Automatic driving heavy-load commercial vehicle queue control method and system based on low-orbit satellite network

Abstract

The invention discloses an automatic driving heavy-load commercial vehicle queue control method and system based on a low-orbit satellite network, high-precision positioning and real-time inter-vehicle communication are carried out by using the low-orbit satellite network, when a front road is a straight road, a queue keeps longitudinal following driving in a lane at a proper vehicle distance, and the front road is a straight road; longitudinal motion control is carried out through distributed model predictive control, when a lane changing demand is generated, a queue changes lanes along a quintic polynomial curve path under the guidance of a pilot car, and transverse motion control is carried out through self-adaptive model predictive control. According to the invention, the stability and safety during longitudinal following driving of the queue can be ensured, and the safety of vehicle transverse movement and queue control is improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation, and particularly to a method and system for controlling an autonomous driving heavy-duty commercial vehicle queue based on a low-earth orbit satellite network. Background Art

[0002] Coal is the main energy source of the country, which is related to the national economy and people's livelihood as well as energy security. With more and more commercial vehicles being applied to transporting coal from Xinjiang to other regions, the practical difficulties encountered in commercial vehicle transportation are becoming more and more obvious. The working conditions in the northwest region are complex and the climate is harsh, which pose great challenges to vehicle transportation efficiency, driver driving skills and physical strength. An autonomous driving commercial vehicle queue is a typical cooperative motion control. The intelligent commercial vehicles in the queue can automatically adjust their motion states according to the state information of surrounding vehicles to achieve formation driving. Applying an autonomous driving commercial vehicle queue in transporting coal from Xinjiang to other regions can improve safety, transportation efficiency and reduce operating costs through multi-vehicle cooperative control, and has many advantages.

[0003] The normal driving of an autonomous driving vehicle fleet depends on the high-speed and stable communication between member vehicles and requires real-time access to cloud data. However, 5G base stations have not been fully covered in the western region, the communication delay between vehicles and between vehicles and the cloud is large, and the road environment is complex, and the construction of infrastructure such as traffic lights is not perfect, which poses a great challenge to the application of autonomous driving vehicle fleets. The low-earth orbit satellite has a low orbit height, resulting in short transmission delay and small path loss, with the advantages of small signal delay and wide coverage. In addition, cooperating with geostationary satellites can provide high-precision positioning information. Therefore, integrating a low-earth orbit satellite network into an autonomous driving vehicle fleet can obtain high-precision positioning and good vehicle-to-vehicle communication effects under complex terrains and harsh working conditions, especially under the condition of the absence of 4G / 5G base stations, so as to realize fleet cooperative control.

[0004] The control method of commercial vehicle queues has been studied since the 1950s and continues to this day. The control methods have experienced classical control, modern control and intelligent control. For heavy-duty commercial vehicles transporting coal, the full load mass can reach more than 30 tons, and the load mass and load distribution often change. Model predictive control (MPC) can better handle the uncertainty and disturbance of the system through real-time optimization and feedback correction, thereby improving the robustness of the system. Among them, distributed model predictive control (DMPC) uses a distributed architecture to coordinate vehicle behaviors by sharing state information, improving the flexibility and reliability of the system; adaptive model predictive control (AMPC) can adjust control parameters in real time according to environmental changes to adapt to system changes. Therefore, comprehensively applying DMPC and AMPC can effectively realize the longitudinal and lateral motion control of heavy vehicle queues. Summary of the Invention

[0005] In view of this, the present invention provides a control method and system for an autonomous driving heavy-duty commercial vehicle queue based on a low-earth orbit satellite network. For the heavy-duty commercial vehicle queue applied to the transportation of coal from Xinjiang to other regions, a five-degree-of-freedom dynamic model and a non-linear tire model suitable for heavy vehicles are established. The low-earth orbit satellite network is used for high-precision positioning and real-time vehicle-to-vehicle communication. A distributed model predictive control algorithm (DMPC) is proposed for longitudinal vehicle distance maintenance control of the commercial vehicle queue, and an adaptive model predictive control algorithm (AMPC) is designed for lateral motion control of the commercial vehicle queue.

[0006] The present invention achieves the above technical objectives through the following technical means.

[0007] Control method for an autonomous driving heavy-duty commercial vehicle queue based on a low-earth orbit satellite network:

[0008] The leading vehicle in the autonomous driving heavy-duty commercial vehicle queue makes a driving state decision by receiving information about the road conditions ahead from the low-earth orbit satellite: when the road ahead is a straight road, the queue maintains longitudinal following driving in the lane at an appropriate vehicle distance, and longitudinal motion control is performed through distributed model predictive control; when a lane change is required, the queue changes lanes along a quintic polynomial curve path under the leadership of the leading vehicle, and lateral motion control is performed through adaptive model predictive control.

[0009] Furthermore, longitudinal motion control is performed through distributed model predictive control, specifically: the distributed model predictive controller decomposes the overall optimization problem of the queue into sub-optimization problems. Each following vehicle in the queue receives information about the vehicle in front and the leading vehicle from the low-earth orbit satellite, and calculates the optimal control variables of each vehicle through MPC optimization, so as to achieve coordinated longitudinal motion control of the queue.

[0010] Even further, the distributed model predictive controller is:

[0011]

[0012] where J ilon (t) is the objective function of the longitudinal controller, J i1 is the following vehicle error cost function between vehicle i and the vehicle in front, and the vehicle distance between following vehicle i and the vehicle in front at time t + k is the position of the vehicle in front, is the position of vehicle i, L is the vehicle length, is the desired vehicle distance between vehicle i and the vehicle in front, v i-1，i (k|t) is the longitudinal speed error between the own vehicle and the vehicle in front at the prediction time, w1 is the following vehicle distance error adjustment coefficient, w2 is the following vehicle speed error adjustment coefficient, J i2 is the following vehicle error cost function between vehicle i and the leading vehicle, and is the position of the leading vehicle at time t + k, is the position of vehicle i, is the desired headway between the leading vehicle and vehicle i, is the assumed longitudinal speed of the leading vehicle, is the predicted longitudinal speed of vehicle i, w3 is the headway error adjustment coefficient with the leading vehicle, and w4 is the speed error adjustment coefficient with the leading vehicle, is the predicted position of vehicle i at t + N P time, is the assumed terminal position of vehicle j in the queue, is the terminal desired spacing between vehicle j and vehicle i, and n is the number of vehicles in the queue, is the predicted speed of vehicle i at t + N P time, is the assumed terminal speed of vehicle j in the queue, v max is the speed upper limit of the heavy commercial vehicle, is the predicted acceleration of vehicle i at the t + k-th moment, a max is the acceleration upper limit, a min is the acceleration lower limit, i = 2, 3, 4,..., n.

[0013] Furthermore, the quintic polynomial curve path is:

[0014]

[0015] where Y i is the lateral position of the lane-changing trajectory of vehicle i, X i is the longitudinal position of the lane-changing trajectory of vehicle i, and the curve shape parameters a0 = a1 = a2 = 0, D lat is the specified value of the longitudinal displacement of the vehicle, D lon is the specified value of the lateral displacement of the vehicle.

[0016] Even further, the lateral position and yaw angle of the quintic polynomial curve path are respectively:

[0017]

[0018] Even further, lateral motion control is performed through adaptive model predictive control, specifically: the adaptive model predictive lateral controller monitors the paths of each vehicle in the queue, and uses the vehicle dynamics model and the MPC controller whose prediction horizon changes with vehicle speed to determine the ideal front wheel steering angle of the vehicle in real time.

[0019] Even further, the adaptive model predictive lateral controller is:

[0020]

[0021] Among them, J ilat is the objective function of the adaptive model predictive lateral controller, and η i (k|t) represents the output of vehicle i, and ΔU i (k|t) represents the control increment of vehicle i, and η ref (k|t) is the reference output, and N P and N c are the sizes of the prediction horizon and the control horizon respectively. M and N are the weight matrices of the output variable and the control increment value ΔU(t). ρ is the weighting factor, ε is the relaxation factor, and u max and u min are the upper and lower boundary values of the control quantity of the adaptive model predictive lateral controller respectively. ΔU max and ΔU min are the upper and lower boundary values of the control increment of the adaptive model predictive lateral controller respectively. η min and η max are the upper and lower boundary values of the output variable respectively. μ is the road surface adhesion coefficient, and g is the acceleration due to gravity, is the lateral acceleration of vehicle i.

[0022] Furthermore, the prediction horizon satisfies:

[0023]

[0024] In the formula, k1, k2, and k3 are all coefficients in the polynomial fitting curve, k4 is the constant term in the polynomial fitting curve, and v xi is the longitudinal vehicle speed of vehicle i.

[0025] Furthermore, the member vehicles of the platoon of autonomous heavy-duty commercial vehicles communicate based on low-earth orbit satellites. Each vehicle receives signals from the low-earth orbit satellites and sends its own status information to the low-earth orbit satellites.

[0026] An autonomous heavy-duty commercial vehicle platoon control system based on a low-earth orbit satellite network, comprising:

[0027] A heavy-duty commercial vehicle dynamics model and a non-linear tire model establishment module, which are used to establish a five-degree-of-freedom vehicle dynamics model and a non-linear tire model of a commercial vehicle respectively;

[0028] A commercial vehicle platoon model establishment module, which is used to establish a commercial vehicle platoon model based on low-earth orbit satellite communication;

[0029] The longitudinal and lateral control module of the queue of autonomous heavy-duty commercial vehicles performs longitudinal motion control through distributed model predictive control and lateral motion control through adaptive model predictive control.

[0030] The beneficial effects of the present invention are as follows:

[0031] (1) The present invention applies low-earth orbit satellites to the control of commercial vehicle queues, obtains the accurate position and speed information of the vehicles in the queue based on low-earth orbit satellite positioning, and conducts highly reliable communication with other vehicles in the queue through low-earth orbit satellites to reduce the risks of instability and collision and improve the safety of queue control.

[0032] (2) Based on the accurate modeling of heavy-duty commercial vehicles, the present invention designs a queue longitudinal control strategy based on DMPC to ensure the stability and safety during longitudinal following driving of the queue.

[0033] (3) In view of the problem of large variations in the load weight and distribution of heavy-duty vehicles, the present invention designs a lateral control algorithm based on AMPC to constrain the front-wheel steering angle and lateral safety acceleration of the vehicle and improve the safety of lateral motion. Description of the Drawings

[0034] Figure 1 is the flowchart of the control method for the queue of autonomous heavy-duty commercial vehicles based on the low-earth orbit satellite network in the present invention;

[0035] Figure 2 is the dynamic model diagram of the commercial vehicle in the present invention;

[0036] Figure 3 is the schematic diagram of the tire dynamic model in the present invention;

[0037] Figure 4 is the schematic diagram of the fleet communication structure based on the low-earth orbit satellite network in the present invention;

[0038] Figure 5 is the schematic diagram of the queue control structure based on the low-earth orbit satellite network in the present invention. Detailed Embodiments

[0039] The following further illustrates the present invention in conjunction with the drawings and specific embodiments, but the protection scope of the present invention is not limited thereto.

[0040] The present invention proposes a control method for the queue of autonomous heavy-duty commercial vehicles based on the low-earth orbit satellite network, and further illustrates the invention in conjunction with the attached Figures 1-5 drawings for further explanation.

[0041] Figure 1 is the flowchart of the control method for the queue of autonomous heavy-duty commercial vehicles based on the low-earth orbit satellite network, including the following steps:

[0042] S1: Establish a dynamic model of heavy-duty commercial vehicles and a non-linear tire model;

[0043] S2: Establish a commercial vehicle platoon model based on low-earth orbit satellite communication;

[0044] S3: Design a longitudinal and lateral control algorithm for platoons based on low-earth orbit satellite networks.

[0045] Step 1: Establish a dynamic model of heavy-duty commercial vehicles and a non-linear tire model. Specifically as follows:

[0046] S1.1, Establish a dynamic model of commercial vehicles

[0047] In the present invention, for heavy-duty commercial vehicles, based on the classical three-degree-of-freedom vehicle model (i.e., considering the vehicle's longitudinal, lateral, and yaw motions), the wheel rotation degree of freedom is introduced to establish a five-degree-of-freedom vehicle model as shown in the appendix. Figure 2 As shown.

[0048] Ignoring the load transfer between wheels and vehicle aerodynamics, according to Newton's second law, the vehicle force and moment balance equations can be obtained:

[0049]

[0050] In the formula, v x , v y and ω are respectively the longitudinal speed, lateral speed, and yaw angular velocity of the vehicle; m is the mass of the vehicle; δ is the front wheel angle of the vehicle; F lf , F lr are respectively the longitudinal forces of the front and rear wheels of the vehicle; F sf , F sr are respectively the lateral forces of the front and rear wheels of the vehicle; l f , l r are respectively the distances from the front and rear axles to the center of mass of the vehicle; I z is the moment of inertia of the vehicle about the z-axis.

[0051] By introducing the wheel rotation degree of freedom into the vehicle model, as shown in the appendix for the tire dynamics model, the following can be obtained: Figure 3 As follows:

[0052]

[0053] In the formula, ω f , ω r are respectively the angular velocities of the front and rear tires of the vehicle; R e is the wheel rotation radius of the vehicle; J f , J r are respectively the moments of inertia of the front and rear wheels of the vehicle; T f , T r are respectively the torques of the front and rear wheels of the vehicle.

[0054] Combining Equation (1) and Equation (2), a five-degree-of-freedom vehicle dynamics model can be obtained:

[0055]

[0056] Let the state variables be the longitudinal velocity, lateral velocity, yaw rate, and front and rear wheel angular velocities, i.e., x = [v x , v y , ω, ω f , ω r T ; the control inputs are the front wheel steering angle and the vehicle torque, i.e., u = [δ, T] T ; the control outputs are the longitudinal velocity, lateral velocity, and yaw rate, i.e., y = [v x , v y , ω] T . Equation (3) can be written in the form of the following state equation:

[0057]

[0058] where f is a smooth function; G is a constant matrix, G = [I 3×3 , 0 3×2 .

[0059] S1.2. Establish a non-linear tire model

[0060] Tires are the key components connecting the vehicle and the driving road surface. For heavy-duty commercial vehicles, due to the frequent changes in the load mass and load distribution, an accurate non-linear tire model needs to be established for subsequent precise control. The tire magic formula is generally expressed as:

[0061] Y(x) = Dsin(Carctan(Bx - E(Bx - arctanBx))) (5)

[0062] where B, C, and D are the stiffness, shape, and peak coefficient of the tire, respectively, which are jointly determined by the vertical force on the tire and the camber angle; E is the curvature factor; Y is the output variable; x is the input variable. When Y is the lateral force, x is the slip angle; when Y is the longitudinal force, x is the longitudinal slip ratio.

[0063] The tire lateral force can be calculated from the slip angle of the tire:

[0064]

[0065] where α f , α r are the front and rear wheel slip angles of the vehicle, respectively; B s , C s , D s ​are the cornering stiffness, cornering shape, and cornering peak coefficient of the vehicle tire, respectively; E s is the cornering curvature factor.

[0066] The tire sideslip angles of the front and rear wheels are:

[0067]

[0068] In the formula, are the longitudinal and lateral speeds of the front and rear wheels in the tire coordinate system, respectively.

[0069] The longitudinal and lateral speeds of the front and rear wheels are:

[0070]

[0071] In the formula, v fx , v fy , v rx , v ry are the longitudinal and lateral speeds of the front and rear wheels in the vehicle coordinate system, respectively, and can be specifically calculated by the following formula:

[0072]

[0073] The longitudinal tire force can be obtained by calculating the slip ratio of the tire:

[0074]

[0075] In the formula, k f , k r are the slip ratios of the front and rear wheels of the vehicle, respectively; B l , C l , D l are the slip stiffness, slip shape, and slip peak coefficient of the vehicle tire, respectively; E l is the slip curvature factor.

[0076] The slip ratio can be calculated by the following formula:

[0077]

[0078] Step 2: Establish a commercial vehicle queue model based on low-earth orbit satellite communication. Specifically as follows:

[0079] The present invention uses a fixed time headway strategy to control the commercial vehicle queue. The fixed time headway means that the ratio of the inter-vehicle distance to the vehicle speed remains constant during the driving of the vehicle fleet, that is:

[0080]

[0081] In the formula, T h represents the time headway; X i represents the longitudinal distance between the host vehicle and the preceding vehicle; Vi Represents the speed of the vehicle in front.

[0082] As can be seen from the formula, when the vehicle is traveling at high speed, X i will increase correspondingly; when the vehicle is traveling at low speed, X i will decrease correspondingly. Considering the case where the vehicle speed is equal to 0, the vehicle spacing D0 when the vehicle is stationary is also set. Therefore, the desired spacing of the vehicle during driving can be expressed as:

[0083] X ref = D0 + T href V i (13)

[0084] Set the vehicle spacing D0 = 5m when the vehicle is stationary; the desired time headway T of the vehicle fleet during driving href = 0.8s.

[0085] In the present invention, the commercial vehicle queue member vehicles communicate based on low-earth orbit satellites. Each vehicle receives signals from low-earth orbit satellites and sends its own status information to low-earth orbit satellites. When studying the use of model predictive control to control the vehicle fleet in the present invention, it is necessary to obtain the relative status inside the vehicle fleet (including relative speed and relative position) in real time, and output a control quantity according to the relative status, so as to achieve string stability inside the vehicle fleet. Considering a vehicle fleet composed of n autonomous heavy commercial vehicles, the following is the car-following strategy for the following vehicle: ensure that the speed of the following vehicle is the same as that of the leading vehicle and the horizontal and vertical spacings between the following vehicle and the vehicle in front are equal to the desired horizontal and vertical spacings, as shown in the following formula:

[0086]

[0087] In the formula, v i (t) is the speed of the following vehicle i (i = 2, 3, 4,..., n); v1(t) is the speed of the leading vehicle; x i (t) is the position of vehicle i; x i-1 (t) is the position of the vehicle in front of vehicle i.

[0088] During the driving of the vehicle fleet, in order to achieve longitudinal string stability, each vehicle in the vehicle fleet needs to obtain the desired spacing, the actual spacing and the speed of the vehicle in front between its own vehicle and the vehicle in front; in order to achieve lateral string stability, each vehicle needs to obtain the lateral distance deviation between its own vehicle and the vehicle in front. Therefore, in order to achieve the transverse and longitudinal string stability inside the vehicle fleet, each following vehicle not only needs to obtain the status information of the leading vehicle but also needs to obtain the status information of the vehicle in front to calculate the relative status inside the vehicle fleet. Therefore, the present invention uses a leading vehicle - leading vehicle following topology based on low-earth orbit satellite communication for vehicle fleet communication, and the communication structure is as shown in the appendix Figure 4As shown, the leading vehicle in the queue (i = 1) receives the information of the road conditions ahead, and sends its own position x1 and speed v1. The following vehicle behind it (i = 2) receives the position x1 and speed v1 of the leading vehicle, and sends its own position x2 and speed v2. The following vehicles after that (i ≥ 3) receive the position x1 and speed v1 of the leading vehicle, the position x i-1 and speed v i-1 , and send its own position x i and speed v i .

[0089] Step 3: Design a longitudinal and lateral control algorithm for the queue based on the low-earth orbit satellite network. Specifically as follows:

[0090] The leading vehicle in the queue makes a driving state decision by receiving the information of the road conditions ahead from the low-earth orbit satellite. When the road ahead is a straight road, the queue should maintain longitudinal following driving in the lane with a suitable inter-vehicle distance, and perform longitudinal motion control through the distributed model predictive control (DMPC) algorithm; when there is a lane-changing demand, the queue changes lanes along a quintic polynomial curve path under the leadership of the leading vehicle, and performs lateral motion control through the adaptive model predictive control (AMPC) algorithm. The overall control structure is as shown in the appendix Figure 5 .

[0091] S3.1, DMPC longitudinal controller

[0092] DMPC decomposes the overall optimization problem of the queue into sub-optimization problems. Each following vehicle in the queue receives the information of the preceding vehicle and the leading vehicle from the low-earth orbit satellite, and calculates its optimal control variables through MPC optimization, so as to realize the cooperative longitudinal motion control of the queue.

[0093] For the queue of heavy commercial vehicles, the goal of car-following is to achieve stability and safety. Stability includes internal stability and string stability. Internal stability means that when the disturbance disappears, the spacing error and speed error between the vehicle and the preceding vehicle should tend to 0; string stability means that the spacing error, speed error and acceleration error of the vehicles arranged behind in the queue show a decaying trend. Safety means that the spacing error and speed error of the queue are kept within a certain range and no collision will occur. The stability and safety of the queue are controlled by designing the cost function and constraint conditions.

[0094] 1) Cost function of the following vehicle's following error with the preceding vehicle

[0095]

[0096] In the formula, J i1 is the cost function of the following error of vehicle i (i = 2, 3, 4,..., n) with the preceding vehicle; is the distance between the following vehicle i (i = 2, 3, 4,..., n) and the vehicle in front at time t + k, where is the position of the vehicle in front, is the position of vehicle i, and L represents the vehicle length; is the desired distance between vehicle i and the vehicle in front; v i-1，i (k∣t) is the longitudinal speed error between the ego vehicle and the vehicle in front at the prediction time; w1 is the distance error adjustment coefficient with the vehicle in front, and here w1 = 10; w2 is the speed error adjustment coefficient with the vehicle in front, and here w2 = 5.

[0097] 2) Following vehicle and leading vehicle following error cost function

[0098]

[0099] In the formula, J i2 is the following error cost function of vehicle i (i = 2, 3, 4,..., n) and the leading vehicle; is the position of the leading vehicle at time t + k; is the position of vehicle i; is the desired distance between the leading vehicle and vehicle i; is the assumed longitudinal speed of the leading vehicle; is the predicted longitudinal speed of vehicle i; w3 is the distance error adjustment coefficient with the leading vehicle, and here w3 = 2.5; w4 is the speed error adjustment coefficient with the leading vehicle, and here w4 = 1.25.

[0100] 3) Constraint conditions

[0101] ① Terminal position constraint

[0102]

[0103] In the formula, is the predicted position of vehicle i at t + N P time, that is, the terminal position; is the assumed terminal position of vehicle j in the queue; is the terminal desired distance between vehicle j and vehicle i; n is the number of vehicles in the queue.

[0104] ② Terminal speed constraint

[0105]

[0106] In the formula, is the predicted speed of vehicle i at t + N P time, that is, the terminal speed; is the assumed terminal speed of vehicle j in the queue.

[0107] ③ Vehicle longitudinal speed constraint

[0108] For heavy-duty commercial vehicles with complex driving conditions, a speed limit v max = 80 km / h is set, and the predicted longitudinal speed of vehicle i at the (t + k)-th moment is constrained as follows:

[0109]

[0110] ④ Vehicle acceleration constraint

[0111] Due to the limitations of the vehicle's own performance, there are limitations on the vehicle's own acceleration and deceleration. For heavy-duty commercial vehicles, to ensure safety during acceleration and braking, the acceleration upper limit a max = 1 m / s 2 , and the acceleration lower limit (braking deceleration upper limit) a min = -1 m / s 2 are taken. Then, the predicted acceleration of vehicle i at the (t + k)-th moment is constrained as follows:

[0112]

[0113] From this, the formula description of the platoon DMPC longitudinal controller is obtained:

[0114]

[0115] In the formula, J ilon (t) is the objective function of the longitudinal controller.

[0116] S3.2, Lane-changing trajectory planning

[0117] When the leading vehicle generates a lane-changing intention, the lane-changing trajectory is planned based on the current state of the host vehicle. A fifth-degree polynomial curve is used as the desired lane-changing trajectory, and its form is as follows:

[0118]

[0119] In the formula, Y i is the lateral position of the lane-changing trajectory of vehicle i; X i is the longitudinal position of the lane-changing trajectory of vehicle i; a k (k = 0, 1, 2,..., 5) are the curve shape parameters, and their values will be introduced later.

[0120] When the longitudinal displacement and lateral displacement of the vehicle reach the specified values D lat and D lon respectively, it is considered that the vehicle has completed a lane change. To enable the vehicle to achieve a better lane-changing effect, D lat = 3.75 and D lon = 100 can be taken. The values of the curve shape parameters in Equation (23) are:

[0121] a0 = a1 = a2 = 0 (24)

[0122]

[0123] Vehicle yaw angle where v xi and ν yi are the longitudinal speed and lateral speed of vehicle i respectively. Then the lateral position and yaw angle of the desired trajectory can be obtained:

[0124]

[0125] S3.3, AMPC lateral controller

[0126] When the vehicle platoon changes lanes or steers, considering that the payload mass and payload distribution of heavy commercial vehicles are prone to change, adaptive model predictive control (AMPC) is used to control the platoon movement. The AMPC lateral controller monitors the desired trajectories related to the yaw angles and lateral positions of the vehicles in the platoon. Using the vehicle dynamics model and the MPC controller whose prediction horizon changes with the vehicle speed, the ideal front-wheel steering angle of the vehicle is determined in real time.

[0127] According to the vehicle dynamics model, the state variable ξ can be expressed by the following formula:

[0128]

[0129] where v yi , v xi are the lateral and longitudinal speeds of vehicle i respectively; is the yaw angle of vehicle i; y i , x i are the lateral and longitudinal positions of vehicle i respectively.

[0130] The lateral control input u and output η of the AMPC lateral controller are defined as follows:

[0131]

[0132] where δ fi is the front-wheel steering angle of vehicle i.

[0133] According to Equation (4) and Equation (29), the discretized lateral control variable equation of the vehicle can be written as the following formula:

[0134]

[0135] From this, the discrete state-space expression can be obtained:

[0136]

[0137] Wherein, P(t) = I + TP(t), Q(t) = TQ(t), I is the identity matrix, T = 1 / T href , P(t) and Q(t) are respectively the Jacobian matrices of the function f of the state variable ξ(t) and the control quantity u(t), and their expressions are as follows:

[0138]

[0139] Wherein:

[0140]

[0141] The objective function J of the AMPC lateral controller ilat is defined as follows:

[0142]

[0143] Wherein, η i (k|t) represents the output of vehicle i, and ΔU i (k|t) represents the control increment of vehicle i, and N P and N c are the sizes of the prediction horizon and the control horizon; η ref (k|t) is the reference output; M and N are respectively the weight matrices of the output variable and the control increment value ΔU(t). In order to generate a feasible solution within the specified time range, the above function is additionally enhanced by using a relaxation factor ε corresponding to the weighting factor ρ, and ε satisfies 0 ≤ ε ≤ M′, where M′ is the upper limit of ε and is taken as 10.

[0144] In the controller, the front wheel steering angle is used as the lateral control quantity, and its increment is restricted to meet the requirements of ride comfort. The constraint conditions are as follows:

[0145]

[0146] Wherein, u max , u min and ΔU max , ΔU min are respectively the upper and lower boundary values of the control quantity and the control increment of the AMPC lateral controller.

[0147] The output quantity should also satisfy the following upper and lower boundary constraints:

[0148] η min ≤ η i (k|t) ≤ η max (38)

[0149] Wherein, η min and η maxThey are the upper and lower boundary values of the output variable, η min and η max The value ranges of min satisfy η max ∈[-0.3, -1], η

[0150] To prevent heavy commercial vehicles from rolling over, a lateral acceleration constraint is added to ensure the safety of the queue, as shown in the following formula:

[0151]

[0152] In the formula, μ is the road surface adhesion coefficient; g is the acceleration due to gravity; is the lateral acceleration of vehicle i.

[0153] To improve the adaptability of the lateral controller to the longitudinal vehicle speed v xi a model predictive control (MPC) controller is designed to adjust the prediction horizon according to the change of the longitudinal vehicle speed. The relationship between the prediction horizon and the longitudinal vehicle speed is fitted to a cubic polynomial curve, N P The specific control law is as follows:

[0154]

[0155] In the formula, k1, k2, k3 and k4 are the coefficients and constant terms in the polynomial fitting curve, and their values are k1 = -0.0000429, k2 = 0.0116, k3 = -0.6944, k4 = 40 respectively.

[0156] From this, the problem description formula of the lateral controller can be obtained:

[0157]

[0158] When a control cycle ends, the next control cycle is entered, and the loop iteration is performed to achieve the tracking of the required lane-changing trajectory.

[0159] The present invention also provides an automatic driving heavy-duty commercial vehicle queue control system based on a low-earth orbit satellite network, including:

[0160] A heavy-duty commercial vehicle dynamics model and a non-linear tire model establishment module, which respectively establish a five-degree-of-freedom vehicle dynamics model and a non-linear tire model of a commercial vehicle;

[0161] A commercial vehicle queue model establishment module, which establishes a commercial vehicle queue model based on low-earth orbit satellite communication;

[0162] The longitudinal and lateral control module of the queue of autonomous heavy-duty commercial vehicles performs longitudinal motion control through distributed model predictive control and lateral motion control through adaptive model predictive control based on the constructed dynamic model, non-linear tire model, and commercial vehicle queue model.

[0163] The described embodiments are the preferred embodiments of the present invention, but the present invention is not limited to the above embodiments. Without departing from the substantial content of the present invention, any obvious improvements, substitutions, or variations that those skilled in the art can make fall within the protection scope of the present invention.

Claims

1. A method for controlling a platoon of heavy-duty commercial vehicles using an autonomous driving system based on a low-orbit satellite network, characterized in that: The lead vehicle in a platoon of autonomous heavy-loaded commercial vehicles makes driving state decisions by receiving information about the road conditions ahead from low-orbit satellites: when the road ahead is a straight road, the platoon maintains longitudinal following driving within the lane with an appropriate vehicle spacing, and performs longitudinal motion control through distributed model predictive control; when there is a need to change lanes, the platoon changes lanes along a quintic polynomial curve path under the guidance of the lead vehicle, and performs lateral motion control through adaptive model predictive control.

2. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 1 is characterized in that: Longitudinal motion control is performed through distributed model predictive control. Specifically, the distributed model predictive controller decomposes the overall optimization problem of the queue into sub-optimization problems. Each follower vehicle in the queue receives information about the preceding vehicle and the pilot vehicle from the low-orbit satellite, and calculates the optimal control variables for each vehicle through MPC optimization, thereby realizing coordinated longitudinal motion control of the queue.

3. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 2 is characterized in that: The distributed model predictive controller is: In the formula, J ilon (t) is the longitudinal controller objective function, J i1 is the error cost function between vehicle i and the preceding vehicle, and The distance between the following vehicle i and the preceding vehicle at time t+k is the position of the front vehicle, is the position of vehicle i, L is the length of the vehicle, is the expected distance between vehicle i and the preceding vehicle, v i-1，i (k|t) is the longitudinal speed error between the vehicle and the preceding vehicle at the prediction moment, w1 is the adjustment coefficient of the distance error with the preceding vehicle, w2 is the adjustment coefficient of the speed error with the preceding vehicle, J i2 is the error cost function between vehicle i and the pilot vehicle, and is the position of the pilot car at the t+kth moment, is the position of vehicle i, is the expected inter-vehicle distance between the pilot vehicle and vehicle i, Assume a longitudinal velocity for the pilot car, is the predicted longitudinal speed of vehicle i, w3 is the error adjustment coefficient with the pilot vehicle, w4 is the error adjustment coefficient with the pilot vehicle speed, is vehicle i at t+N P The predicted position at the time, is the assumed terminal position of vehicle j in the queue, is the expected terminal distance between vehicle j and vehicle i, n is the number of vehicles in the queue, is vehicle i at t+N P The predicted speed at the moment, is the assumed terminal velocity of vehicle j in the queue, ν max The upper speed limit for heavy-duty commercial vehicles. is the predicted acceleration of vehicle i at time t+k, a max is the upper limit of acceleration, a min is the lower limit of acceleration, i=2, 3, 4, ..., n.

4. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 1, characterized in that: The quintic polynomial curve path is: Where Y i is the lateral position of the lane-changing trajectory of vehicle i, X i is the longitudinal position of the lane-changing trajectory of vehicle i, and the curve shape parameters a0=a1=a2=0, D lat Specify a value for the longitudinal displacement of the vehicle, D lon Specifies a value for the lateral displacement of the vehicle.

5. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 4 is characterized in that: The lateral position and yaw angle of the fifth-order polynomial curve path are respectively:

6. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 5 is characterized in that: Lateral motion control is performed through adaptive model predictive control, specifically: an adaptive model predictive lateral controller monitors the path of each vehicle in the queue and determines the ideal front wheel steering angle of the vehicle in real time using a vehicle dynamics model and an MPC controller that predicts changes in time domain with vehicle speed.

7. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 6 is characterized in that: The adaptive model predicts the lateral controller as: Among them, J ilat The adaptive model predicts the objective function of the lateral controller, η i (k|t) represents the output of vehicle i, ΔU i (k|t) represents the control increment of vehicle i, η ref (k|t) is the reference output, N P 、N c are the sizes of the prediction time domain and the control time domain, M and N are the weight matrices of the output variable and the control increment value ΔU(t), ρ is the weighting factor, ε is the relaxation factor, and u max 、u min are the upper and lower boundary values ​​of the control quantity of the adaptive model predicting the lateral controller, ΔU max , ΔU min are the upper and lower boundary values ​​of the control increment of the adaptive model predictive lateral controller, η min , η max are the upper and lower boundary values ​​of the output variable, μ is the road adhesion coefficient, g is the gravitational acceleration, is the lateral acceleration of vehicle i.

8. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 7 is characterized in that: The prediction time domain satisfies: In the formula, k1, k2, k3 are coefficients in the polynomial fitting curve, k4 is the constant term in the polynomial fitting curve, and v xk is the longitudinal speed of vehicle i.

9. The method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network according to claim 1, characterized in that: The member vehicles of the autonomous heavy-duty commercial vehicle fleet communicate based on low-orbit satellites. Each vehicle receives signals from the low-orbit satellite and sends its own status information to the low-orbit satellite.

10. A system for implementing the method for controlling a platoon of heavy-duty commercial vehicles based on an autonomous driving low-orbit satellite network as described in any one of claims 1 to 9, characterized in that: include: Heavy-duty commercial vehicle dynamics model and nonlinear tire model establishment module, used to establish a five-degree-of-freedom vehicle dynamics model and a nonlinear tire model for commercial vehicles respectively; Commercial vehicle platoon model building module, used to build a commercial vehicle platoon model based on low-orbit satellite communications; The longitudinal and lateral control module of the autonomous driving heavy-load commercial vehicle platoon is based on the constructed dynamic model, nonlinear tire model, and commercial vehicle platoon model. It performs longitudinal motion control through distributed model predictive control and lateral motion control through adaptive model predictive control.

Images