A vehicle active collision avoidance control method based on model predictive control algorithm

Abstract

The present invention discloses a vehicle active collision avoidance control method based on a model predictive control algorithm, comprising the following steps: S1, collecting controlled vehicle information, vehicle position information and global path information; S2, establishing an obstacle avoidance function; S3, establishing a nonlinear vehicle kinematic model based on a point mass model; S4, obtaining the vehicle collision avoidance trajectory; S5, establishing a state prediction model based on a two-degree-of-freedom vehicle dynamics model; S6, obtaining the vehicle front wheel angle control amount and sending it to the controlled vehicle to achieve collision avoidance trajectory tracking control; S7, establishing a predicted time domain control law to update the predicted time domain value, and passing it to an active collision avoidance controller to achieve active steering collision avoidance control. The present invention is based on a model predictive control algorithm and realizes autonomous obstacle bypassing and tracking control. At the same time, the relationship between the predicted time domain and the speed in the controller is fully considered, which not only further improves the control accuracy of the controller, but also improves the adaptability and stability of the controller.

Description

Technical Field

[0001] The invention relates to the field of automobile active safety, and in particular to a vehicle active collision avoidance control method based on model predictive control. Background Art

[0002] With the rapid growth of the number of cars in my country, traffic accidents are frequent and urban traffic congestion is becoming increasingly serious. According to statistics, among all traffic accidents, car collision accidents account for a large proportion (60-70%). If the driver can be reminded of the danger and take necessary safety measures before the traffic accident occurs, it will be very useful to reduce the traffic accident rate. The active obstacle avoidance control system is an effective means to achieve this function.

[0003] The active collision avoidance control system of the vehicle mainly obtains the road traffic environment ahead of the vehicle through the sensor system, and judges whether there are safety hazards ahead in combination with the vehicle's own driving status (vehicle speed, road conditions, etc.), and issues corresponding prompts or alarm information. When necessary, the control system actively takes over the driver's work and controls the vehicle to automatically avoid obstacles to ensure vehicle safety. In the field of intelligent vehicles, the research on active collision avoidance control technology of vehicles has broad application prospects, important scientific research value, and strong market competitiveness.

[0004] In recent years, domestic and foreign scholars have conducted a lot of research on the obstacle avoidance path planning and trajectory tracking control of intelligent vehicles. In terms of obstacle avoidance trajectory planning, methods such as quintic polynomials, geometric path planning, and artificial potential energy algorithms are often used to achieve vehicle obstacle avoidance path planning; the commonly used control methods mainly include model predictive control algorithms, preview feedforward, and state feedback control algorithms. However, due to the complexity and variability of the operating environment of intelligent driving vehicles, there is still a large gap between existing research and actual application. There are still some technical issues that need further research, which are mainly summarized as follows:

[0005] Active collision avoidance does not take obstacle information into consideration enough. Usually, obstacles are regarded as a point or an area in the design of active collision avoidance systems, and the relative position relationship between the obstacle and the vehicle is evaluated through a safe distance model. However, in the actual steering collision avoidance process, the control system is required to not only autonomously bypass the obstacle, but also return to the expected driving trajectory or area as soon as possible. Obviously, simple obstacle position information cannot meet the needs of obstacle avoidance and fast tracking at the same time.

[0006] The applicability of collision avoidance path planning is low. For most lateral collision avoidance systems, their collision avoidance paths are mostly pre-set, or the path planning calculations are complex, and the ability to respond to changes in the driving environment is insufficient. The path cannot be changed according to vehicle status and environmental changes, and it cannot be guaranteed that the intelligent vehicle can accurately handle any collision avoidance trajectory tracking problems.

[0007] The control parameters of the tracking controller lack variability. The control parameters of the model predictive control algorithm mainly include the prediction time domain, the control time domain and the sampling parameters, which have a certain impact on the tracking control accuracy of the controller. However, most scholars currently use controllers with fixed control parameters when using the model predictive control algorithm to study tracking problems, which cannot effectively guarantee the tracking control accuracy of the controller at different speeds. Summary of the invention

[0008] In view of the problems existing in the above-mentioned prior art, the present invention studies active collision avoidance control from three aspects: vehicle model, trajectory planning method and trajectory tracking control hierarchical design, and proposes an active collision avoidance control method which reduces the complexity of the control system and improves the adaptability of the controller while ensuring the accuracy of the control system. The specific details are as follows.

[0009] An active collision avoidance control method based on a model predictive control algorithm comprises the following steps:

[0010] S1: Use the vehicle-mounted camera, millimeter-wave radar, and vehicle-mounted navigation equipment to obtain the controlled vehicle information, vehicle positioning information, and obstacle information, and collect the vehicle's driving road information to obtain the global path information as the reference trajectory of the vehicle.

[0011] S2: After obtaining the information required for vehicle collision avoidance trajectory planning, the distance deviation between the controlled vehicle and the obstacle is obtained according to the road information, obstacle information, and controlled vehicle information. The influence of vehicle speed and collision avoidance weight on obstacle avoidance is comprehensively considered to establish an obstacle avoidance function.

[0012] S3: Establish a local collision avoidance trajectory planner; the local collision avoidance trajectory planner uses the nonlinear vehicle kinematic model established based on the point mass model as a prediction model of the model predictive control algorithm;

[0013] S4: Combine the obstacle avoidance function, add vehicle soft constraints, set the objective function, and obtain the position coordinates of the vehicle collision avoidance process, thereby obtaining the planned collision avoidance trajectory.

[0014] S5: Based on the current state information of the vehicle, a two-degree-of-freedom vehicle dynamics model is established. Through Taylor formula linearization and forward Euler discretization, a discretized linear vehicle dynamics model is obtained as the prediction model of the model predictive control algorithm in the lower-level tracking controller.

[0015] S6: Combined with the vehicle dynamics constraints, the front wheel steering angle increment is used as the control variable, the objective function is designed, and the model predictive control algorithm is used to solve the optimal front wheel steering angle control variable, which is input to the controlled vehicle to achieve collision avoidance trajectory tracking control.

[0016] S7: Establish a prediction time domain control law. Through simulation testing, establish the functional relationship between vehicle speed and prediction time domain, so as to update the prediction time domain value in real time according to the current vehicle speed, pass it to the upper-level collision avoidance trajectory planner, and combine it with the next trajectory tracking controller to realize the vehicle's active steering collision avoidance control.

[0017] Furthermore, in S1, the controlled vehicle information includes the current state operation information of the controlled vehicle; obstacle information includes the number, size, and position of obstacles; and the vehicle positioning information includes angle information, acceleration information, speed information, and position information.

[0018] Furthermore, in S2, the obstacle avoidance function is established by acquiring vehicle information through vehicle-mounted equipment, acquiring front obstacle information and position relationship through millimeter wave radar, and acquiring vehicle speed information through vehicle-mounted sensors to establish the obstacle avoidance function.

[0019] Furthermore, in S3, the vehicle state information, vehicle position information and point mass model are used to establish a vehicle kinematic state equation for the vehicle's current operating state, and forward Euler discretization is used to obtain a discrete nonlinear vehicle kinematic state equation.

[0020] Further, the S4 includes the following steps:

[0021] S4.1. Use lateral acceleration as a soft restraint on the vehicle;

[0022] S4.2. According to the vehicle kinematic state equation of the vehicle's current running state, combined with the vehicle's soft constraints and obstacle avoidance function, the lateral acceleration is used as the control variable, and the objective function of the upper nonlinear model predictive controller is set in the following specific form:

[0023]

[0024] st.a min ≤a v (j)≤a max , (j=1, 2, ... Nc)

[0025] Where: k represents the kth moment, Q and R are the output weight matrix and the control weight matrix respectively, Nc is the control time domain, Np is the prediction time domain, χ(k+i|k) is the predicted trajectory point, χ ref (k+i|k) is the reference trajectory point; a y is the lateral acceleration; J obs,i is the obstacle avoidance function at sampling time i; ρ is the relaxation factor weight, ε is a non-zero positive number used to prevent the denominator from being zero; a min 、a maxThey represent the minimum and maximum values ​​of the lateral acceleration respectively, and the specific form of the obstacle avoidance function is as follows:

[0026]

[0027] Where: S obs is the obstacle avoidance weight coefficient; is the vehicle speed at time i; (x 0 ,y 0 ) is the target position of the vehicle when avoiding obstacles; (x i ,y i ) is the vehicle position at time i; γ is the angle between the obstacle and the vehicle heading, which is used to indicate the position of the obstacle relative to the vehicle;

[0028] S4.3. The objective function is transformed into a quadratic programming (QP) problem, and the optimal solution is obtained using the effective set method. A series of optimal lateral acceleration control quantities are obtained by rolling optimization, and then discrete planning trajectory data points are obtained;

[0029] S4.4. The obtained discrete data points are selected, and the horizontal coordinate and yaw angle of the local path obtained by fitting the fifth-order polynomial are passed to the tracking control layer in the following form.

[0030]

[0031]

[0032] Where: X is the longitudinal displacement of the vehicle in the geodetic coordinate system in the prediction time domain; c i and d i are the lower layer reference trajectory fitting coefficient and reference yaw angle fitting coefficient respectively.

[0033] Further, the S6 comprises the following steps:

[0034] S6.1. According to the established vehicle prediction model at the current moment, the front wheel angle increment is used as the control variable, and the objective function of the model prediction control algorithm is set in the following form:

[0035]

[0036] in, Vehicle status information; is the control input of the vehicle at time k-1; the control increment ΔU(k) ​​in the control time domain; δ f,k and Δδ f,k are the front wheel steering angle control value and the front wheel steering angle control increment at time k respectively; the hard constraint y hIncluding tire slip angle constraint, center of mass slip angle constraint and tire slip angle constraint; Δη(k+i|k) is the difference between the actual output in the prediction time domain and the known reference trajectory; is the control increment at time k+i, is the control input of the vehicle at k+i; δ f,min , δ f,max They are the minimum and maximum front wheel turning angles of the vehicle respectively; Δδ f,min , Δδ f,max are the minimum and maximum values ​​of the vehicle's front wheel angle increment respectively; y h,min ,y h,max are the minimum and maximum values ​​of the hard constraints respectively; Q and R are the output weight matrix and the control weight matrix respectively; ρ is the weight coefficient, and ε is the system relaxation factor;

[0037] S6.2. Convert the objective function into a quadratic programming problem and solve it using the interior point method to obtain the control increment ΔU(k) ​​in the control domain = [Δu * (k),Δu * (k+1),…,Δu * (k+Nc-1)], the first element of the control increment sequence is taken as the actual control quantity of the controlled object. When it comes to the next time k+1, the above process is repeated, and the constrained optimization problems are completed one by one in a rolling manner to achieve continuous control of the controlled object.

[0038] Furthermore, in S7, the functional relationship between the predicted time domain and the speed is obtained through cluster analysis and comprehensive evaluation, and the predicted time domain control law is established. The predicted time domain value is updated in real time according to the current vehicle speed and input into the active collision avoidance controller to realize the active steering collision avoidance control of the vehicle. The function form is as follows:

[0039]

[0040] Where: Np is the predicted time domain value, v is the actual speed value of the vehicle.

[0041] Compared with the existing methods, the advantages of this method are mainly reflected in the following aspects:

[0042] The vehicle active collision avoidance control method proposed in the present invention is based on a double-layer model predictive control algorithm to establish an active collision avoidance controller, which realizes autonomous obstacle bypassing and tracking control. At the same time, the relationship between the prediction time domain and the speed in the controller is fully considered, which not only further improves the control accuracy of the controller, but also improves the adaptability and stability of the controller.

[0043] In addition, the present invention also provides a reference method for research in the same field, which can be further extended to other related collision avoidance control fields based on this method, and has high practicality and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The framework of active collision avoidance control method based on model predictive control algorithm;

[0045] Figure 2 Sensor module;

[0046] Figure 3 Schematic diagram of the MPC local obstacle avoidance trajectory planner;

[0047] Figure 4 Schematic diagram of penalty function;

[0048] Figure 5 Schematic diagram of the positional relationship between obstacles and vehicles;

[0049] Figure 6 Schematic diagram of MPC trajectory tracking controller; DETAILED DESCRIPTION

[0050] The present invention proposes an active collision avoidance control method based on a model predictive control algorithm, and its specific structure is as follows: Figure 1 As shown, the steps are:

[0051] S1: Use the on-board camera, millimeter-wave radar, and on-board equipment to obtain the controlled vehicle information, vehicle positioning information, and forward obstacle information, and collect the vehicle's driving road information to obtain road section information as the preliminary global reference trajectory of the vehicle.

[0052] The controlled vehicle information includes the current operating status information of the controlled vehicle; obstacle information includes the number, size and position of obstacles; and the vehicle positioning information includes angle information, acceleration information, speed information and position information.

[0053] S2: After obtaining the information required for vehicle collision avoidance trajectory planning, an obstacle avoidance function based on a penalty function is established according to the road information, obstacle information, and controlled vehicle information. Here, the vehicle information is obtained through the vehicle-mounted equipment, the obstacle information and position relationship in front are obtained through the millimeter-wave radar, and the speed information of the vehicle is obtained through the vehicle-mounted sensor to establish the obstacle avoidance function.

[0054] S3: Using the vehicle state information, vehicle position information and point mass model, establish the vehicle kinematic state equation for the vehicle's current operating state, and use forward Euler discretization to obtain the discrete nonlinear vehicle kinematic state equation as the prediction model of the model predictive control algorithm of the upper planner of the active collision avoidance controller. Figure 3 shown.

[0055] S4: Based on the penalty function, considering the vehicle speed, obstacle avoidance coefficient and the positional relationship between the vehicle and the obstacle, an obstacle avoidance function is established. The penalty function diagram and the positional relationship diagram between the obstacle and the vehicle are shown in the figure below. Figure 4 and Figure 5 As shown in the figure, the lateral acceleration is used as the vehicle soft constraint. According to the vehicle kinematic state equation of the vehicle's current running state, combined with the vehicle soft constraint and obstacle avoidance function, the lateral acceleration is used as the control quantity, and the objective function of the upper nonlinear model predictive controller is set as follows:

[0056]

[0057] sta min ≤a v (j)≤a max , (j=1, 2, ... Nc)

[0058] Where: k represents the kth moment, Q and R are the output weight matrix and the control weight matrix respectively, Nc is the control time domain, Np is the prediction time domain, χ(k+i|k) is the predicted trajectory point, χ ref (k+i|k) is the reference trajectory point; a y is the lateral acceleration; J obs,i is the obstacle avoidance function at sampling time i; ρ is the relaxation factor weight, ε is a non-zero positive number used to prevent the denominator from being zero; a min 、a max Respectively represent the minimum and maximum values ​​of lateral acceleration. The specific form of the obstacle avoidance function is as follows:

[0059]

[0060] Where: S obs is the obstacle avoidance weight coefficient; is the vehicle speed at time i; (x 0 ,y 0 ) is the target position of the vehicle when avoiding obstacles; (x i ,y i ) is the vehicle position at time i; γ is the angle between the obstacle and the vehicle heading, which is used to indicate the position of the obstacle relative to the vehicle;

[0061] S4.1. Convert the objective function into a quadratic programming (QR) problem, use the effective set method to perform the optimal solution, obtain the optimal lateral acceleration control value, and then obtain the discrete planning trajectory data points.

[0062] S4.2. The obtained discrete data points are selected, and the horizontal coordinates and yaw angles of the local path obtained by fitting the fifth-order polynomial are passed to the tracking control layer in the following form. The workflow is as follows: Figure 3 shown.

[0063]

[0064]

[0065] Where: X is the longitudinal displacement of the vehicle in the geodetic coordinate system in the prediction time domain; c i and d i are the lower layer reference trajectory fitting coefficient and reference yaw angle fitting coefficient respectively.

[0066] S5: According to the current state information of the vehicle, a two-degree-of-freedom vehicle dynamics model is established. Through Taylor formula linearization and forward Euler discretization, a discretized linear vehicle dynamics model is obtained as the prediction model of the model predictive control algorithm in the lower tracking controller. The specific form of the prediction model is as follows:

[0067]

[0068] Where: is the state of the vehicle at time k, is the control input of the vehicle at k-1; C k is the output matrix coefficient, is the control increment at time k, and the control output in the entire prediction time domain Np is iterated to obtain the output matrix Y at time k c The expression is:

[0069]

[0070] Where: k and θ k are the state matrix and output matrix weights, and ΔU(k) ​​is the control increment in the control domain.

[0071] S6: Combined with the vehicle dynamics constraints, the front wheel steering angle increment is used as the control quantity, the objective function is designed, and the model predictive control algorithm is used to solve the optimal front wheel steering angle control quantity, which is input to the controlled vehicle to achieve real-time collision avoidance trajectory tracking control. The specific process is as follows: Figure 6 shown.

[0072] The specific steps of S6 are as follows:

[0073] S6.1. Establish the current vehicle state matrix based on the vehicle state information and vehicle positioning information, use the front wheel angle increment as the control variable, and set the objective function of the model predictive control algorithm in the following form:

[0074]

[0075] in, Vehicle status information; is the control input of the vehicle at time k-1; the control increment ΔU(k) ​​in the control time domain; δ f,k and Δδ f,k are the front wheel steering angle control value and the front wheel steering angle control increment at time k respectively; the hard constraint y h Including tire slip angle constraint, center of mass slip angle constraint and tire slip angle constraint; Δη(k+i|k) is the difference between the actual output in the prediction time domain and the known reference trajectory; is the control increment at time k+i, is the control input of the vehicle at k+i; δ f,min , δ f,max They are the minimum and maximum front wheel turning angles of the vehicle respectively; Δδ f,min , Δδ f,max are the minimum and maximum values ​​of the vehicle's front wheel angle increment respectively; y h,min ,y h,max are the minimum and maximum values ​​of the hard constraints respectively; Q and R are the output weight matrix and the control weight matrix respectively; ρ is the weight coefficient, and ε is the system relaxation factor;

[0076] S6.2. Convert the objective function into a quadratic programming problem and solve it using the interior point method to obtain the change in the front wheel steering angle of the vehicle. The specific conversion form is as follows:

[0077]

[0078] Where: is a positive definite matrix; is the constraint matrix; Q and R are the output weight matrix and the control weight matrix respectively; ΔU(k) ​​is the control increment in the control time domain; Vehicle status information; is the control input of the vehicle at time k-1; ε is the system relaxation factor, a positive number not less than 0.

[0079] Solve to obtain the control increment ΔU(k) ​​in the control time domain = [Δu * (k),Δu * (k+1),…,Δu * (k+Nc-1)], the control quantity of the first element of the control increment sequence at the current moment is taken as the actual control quantity of the controlled object. When it comes to the next moment k+1, the above process is repeated, and the constrained optimization problems are completed one by one in a rolling manner to achieve continuous control of the controlled object.

[0080] S7: Through cluster analysis and comprehensive evaluation, a prediction time domain control law is established, and the prediction time domain value is updated in real time according to the current vehicle speed and input into the active collision avoidance controller to improve the adaptability of the controller. The function between the prediction time domain and the speed is as follows:

[0081]

[0082] Where: Np is the predicted time domain value, v is the actual speed value of the vehicle.

[0083] In summary, the vehicle active collision avoidance control method proposed in the present invention is based on the double-layer model predictive control algorithm to establish an active collision avoidance controller, which realizes autonomous obstacle bypassing and tracking control. At the same time, the relationship between the prediction time domain and the speed in the controller is fully considered, which not only further improves the control accuracy of the controller, but also improves the adaptability and stability of the controller.

[0084] In addition, the present invention also provides a reference method for research in the same field, which can be further extended to other related collision avoidance control fields based on this method, and has high practicality and promotion value.

Claims

1. An active collision avoidance control method based on a model predictive control algorithm, characterized in that: The following steps are involved: S1: Use the vehicle camera, millimeter wave radar, and vehicle navigation equipment to obtain the controlled vehicle information, vehicle positioning information, and obstacle information, and collect the vehicle driving road information to obtain the global path information as the reference trajectory of the vehicle; S2: After obtaining the information required for vehicle collision avoidance trajectory planning, the distance deviation between the controlled vehicle and the obstacle is obtained according to the road information, obstacle information, and controlled vehicle information, and the influence of vehicle speed and collision avoidance weight on obstacle avoidance is comprehensively considered to establish an obstacle avoidance function; S3: Establish a local collision avoidance trajectory planner; the local collision avoidance trajectory planner uses the nonlinear vehicle kinematic model established based on the point mass model as a prediction model of the model predictive control algorithm; In S3, a vehicle kinematic state equation for the current running state of the vehicle is established by using the vehicle state information, the vehicle position information and the point mass model, and a discrete nonlinear vehicle kinematic state equation is obtained by using forward Euler discretization; S4: Combine the obstacle avoidance function, add vehicle soft constraints, set the objective function, and obtain the position coordinates of the vehicle collision avoidance process, thereby obtaining the planned collision avoidance trajectory; S5: According to the current state information of the vehicle, a two-degree-of-freedom vehicle dynamics model is established, and a discretized linear vehicle dynamics model is obtained through Taylor formula linearization and forward Euler discretization, which is used as the prediction model of the model predictive control algorithm in the lower tracking controller; S6: Combined with the vehicle dynamics constraints, the front wheel steering angle increment is used as the control variable, the objective function is designed, and the model predictive control algorithm is used to solve the optimal front wheel steering angle control variable, which is input to the controlled vehicle to achieve collision avoidance trajectory tracking control; The S6 comprises the following steps: S6.

1. Establish the current vehicle state matrix based on the vehicle state information and vehicle positioning information, use the front wheel angle increment as the control quantity, and set the objective function of the model predictive control algorithm The form is as follows: stΔδ f，min ≤Δδ f，k ≤Δδ f,max Control Increment Constraints δ f，min ≤AΔδ f，k +δ f，k ≤δ f，max Control quantity constraint y h,min ≤y h ≤y h，max Hard Constraints a y,min -ε≤a y ≤a y,max +ε soft constraint in, Vehicle status information; is the control input of the vehicle at time k-1; the control increment ΔU(k) ​​in the control time domain; δ f,k and Δδ f,k are the front wheel steering angle control value and the front wheel steering angle control increment at time k respectively; the hard constraint y h Including tire slip angle constraint, center of mass slip angle constraint and tire slip angle constraint; Δη(k+i|k) is the difference between the actual output in the prediction time domain and the known reference trajectory; is the control increment at time k+i, is the control input of the vehicle at k+i; δ f,min , δ f,max They are the minimum and maximum front wheel turning angles of the vehicle respectively; Δδ f,min , Δδ f,max are the minimum and maximum values ​​of the vehicle's front wheel angle increment respectively; y h,min ,y h,max are the minimum and maximum values ​​of the hard constraints respectively; Q and R are the output weight matrix and the control weight matrix respectively; ρ is the weight coefficient, and ε is the system relaxation factor; S6.

2. The objective function is converted into a quadratic programming problem, and the quadratic programming problem is solved using the interior point method to obtain a control sequence of the vehicle's front wheel angle change. The first element of the control sequence is used as the actual control amount of the controlled object. When the next time k+1 comes, the above process is repeated, and the constrained optimization problems are completed one by one in a rolling manner to achieve continuous control of the controlled object. S7: Establish a prediction time domain control law. Through simulation testing, establish a functional relationship between vehicle speed and prediction time domain, so as to update the prediction time domain value in real time according to the current vehicle speed, pass it to the collision avoidance trajectory planner, and combine it with the next trajectory tracking controller to realize the vehicle's active steering collision avoidance control.

2. The vehicle active collision avoidance control method based on the model predictive control algorithm according to claim 1, characterized in that: In said S1, the controlled vehicle information includes the current state operation information of the controlled vehicle; The obstacle information includes the number, size and position of the obstacles; the vehicle positioning information includes angle information, acceleration information, speed information and position information.

3. The vehicle active collision avoidance control method based on model predictive control algorithm according to claim 1, characterized in that: In S2, the obstacle avoidance function is established by obtaining vehicle information through vehicle-mounted equipment, obtaining front obstacle information and position relationship through millimeter-wave radar, and obtaining vehicle speed information through vehicle-mounted sensors to establish the obstacle avoidance function.

4. The vehicle active collision avoidance control method based on model predictive control algorithm according to claim 1, characterized in that: The S4 includes the following steps: S4.

1. Use lateral acceleration as a soft restraint on the vehicle; S4.

2. According to the vehicle kinematic state equation of the vehicle's current running state, combined with the vehicle's soft constraints and obstacle avoidance function, the lateral acceleration is used as the control variable, and the objective function J of the upper nonlinear model predictive controller is set in the following specific form: Where: k represents the kth moment, Q and R are the output weight matrix and the control weight matrix respectively, Nc is the control time domain, Np is the prediction time domain, χ(k+i|k) is the predicted trajectory point, χ ref (k+i|k) is the reference trajectory point; a y is the lateral acceleration; J obs,i is the obstacle avoidance function at sampling time i; ρ is the relaxation factor weight, ε is a non-zero positive number used to prevent the denominator from being zero; a min 、a max They represent the minimum and maximum values ​​of the lateral acceleration respectively, and the specific form of the obstacle avoidance function is as follows: Where: S obs is the obstacle avoidance weight coefficient; is the vehicle speed at time i; (x0, y0) is the target position of the vehicle when avoiding obstacles; (x i ,y i ) is the vehicle position at time i; γ is the angle between the obstacle and the vehicle heading, which is used to indicate the position of the obstacle relative to the vehicle; S4.

3. The objective function is transformed into a quadratic programming (QP) problem, and the optimal solution is obtained using the effective set method. A series of optimal lateral acceleration control quantities are obtained by rolling optimization, and then discrete planning trajectory data points are obtained; S4.

4. The obtained discrete data points are selected, and the abscissa and yaw angle of the local path obtained by fitting the fifth-order polynomial are passed to the tracking control layer in the following form; Where: X is the longitudinal displacement of the vehicle in the geodetic coordinate system in the prediction time domain; c i and d i are the lower layer reference trajectory fitting coefficient and reference yaw angle fitting coefficient respectively.

5. The vehicle active collision avoidance control method based on model predictive control algorithm according to claim 1, characterized in that: In S7, the functional relationship between the predicted time domain and the speed is obtained through cluster analysis and comprehensive evaluation, and the predicted time domain control law is established. The predicted time domain value is updated in real time according to the current vehicle speed and input into the active collision avoidance controller to realize the active steering collision avoidance control of the vehicle. The function form is as follows: Among them: Np is the predicted time domain value, v is the actual speed value of the vehicle.

Images