within a preset time length according to the set time step length, and sending the predicted track curve to a longitudinal and transverse collision avoidance decision module.

In the pedestrian trajectory prediction module, for a pedestrian not paying attention to the vehicle, the future position and speed of the pedestrian depend on the current position and speed of the pedestrian, so that the state description of the pedestrian can be obtained:

State_ped＝(x_p(t),y_p(t),v_x-p(t),v_y-p(t))

in the formula, State_pedIs a pedestrian state; v. v_x-p(t) and v_y-p(t) shows the speed of the pedestrian at time t in the X-direction and Y-direction, respectively, Δ v_x-pAnd Δ v_y-pRepresenting the velocity increase at time t, v, in the X and Y directions, respectively_x-p(t + Δ t) and v_y-p(t + Δ t) represents the speed of the pedestrian at the time t + Δ t in the X-direction and Y-direction, respectively, k_xAnd k_yThe number of the symbols representing the constant number,

and

respectively representing the desired speed, epsilon, of the pedestrian in the X-direction and in the Y-direction_xAnd ε_yRepresenting random disturbances, X, of the pedestrian's velocity increment in the X and Y directions, respectively_p(t) and y_p(t) represents the displacement function of the pedestrian in the X and Y directions with respect to time t, p_mAnd (t) represents the position of the pedestrian at t predicted by the Markov pedestrian model.

The pedestrian notices the vehicle and intends to walk facing the coming vehicle, and introduces a social force model for the motion prediction of the pedestrian at the moment, and the specific steps comprise:

the pedestrian moves to the target point and has a driving force, the pedestrian can receive the repulsion force of the vehicle when facing the coming vehicle, the road can also apply a hidden boundary force to the pedestrian, the forces are accumulated to form a resultant force, and the position of the pedestrian is recurred along with the time step under the action of the social force. The expression formula of the resultant force and the position of the pedestrian is as follows:

in the formula, F_ped(t) social force resultant force to which pedestrian is subjected, F_dRepresenting the driving force received by the pedestrian moving towards the target point, F_vpIndicating the repulsive force of the pedestrian facing the incoming vehicle, F_eIndicating that the road is applying a hidden boundary force to the pedestrian. v. of_n(t) represents the speed of the pedestrian at t, p_s(t) the position of the pedestrian at t predicted by the social force model, p_s(t + Δ t) represents the position of the pedestrian at t + Δ t, Δ t represents a time step, and m represents the mass of the pedestrian.

Regarding pedestrians who pay attention to vehicles and continue to walk, regarding the movement of the pedestrian crossing the street as the combination of free movement and interference movement of the coming vehicles, and performing weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain a corrected pedestrian position, which is expressed by the following formula:

p(t)＝τ_m·p_m(t)+τ_s·p_s(t)

wherein p (t) represents the pedestrian position at t obtained by fusing and correcting the Markov pedestrian model and the social force model, and τ_mAnd τ_sRespectively represents p_m(t) and p_s(t) weight coefficient.

And respectively obtaining the predicted track curve of the pedestrian who does not notice the vehicle and the predicted track curve of the pedestrian who notices the vehicle and continues to walk within the preset time length according to the set time step.

And the longitudinal and transverse collision avoidance decision module receives the pedestrian track predicted by the pedestrian track prediction module, evaluates the feasibility of longitudinal collision avoidance and transverse collision avoidance through the safety state analysis of pedestrians and vehicles, and decides a proper collision avoidance strategy of the vehicles under the condition of ensuring the safety. The longitudinal and transverse collision avoidance selection strategy in the longitudinal and transverse collision avoidance decision module comprises the following steps:

(1) dividing the position of the pedestrian on the road into a dangerous area, a high risk area and a safe area; FIG. 5 is a pedestrian hazard zone division view, the zone from the right side edge of the vehicle to the right side edge of the road being a hazard zone; the area directly in front of the vehicle may be considered a high risk area; other areas on the road may be considered safe areas.

(2) And calculating the longitudinal collision avoidance time of the vehicle. Considering the position fluctuation of the pedestrian position in the longitudinal direction, the obtained longitudinal collision avoidance time of the vehicle is a time zone range:

in the formula, t_vIndicating longitudinal time to collision, v, of the vehicle_vehicleRepresenting the vehicle speed, as representing the longitudinal relative distance between the vehicle and the pedestrian, epsilon representing the longitudinal offset of the pedestrian crossing the street, and kappa representing the time elasticity factor.

(3) And uniformly sampling the obtained time region segment to obtain a series of time sequence points:

{t-κ,t-κ+Δt’,t-κ+2Δt’,…,t-κ+(n-1)Δt’,t+κ}

where Δ t' is the time interval of uniform sampling.

(4) Substituting the time series points into the position expression of the prediction track to generate a position series in the time region:

{P_t-κ,P_t-κ+Δt’,P_t-κ+2Δt’,…,P_t+κ-Δt’,P_t+κ}

(5) according to the number of the position sequence points in three divided areas (a dangerous area, a high-risk area and a safe area), deciding a collision avoidance strategy of the vehicle:

1) if all the sequence points fall in the dangerous area, performing transverse collision avoidance operation;

2) if all the sequence points fall in the high risk area, performing longitudinal collision avoidance operation;

3) and if only one sequence point falls in the high risk area, performing longitudinal collision avoidance operation.

If the transverse collision avoidance operation is determined, the transverse planning layer plans a transverse collision avoidance path according to the state of the pedestrian relative to the vehicle; if the longitudinal collision avoidance operation is determined, an emergency braking or deceleration method is adopted to achieve the purpose of the longitudinal collision avoidance operation.

More specifically, the specific steps of the transverse planning layer for planning the transverse collision avoidance path are as follows:

(1) combining the predicted pedestrian track of S3 or S4, marking out a transverse collision avoidance path by adopting an artificial potential field rule; constructing an attractive force potential field, a road boundary repulsive force field and an elliptical obstacle repulsive force field:

(2) obtaining negative gradients of the attractive force potential field, the road boundary repulsive force field and the elliptical obstacle repulsive force field to obtain a potential field force corresponding to each potential field, and the method comprises the following steps:

and adding the three potential field forces to obtain a resultant force applied to the vehicle:

F_total＝F_alt+F_road+F_obs

in the formula of U_altRepresenting the gravitational potential field to which the vehicle is subjected, U_roadIndicating the road boundary repulsive potential field to which the vehicle is subjected，U_obsRepresenting the repulsive potential field of an obstacle to which the vehicle is subjected, F_altDenotes the gravitational force, F_roadDenotes road boundary repulsion, F_obsDenotes the obstacle repulsion, κ_altRepresenting the gravitational potential field gain coefficient, X representing the real-time coordinates of the vehicle, X_gRepresenting the coordinates of the target point of the vehicle, k_roadRepresenting road boundary constraint coefficients, x representing the coordinates of the vehicle in the x-direction, y representing the coordinates of the vehicle in the y-direction, y_road,iRepresents the ordinate of the ith road side boundary, W represents the vehicle width, (x)_obs,y_obs) Coordinates representing obstacles, σ_xAnd σ_yIndicating the distance-influencing factor of the effect of the obstacle on the vehicle,

for representing gradient calculations;

(3) the position points where the vehicle moves under the action of resultant force can be obtained by the stress balance of the potential field force in the transverse direction, and the transverse planning path of the obstacle avoidance can be obtained by curve fitting the points;

(4) according to the position change of the pedestrian in the street crossing process, collision risk analysis is carried out on the vehicle and the pedestrian in real time, whether the path needs to be re-planned or not is judged, and therefore the real-time transverse collision avoidance path is obtained. In this embodiment, the analysis of the risk of collision between the vehicle and the pedestrian is performed according to whether the location area where the pedestrian may appear and the location area where the vehicle may appear overlap in the planning period, and if the overlap exists, it is determined that the risk of collision exists and the path needs to be re-planned, otherwise, it is determined that the risk of collision does not exist and the path does not need to be re-planned.

More specifically, the method for emergency braking or deceleration comprises the following specific steps:

(1) if emergency braking is adopted, a fuzzy controller is established, the relative speed and the relative distance between the vehicle and the pedestrian are used as input, and expected deceleration is output;

(2) if deceleration is adopted to avoid the pedestrian and the pedestrian normally runs after passing through the road, the expected minimum deceleration of the vehicle is obtained by the following formula:

wherein:

in the formula, t_passIndicating the time, L, at which the pedestrian crosses the road_pathIndicating the length of the road, y_pCoordinates in the Y direction, v, representing a pedestrian_p-yIndicating the speed, v, of the pedestrian in the Y direction_vehicleIndicating the speed, v, of the vehicle_pedRepresenting the speed of the pedestrian, deltas the longitudinal relative distance of the pedestrian from the vehicle, and a the desired minimum deceleration.

The above embodiments are only used for illustrating the design idea and features of the present invention, and the purpose of the present invention is to enable those skilled in the art to understand the content of the present invention and implement the present invention accordingly, and the protection scope of the present invention is not limited to the above embodiments. Therefore, all equivalent changes and modifications made in accordance with the principles and concepts disclosed herein are intended to be included within the scope of the present invention.

Claims (10)

1. A pedestrian track prediction and active collision avoidance method considering human-vehicle interaction is characterized by comprising the following steps:

s1, acquiring the driving state information of the vehicle, the motion state information of the pedestrian and the front image of the vehicle;

s2, extracting a head region of the pedestrian and identifying the face direction of the pedestrian according to the acquired image in front of the vehicle, and judging whether the pedestrian notices the coming vehicle according to the face direction of the pedestrian; fusing pedestrian motion state information, pedestrian face orientation and vehicle motion state information based on a logistic regression model, and judging the intention of the pedestrian;

s3, according to the face orientation of the pedestrian, if the pedestrian is judged not to notice the vehicle, predicting the track of the pedestrian in a future preset time length by using a Markov pedestrian model; the process of predicting the track of the pedestrian in the future preset time length by using the Markov pedestrian model is as follows:

the pedestrian does not notice the vehicle, the vehicle on the road can be regarded as the pedestrian of noninterference at this moment, the pedestrian crossing freely moves and accords with the Markov process under the condition without external disturbance, the pedestrian future position and speed depend on his present position and speed, therefore can get the state description of the pedestrian:

State_ped＝(x_p(t),y_p(t),v_x-p(t),v_y-p(t))

in the formula, State_pedIs a pedestrian state, v_x-p(t) and v_y-p(t) shows the speed of the pedestrian at time t in the X-direction and Y-direction, respectively, Δ v_x-pAnd Δ v_y-pRepresenting the velocity increase at time t, v, in the X and Y directions, respectively_x-p(t + Δ t) and v_y-p(t + Δ t) represents the speed of the pedestrian at the time t + Δ t in the X-direction and Y-direction, respectively, k_xAnd k_yThe number of the symbols representing the constant number,

and

respectively representing the desired speed, epsilon, of the pedestrian in the X-direction and in the Y-direction_xAnd ε_yRepresenting random disturbances, X, of the pedestrian's velocity increment in the X and Y directions, respectively_p(t) and y_p(t) represents the displacement function of the pedestrian in the X and Y directions with respect to time t, p_m(t) represents the position of the pedestrian at t predicted by the Markov pedestrian model;

s4, according to the face direction and intention judgment result of the pedestrian, if the pedestrian is judged to pay attention to the vehicle and continue walking, introducing a social force model to predict the movement of the pedestrian; and a Markov pedestrian model is introduced to predict in the motion process of the pedestrians; carrying out weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain a corrected pedestrian position; obtaining a track curve within a preset time length according to a set time step; the method for predicting the motion of the pedestrian by utilizing the social force model comprises the following specific steps:

the pedestrian moves to the target point and has a driving force, the pedestrian can be subjected to the repulsive force of the vehicle when facing the coming vehicle, the road can also apply a hidden boundary force to the pedestrian, the forces are accumulated to form a resultant force, and the position of the pedestrian is recurred along with the time step under the action of social force; the expression formula of the resultant force and the position of the pedestrian is as follows:

in the formula, F_ped(t) social force resultant force to which pedestrian is subjected, F_dRepresenting the driving force received by the pedestrian moving towards the target point, F_vpIndicating the repulsive force of the pedestrian facing the incoming vehicle, F_eIndicating that the road is applying a hidden boundary force to the pedestrian. v. of_n(t) represents the speed of the pedestrian at t, p_s(t) the position of the pedestrian at t predicted by the social force model, p_s(t + Δ t) represents the position of the pedestrian at t + Δ t, Δ t represents a time step, and m represents the mass of the pedestrian;

and S5, judging the safety state according to the predicted pedestrian track, and deciding a proper collision avoidance strategy of the vehicle under the condition of ensuring the safety.

2. The method of claim 1, wherein in step S2, the step of identifying the face orientation of the pedestrian comprises:

(1) detecting the head area of the pedestrian by adopting a Yolo algorithm according to the image in front of the vehicle to obtain a head image;

(2) and (3) adopting a convolutional neural network to build a forward/lateral/backward classifier of the face of the pedestrian, and determining the face Orientation of the pedestrian.

3. The method as claimed in claim 2, wherein in S2, the logistic regression model for pedestrian intention determination comprises the following steps:

(1) at pedestrian speed v_pedOrientation of pedestrian face, vehicle speed v_vehicleThe Distance between the pedestrian and the vehicle is an independent variable, the intention of the pedestrian to select walking or stopping facing the coming vehicle is a dependent variable in the human-vehicle interaction process, and a function expression in a logistic regression model for judging the intention of the pedestrian is as follows:

wherein the independent variable is x ═ c, v_ped,Orientation,v_vehicle,Distance]^TC is any constant term; theta is ═ theta₀,θ₁,θ₂,θ₃,θ₄]^TIs a set of coefficients, θ_iIs the coefficient corresponding to the ith argument, i ═ 0, 1, 2, 3, 4; determining the minimum value theta of the cost function by means of a gradient descent method_iIs the coefficient corresponding to the ith argument; the expression formula of the cost function is as follows:

where J (θ) is the cost function, m is the number of samples, H_θTo assume the value, y⁽ⁱ⁾Is the ith true value, x⁽ⁱ⁾Is the ith independent variable, x⁽ⁱ⁾∈x；

(2) Judging the intention of the pedestrian by using the trained logistic regression model, and if H is the intention of the pedestrian, judging the intention of the pedestrian_θ> 0.5 indicates pedestrian walking, H_θ< 0.5 indicates that the pedestrian stops.

4. The pedestrian trajectory prediction and active collision avoidance method considering human-vehicle interaction according to claim 1, wherein the pedestrian crossing motion at that time is considered as a combination of free interference-free motion and incoming vehicle interference motion, and the pedestrian position predicted by the markov pedestrian model and the pedestrian position predicted by the social force model are weighted to obtain the predicted pedestrian position, which is represented by the following formula:

p(t)＝τ_m·p_m(t)+τ_s·p_s(t)

wherein p (t) represents the pedestrian position at t obtained by fusing and correcting the Markov pedestrian model and the social force model, and τ_mAnd τ_sRespectively represents p_m(t) and p_s(t) weight coefficient.

5. The method for pedestrian trajectory prediction and active collision avoidance considering human-vehicle interaction as claimed in claim 1, wherein in S5, the method for determining the safety status of the pedestrian and the vehicle comprises:

(1) dividing the position of the pedestrian on the road into a dangerous area, a high risk area and a safe area;

(2) calculating the longitudinal collision avoidance time of the vehicle, and considering the position fluctuation of the pedestrian position in the longitudinal direction, wherein the obtained longitudinal collision avoidance time of the vehicle is a time region range:

in the formula, t_vIndicating longitudinal time to collision, v, of the vehicle_vehicleRepresenting the vehicle speed, deltas representing the longitudinal relative distance between the vehicle and the pedestrian, epsilon representing the longitudinal offset when the pedestrian crosses the street, and kappa representing the time elasticity factor;

(3) and uniformly sampling the obtained time region segment to obtain a series of time sequence points:

{t-κ,t-κ+Δt’,t-κ+2Δt’,…,t-κ+(n-1)Δt’,t+κ}

wherein Δ t' is the time interval of uniform sampling;

(4) substituting the time series points into the position expression of the prediction track to generate a position series in the time region:

{P_t-κ,P_t-κ+Δt，,P_t-κ+2Δt，,…,P_t+κ-Δt，,P_t+κ}

(5) and deciding a collision avoidance strategy of the vehicle according to the number of the position sequence points in the three divided regions.

6. The pedestrian trajectory prediction and active collision avoidance method considering human-vehicle interaction according to claim 5, wherein the collision avoidance policy of the vehicle is:

1) if all the sequence points fall in the dangerous area, performing transverse collision avoidance operation;

2) if all the sequence points fall in the high risk area, performing longitudinal collision avoidance operation;

3) if only one sequence point falls in the high risk area, longitudinal collision avoidance operation is carried out;

if the transverse collision avoidance operation is determined, the transverse planning layer plans a transverse collision avoidance path according to the state of the pedestrian relative to the vehicle; if the longitudinal collision avoidance operation is determined, an emergency braking or deceleration method is adopted to achieve the purpose of the longitudinal collision avoidance operation.

7. The pedestrian trajectory prediction and active collision avoidance method considering human-vehicle interaction according to claim 6, wherein the specific steps of the transverse planning layer to plan the transverse collision avoidance path are as follows:

(1) combining the predicted pedestrian track of S3 or S4, marking out a transverse collision avoidance path by adopting an artificial potential field rule; constructing an attractive force potential field, a road boundary repulsive force field and an elliptical obstacle repulsive force field:

(2) obtaining negative gradients of the attractive force potential field, the road boundary repulsive force field and the elliptical obstacle repulsive force field to obtain a potential field force corresponding to each potential field, and the method comprises the following steps:

and adding the three potential field forces to obtain a resultant force applied to the vehicle:

F_total＝F_alt+F_road+F_obs

in the formula of U_altRepresenting the gravitational potential field to which the vehicle is subjected, U_roadIndicating the road boundary repulsive potential field, U, experienced by the vehicle_obsRepresenting the repulsive potential field of an obstacle to which the vehicle is subjected, F_altDenotes the gravitational force, F_roadDenotes road boundary repulsion, F_obsDenotes the obstacle repulsion, κ_altRepresenting the gravitational potential field gain coefficient, X representing the real-time coordinates of the vehicle, X_gRepresenting the coordinates of the target point of the vehicle, k_roadRepresenting road boundary constraint coefficients, x representing the coordinates of the vehicle in the x-direction, y representing the coordinates of the vehicle in the y-direction, y_road,iRepresents the ordinate of the ith road side boundary, W represents the vehicle width, (x)_obs,y_obs) Coordinates representing obstacles, σ_xAnd σ_yIndicating the distance-influencing factor of the effect of the obstacle on the vehicle,

for representing gradient calculations;

(3) the position points where the vehicle moves under the action of resultant force can be obtained by the stress balance of the potential field force in the transverse direction, and the transverse planning path of the obstacle avoidance can be obtained by curve fitting the points;

(4) according to the position change of the pedestrian in the street crossing process, carrying out collision risk analysis on the vehicle and the pedestrian in real time, and judging whether a path needs to be re-planned or not so as to obtain a real-time transverse collision avoidance path; the analysis of the collision risk of the vehicle and the pedestrian is carried out according to whether a position area where the pedestrian is likely to appear and a position area where the vehicle is likely to appear are overlapped in a planning period, if the position areas are overlapped, the situation that the collision risk exists is determined to require re-planning of a path, and if the position areas are not overlapped, the situation that the collision risk does not exist and the path does not need to be re-planned.

8. The pedestrian trajectory prediction and active collision avoidance method considering human-vehicle interaction according to claim 6, wherein the method of emergency braking or deceleration comprises the following specific steps:

(1) if emergency braking is adopted, a fuzzy controller is established, the relative speed and the relative distance between the vehicle and the pedestrian are used as input, and expected deceleration is output;

(2) if deceleration is adopted to avoid the pedestrian and the pedestrian normally runs after passing through the road, the expected minimum deceleration of the vehicle is obtained by the following formula:

wherein:

in the formula, t_passIndicating the time, L, at which the pedestrian crosses the road_pathIndicating the length of the road, y_pCoordinates in the Y direction, v, representing a pedestrian_p-yIndicating the speed, v, of the pedestrian in the Y direction_vehicleIndicating the speed, v, of the vehicle_pedRepresenting the speed of the pedestrian, deltas the longitudinal relative distance of the pedestrian from the vehicle, and a the desired minimum deceleration.

9. A pedestrian track prediction and active collision avoidance system considering human-vehicle interaction is characterized by specifically comprising an environment sensing module, a pedestrian intention judging module, a pedestrian track prediction module and a longitudinal and transverse collision avoidance decision module;

the environment perception module is used for acquiring the driving state information of the vehicle, the motion state information of pedestrians and the image in front of the vehicle, and is respectively connected with the pedestrian intention judgment module, the pedestrian track prediction module and the longitudinal and transverse collision avoidance decision module;

the pedestrian intention judging module detects the head of a pedestrian in the image in front of the vehicle by using the image processing unit according to the driving state information of the vehicle, the motion state information of the pedestrian and the image in front of the vehicle, which are acquired by the environment sensing module, so as to identify the face orientation of the pedestrian crossing the street; judging the intention of the pedestrian crossing the street based on the identification result of the face orientation of the pedestrian, and sending the intention judgment result to a pedestrian track prediction module;

the pedestrian trajectory prediction module receives the vehicle running state information, the pedestrian movement state information and the judgment result of the pedestrian intention judgment module sent by the environment sensing module, and predicts the trajectory of a pedestrian in a future preset time length by using a Markov pedestrian model for the pedestrian not paying attention to the vehicle; for pedestrians who notice vehicles and continue to walk, the movement of the pedestrian crossing the street is regarded as a combination of free movement and interference movement of the coming vehicles; introducing a social force model into the motion prediction of the pedestrian, and performing weighted fusion on the pedestrian position predicted by the Markov pedestrian model and the pedestrian position predicted by the social force model to obtain a corrected pedestrian position; obtaining a track curve within a preset time length according to a set time step, and sending the predicted track curve to a longitudinal and transverse collision avoidance decision module;

and the longitudinal and transverse collision avoidance decision module receives the pedestrian track predicted by the pedestrian track prediction module, evaluates the feasibility of longitudinal collision avoidance and transverse collision avoidance through the safety state analysis of pedestrians and vehicles, and decides a proper collision avoidance strategy of the vehicles under the condition of ensuring the safety.

10. The system for pedestrian trajectory prediction and active collision avoidance considering human-vehicle interaction as claimed in claim 9, wherein the environment sensing module comprises a GPS, a speed sensor, a lidar and a monocular camera mounted on the vehicle, and obtains the position information and the speed information of the vehicle and the position information and the speed information of the pedestrian crossing the street relative to the vehicle in real time; the monocular camera captures images in front of the vehicle.

Images