Longitudinal formation control method for multiple autonomous platforms under dynamic communication topology

Abstract

The invention discloses a longitudinal formation control method for multiple autonomous platforms under dynamic communication topology, which comprises the following steps: firstly, constructing a target function, then describing a multi-platform formation control problem as an optimization problem P1 of the target function, solving the optimization problem P1 based on a distributed model prediction control method, obtaining an optimal control sequence, and finally obtaining a multi-platform formation control result. And autonomous platform longitudinal formation control under the dynamic communication topology condition is realized. Multi-platform longitudinal formation control can still be successfully carried out under the condition that the communication topology changes in real time, meanwhile, longitudinal formation control of the autonomous platform under the dynamic communication topology is verified through a simulation test, engine torque tracking control of a leader autonomous platform and distance control of the autonomous platform and a front platform are achieved, and the reliability of the autonomous platform is improved. According to the method, the speed of an autonomous platform and engine torque follow-up control are followed, inevitable state constraints, control input constraints and collision avoidance constraints in practical application can be processed, performance indexes are optimized, and the method has good practicability.

Description

Technical Field

[0001] The present invention belongs to the technical field of longitudinal control of multiple autonomous platforms under dynamic communication topology, and in particular relates to a longitudinal formation control method of multiple autonomous platforms under dynamic communication topology. Background Art

[0002] Multi-autonomous platform longitudinal formation control is an advanced traffic management technology that aims to enable a group of autonomous platforms to form a tight formation on the road and maintain the desired spacing and the same speed through communication and collaboration between intelligent systems and autonomous platforms. First, each platform in the multi-autonomous platform senses the position and other information of the autonomous platforms in front and around it in real time through the sensor system it carries, and then calculates the control input based on the algorithm. The autonomous platform adjusts its own speed and direction through the automatic control system to maintain the desired spacing between the autonomous platforms in front and behind, thereby forming a stable formation state. At the same time, the autonomous platforms need to maintain the same speed, and achieve speed synchronization and coordination through the communication system between the autonomous platforms to ensure the stability and safety of the formation operation.

[0003] The control of multiple autonomous platforms is usually divided into two categories: centralized control and distributed control. Centralized multi-autonomous platform column control refers to the formation and control of the entire multi-autonomous platform by a central controller. The controller is responsible for guiding the driving speed and spacing of all autonomous platforms to achieve the column operation of multiple autonomous platforms. The centralized controller can globally consider the operating status and road conditions of multiple autonomous platforms to achieve more flexible scheduling and formation control. However, it has high requirements for the reliability and processing power of the central controller. Once the controller fails, the operation of the entire multi-autonomous platform may be seriously affected. A good communication infrastructure and reliable communication connection are required to timely transmit control instructions and autonomous platform status feedback information. Distributed multi-autonomous platform column control, on the other hand, realizes the autonomous adjustment and formation control of multiple autonomous platforms through communication and coordination between autonomous platforms. The distributed control method has high fault tolerance. Each platform can make decisions and adjustments independently. The failure of one platform will not have a significant impact on other autonomous platforms. It does not rely on the central controller, which reduces the complexity of the system and the need for centralization. However, the distributed control method has high requirements for the communication and algorithm design of the autonomous platform ship, and it is necessary to ensure the accuracy and reliability of information transmission and coordination between autonomous platforms. Distributed control design The negotiation and decision-making of autonomous platform ships may lead to delays and inconsistencies in the formation process.

[0004] The formation control of autonomous multi-platforms, whether centralized or distributed, requires information exchange between autonomous platforms. Nowadays, based on new communication technologies represented by 5G, intelligent transportation systems are also developing rapidly. Existing research results all assume that multiple autonomous platforms have a fixed communication topology, but in actual environments, network communication topology occasionally fails. Once a failure occurs, the control method will no longer be applicable, resulting in failure of autonomous platform column control. Therefore, existing research is not very suitable for solving this actual situation.

[0005] Model predictive control is a method that can handle various constraints, such as input-output constraints, state constraints, etc., to ensure that the system operates within a safe boundary. Distributed model predictive control combines the advantages of model predictive control and distributed control. It can optimize performance indicators while satisfying constraints. It is also an online solution method that can cope with real-time changes and new requirements of the system during the task process. Summary of the invention

[0006] The purpose of the present invention is to provide a longitudinal formation control method for multiple autonomous platforms under dynamic communication topology, aiming to solve the above problems and realize longitudinal formation control of autonomous platforms under dynamic communication topology by using distributed model predictive control. The autonomous platform can be an unmanned autonomous platform.

[0007] The present invention is mainly achieved through the following technical solutions:

[0008] A longitudinal formation control method for multiple autonomous platforms under a dynamic communication topology is provided, which realizes longitudinal formation control of multiple autonomous platforms based on distributed model predictive control, and comprises the following steps:

[0009] Step S1: Construct the objective function based on the control target:

[0010]

[0011] Where: j = 1 means it is the leader’s autonomous platform. 1j =1,c 2j =0;

[0012] j≠1 means it is following the autonomous platform. 1j =0,c 2j =1;

[0013] c 1j is the coefficient of the difference between the engine torque of the leader autonomous platform and the expected engine torque;

[0014] c 2j is the coefficient of the engine torque difference term between the rear platform and the front platform;

[0015] d is the expected distance between the front platform and the rear platform;

[0016] Q1-Q4 are the weight matrices for each item in the objective function;

[0017] T q,Lead The engine torque for the leader’s autonomous platform;

[0018] is the expected engine torque of the leader autonomous platform;

[0019] T q,j-1 is the engine torque of the j-1th autonomous platform;

[0020] T q,j is the engine torque of the jth autonomous platform;

[0021] v j-1 is the speed of the j-1th autonomous platform;

[0022] v j is the speed of the jth autonomous platform;

[0023] s j-1 is the longitudinal distance of the j-1th autonomous platform;

[0024] s j is the longitudinal distance of the jth autonomous platform;

[0025] u j is the control input;

[0026] At each sampling time k, the communication status is detected, and α(k) and β(k) represent communication indication signals; α(k) = 1 means that the autonomous platform can exchange information with the front platform at time k;

[0027] α(k) = 0 means that the autonomous platform cannot communicate with the front platform at time k;

[0028] β(k) = 1 means that the autonomous platform can exchange information with the leader autonomous platform at time k; β(k) = 0 means that it cannot communicate with the leader autonomous platform;

[0029] Step S2: Describe the multi-autonomous platform control problem as an optimization problem P1 of the objective function:

[0030]

[0031] Define the state vector x(k) = [sv T] T , then the state update equation is:

[0032] x j '(t)=ax j (k)+bu j (k)(7

[0033] Where: a and b are the system matrix and control matrix of the continuous system respectively;

[0034] State constraints:

[0035] x j (k)∈X j (8)

[0036] Where: X j is the state constraint set;

[0037] Control input constraints:

[0038] u j (k)∈U j (9)

[0039] Among them: U j is the control quantity constraint set;

[0040] State initialization equation:

[0041] x j (0|k)=x j (k)(10)

[0042] Safety distance constraints between autonomous platforms:

[0043] ||S j-1 -S j ||>R (11)

[0044] Where: R is the safety distance;

[0045] Step S3: Solve the optimization problem P1 based on the distributed model predictive control method, obtain the optimal control sequence, and realize the longitudinal formation control of the autonomous platform based on the optimal control sequence.

[0046] In order to better implement the present invention, further, in step S2, T is taken as the sampling period, and the following discretized system matrices A and B are obtained:

[0047]

[0048] Where: a and b are the system matrix and control matrix of the continuous system respectively;

[0049] I is the identity matrix;

[0050] T is the sampling period.

[0051] Thus, the continuous system can be discretized into the following form:

[0052] x j (k+1)=Ax j (k)+Buj (k).

[0053] In order to better implement the present invention, further, in step S1, the goal is to track the speed and engine torque of the predecessor while maintaining a predefined spacing, and consider the engine torque tracking of the leader autonomous platform. The leader autonomous platform considers the engine torque tracking control and constructs the control target:

[0054]

[0055] in: is the desired engine torque of the autonomous platform.

[0056] In order to better implement the present invention, further, in the step S1, the control target also includes the control of fuel performance indicators.

[0057] In order to better implement the present invention, further, it is applied to the coordinated formation control of multiple unmanned autonomous platforms, including the following steps:

[0058] Step A1: Initialize state information x j (0|k)=x j (k);

[0059] Step A2: Based on the onboard equipment detecting the communication situation, determining the communication indication signals α(k) and β(k), and receiving the status information of the previous unmanned platform or the leader;

[0060] Step A3: Solve the optimization problem P1 to obtain the optimal control sequence;

[0061] Step A4: Apply the first control quantity in the optimal control sequence to the current system, encapsulate its own state information, and transmit it to the subsequent unmanned platform;

[0062] Step A5: Determine whether the end condition is met. If not, return to step A1. If so, end.

[0063] In order to better implement the present invention, further, the unmanned autonomous platform determines Q1, Q2, Q3, Q4, R, d, T in the offline stage. Leadd (k) Parameter.

[0064] The beneficial effects of the present invention are as follows:

[0065] (1) The present invention is applicable to the longitudinal formation control of multiple autonomous platforms under dynamic communication topology, and adopts distributed model predictive control, which can handle the inevitable state constraints, control input constraints, and collision avoidance constraints in practical applications, and optimize the performance indicators. The present invention can successfully perform multi-autonomous platform formation control when the communication topology changes in real time. Simulation experiments verify the longitudinal formation control of 5 platforms under dynamic communication topology, realize the engine torque tracking control of the leading autonomous platform, the distance control between the autonomous platform and the front platform, and the speed and engine torque control of the following autonomous platform, which has good practicality.

[0066] (2) The present invention is based on distributed model predictive control. Each autonomous platform autonomously uses local models and prediction information to achieve multi-autonomous platform platoon control. Model predictive control uses the dynamic model of the system to formulate the optimal control strategy by predicting the future state. This enables the system to be optimized within the prediction range to achieve better performance. In addition, model predictive control can handle nonlinear and complex systems because it can use various models to describe the dynamic characteristics of the system and can be optimized for different models, which has good practicality. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 Schematic diagram of the communication topology of multiple autonomous platforms for complete communication;

[0068] Figure 2 A schematic diagram of the communication topology of multiple autonomous platforms that can only communicate with the leader autonomous platform;

[0069] Figure 3 A schematic diagram of the communication topology of multiple autonomous platforms that can only communicate with the front platform;

[0070] Figure 4 A schematic diagram of the communication topology of multiple autonomous platforms that do not communicate at all;

[0071] Figure 5 It is a schematic diagram of the structure of the distributed multi-autonomous platform longitudinal control system;

[0072] Figure 6 It is the motion trajectory curve of the unmanned autonomous platform;

[0073] Figure 7 It is the distance change curve between unmanned autonomous platforms;

[0074] Figure 8 This is the speed change curve of the unmanned autonomous platform;

[0075] Fig. 9 The engine torque curve for the leader;

[0076] Fig.10The engine torque variation curve for all unmanned autonomous platforms;

[0077] Fig.11 It is the control input variation curve. DETAILED DESCRIPTION

[0078] Embodiment 1:

[0079] A longitudinal formation control method for multiple autonomous platforms under a dynamic communication topology comprises the following steps:

[0080] (1) Communication topology

[0081] like Figure 1-Figure 4 As shown, the present invention will assume four communication topologies, including perfect communication, communication only with the leader autonomous platform, communication only with the front platform, and no communication at all. Figure 1 As shown in the figure, when there is no communication failure in each autonomous platform, the position, speed and engine torque information of the leading autonomous platform and the front platform can be received through communication and platform-mounted sensors. Considering the actual multi-autonomous platform driving scenario, various communication problems may exist, which may lead to information exchange failure or communication link interruption. Figure 1 The communication situation degenerates to Figure 2 In the communication situation shown in the figure, each autonomous platform can only exchange information with the leader platform, and it degenerates into Figure 3 In the communication situation shown, the autonomous platform can only exchange information with the front platform. Figure 4 As shown in the figure, the communication is completely failed, and the autonomous platform can only perceive the speed and position information of the platform in front through the platform-mounted sensors.

[0082] (2) Problem Modeling

[0083] Consider the following form of the autonomous platform motion equation model:

[0084]

[0085] Where S j (t) represents the longitudinal distance of the jth autonomous platform, v j (t) represents speed, T q,j represents the autonomous platform engine torque; is the derivative of the distance of the jth autonomous platform at time t;

[0086] is the derivative of the velocity of the jth autonomous platform at time t;

[0087] is the derivative of the engine torque of the jth autonomous platform at time t;

[0088] τ is the force coefficient;

[0089] i o,j is a scalar coefficient value;

[0090] η j for mechanical efficiency;

[0091] m j is the quality of the jth autonomous platform;

[0092] are the structural parameters of the autonomous platform;

[0093] u j is the control input;

[0094] The above model can be rewritten as follows:

[0095]

[0096] in represents the state vector, u j represents the control input, and the following system matrices a and b:

[0097]

[0098] The above continuous system equation is discretized using the zero-order hold method and can be transformed into the following form:

[0099] x j (k+1)=Ax j (k)+Bu j (k) (3)

[0100] Taking T as the sampling period, the following discretized system matrices A and B can be obtained:

[0101]

[0102] Where: I is the identity matrix;

[0103] T is the sampling period.

[0104] (3) Design of multi-platform longitudinal formation control algorithm based on distributed model predictive control

[0105] Distributed model predictive controllers transmit some information between controllers in a distributed control structure, so that local controllers have some understanding of the other's behavior. Figure 5 As shown, in the multi-autonomous platform system of the present invention, these stand-alone subsystems will make autonomous decisions and controls based on their respective local models and prediction information, and in order to achieve coordination and overall optimization, these subsystems need to interact to jointly achieve system-level control goals.

[0106] An important goal of the multi-autonomous platform system control is to track the speed and engine torque of the predecessor while maintaining a predefined spacing. The engine torque tracking of the leader autonomous platform is also considered. The leader autonomous platform considers engine torque tracking control and defines The desired engine torque curve for the autonomous platform is constructed as follows:

[0107]

[0108] Where: s j-1 is the longitudinal distance of the j-1th autonomous platform;

[0109] s j is the longitudinal distance of the jth autonomous platform;

[0110] v j-1 is the speed of the j-1th autonomous platform;

[0111] v j is the speed of the jth autonomous platform;

[0112] T q,j-1 is the engine torque of the j-1th autonomous platform;

[0113] T q,j is the engine torque of the jth autonomous platform;

[0114] T q,Lead The engine torque for the leader’s autonomous platform;

[0115] is the expected engine torque of the leader autonomous platform;

[0116] At the same time, in order to save fuel consumption in heavy autonomous platform applications, the present invention also takes fuel performance indicators into consideration, so the objective function of the distributed model predictive control can be constructed as follows:

[0117]

[0118] When j = 1, it indicates the leader’s autonomous platform. 1j =1,c 2j = 0, when j≠1, it means following the autonomous platform. At this time, c 1j =0,c 2j =1.

[0119] At each sampling time k, the platform-mounted device detects the communication status, and uses α(k) and β(k) to represent the communication indication signal:

[0120] α(k) = 1 represents the situation where the autonomous platform can exchange information with the front platform at time k.

[0121] α(k)=0 means that it is impossible to communicate with the previous platform.

[0122] β(k) = 1 means that at time k, the autonomous platform can exchange information with the leader’s autonomous platform.

[0123] β(k) = 0 means that it is impossible to communicate with the leader autonomous platform.

[0124] In summary, the multi-platform longitudinal formation control problem based on model predictive control can be described in the following form:

[0125]

[0126] stx j (k+1)=Ax j (k)+Bu j (k) (7)

[0127] x j (k)∈X j (8)

[0128] u j (k)∈U j (9)

[0129] x j (0|k)=x j (k) (10)

[0130] ||S j-1 -S j ||>R (11)

[0131] Wherein, formula (7) is the state update equation, formulas (8)-(9) represent the state constraints and control input constraints, formula (10) is the state initialization, and formula (11) is the safety distance constraint between autonomous platforms.

[0132] Embodiment 2:

[0133] A longitudinal formation control method for multiple autonomous platforms under a dynamic communication topology. To achieve collaborative formation control of multiple unmanned autonomous platforms, the problem is first modeled and constructed as a mathematical problem. This step has been described in the above research content. The following takes the jth unmanned autonomous platform as an example to introduce the problem-solving steps in detail.

[0134] Offline stage: Determine that the system parameters have controlled the constant parameters, including Q1, Q2, Q3, Q4, R, d,

[0135] Online stage: The specific steps are as follows:

[0136] ① Initialize status information x j (0|k)=x j (k);

[0137] ② Based on the communication status detected by the airborne equipment, the communication indication signals α(k) and β(k) are determined, and the status information of the unmanned platform or the leader is received;

[0138] ③Solve the optimization problem P1 and obtain the optimal control sequence;

[0139] ④ Apply the first control quantity in the optimal control sequence to the current system, and encapsulate its own state information and transmit it to the subsequent unmanned platform;

[0140] ⑤ Determine the simulation end condition. If it is not met, return to step ①. If it is met, end.

[0141] like Figure 6 As shown in the figure, it can be seen that the five trajectories have no intersection, indicating that collision avoidance between unmanned platforms has been achieved. The five curves are basically parallel and the spacing is basically equal. The longitudinal formation control under the dynamic communication topology is achieved under the proposed algorithm. Figure 7 As shown in Figure 1, the distance between the unmanned autonomous platforms is controlled to the desired distance of about 25. Figure 8 As shown in the figure, it can be seen that the speed values ​​of the five unmanned platforms are very different and the change trends are consistent, indicating that the proposed algorithm realizes the speed cooperative control under the dynamic communication topology. Fig. 9 As shown in the figure, the solid line is the expected change curve, and the dotted line is the actual value under the control of the proposed algorithm. It can be seen that the engine torque control effect of the leader is very good. Fig.10 As shown in the figure, the unmanned autonomous platform will track the engine torque of the previous unmanned platform or the leader. The effectiveness of the proposed algorithm can be verified from the figure. Fig.11 As shown, the control input change satisfies the constraint of [-120,120].

[0142] The above description is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any simple modification or equivalent change made to the above embodiment based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A longitudinal formation control method for multiple autonomous platforms under dynamic communication topology, characterized in that: The longitudinal formation control of multiple autonomous platforms is realized based on distributed model predictive control, including the following steps: Step S1: Construct the objective function based on the control target: Where: j = 1 means it is the leader’s autonomous platform. 1j =1,c 2j =0; j≠1 means it is following the autonomous platform. 1j =0,c 2j =1; c 1j is the coefficient of the difference between the engine torque of the leader autonomous platform and the expected engine torque; c 2j is the coefficient of the engine torque difference term between the rear platform and the front platform; d is the expected distance between the front platform and the rear platform; Q1-Q4 are the weight matrices for each item in the objective function; T q,Lead The engine torque for the leader’s autonomous platform; is the expected engine torque of the leader autonomous platform; T q,j-1 is the engine torque of the j-1th autonomous platform; T q,j is the engine torque of the jth autonomous platform; v j-1 is the speed of the j-1th autonomous platform; v j is the speed of the jth autonomous platform; s j-1 is the longitudinal distance of the j-1th autonomous platform; s j is the longitudinal distance of the jth autonomous platform; u j is the control input; At each sampling time k, the communication status is detected, and α(k) and β(k) represent the communication indication signals; α(k) = 1 means that the autonomous platform can exchange information with the front platform at time k; α(k) = 0 means that the autonomous platform cannot communicate with the front platform at time k; β(k) = 1 means that the autonomous platform can exchange information with the leader’s autonomous platform at time k; β(k) = 0 means that it is impossible to communicate with the leader autonomous platform; Step S2: Describe the multi-autonomous platform control problem as an optimization problem P1 of the objective function: Define the state vector x(k) = [sv T] T , then the state update equation is: x j '(t)=ax j (k)+bu j (k) (7 Where: a and b are the system matrix and control matrix of the continuous system respectively; State constraints: x j (k)∈X j (8) Where: X j is the state constraint set; Control input constraints: in j (k)∈U j (9) Among them: U j is the control quantity constraint set; State initialization equation: x j (0|k)=x j (k) (10) Safety distance constraints between autonomous platforms: ||S j-1 -S j ||>R (11) Where: R is the safety distance; Step S3: Solve the optimization problem P1 based on the distributed model predictive control method, obtain the optimal control sequence, and realize the longitudinal formation control of the autonomous platform based on the optimal control sequence.

2. The longitudinal formation control method of multiple autonomous platforms under a dynamic communication topology according to claim 1 is characterized in that: In step S2, T is taken as the sampling period, and the following discretized system matrices A and B are obtained: Where: a and b are the system matrix and control matrix of the continuous system respectively; I is the identity matrix; T is the sampling period.

3. The longitudinal formation control method of multiple autonomous platforms under a dynamic communication topology according to claim 1 is characterized in that: In step S1, the goal is to track the speed and engine torque of the predecessor while maintaining a predefined spacing, and consider the engine torque tracking of the leader autonomous platform. The leader autonomous platform considers the engine torque tracking control and constructs the control target: in: is the desired engine torque of the autonomous platform.

4. The longitudinal formation control method of multiple autonomous platforms under a dynamic communication topology according to claim 3 is characterized in that: In the step S1, the control target also includes the control of fuel performance indicators.

5. The longitudinal formation control method of multiple autonomous platforms under a dynamic communication topology according to any one of claims 1 to 4, characterized in that: The collaborative formation control applied to multiple unmanned autonomous platforms includes the following steps: Step A1: Initialize state information x j (0|k)=x j (k); Step A2: Based on the onboard equipment detecting the communication situation, determining the communication indication signals α(k) and β(k), and receiving the status information of the previous unmanned platform or the leader; Step A3: Solve the optimization problem P1 to obtain the optimal control sequence; Step A4: Apply the first control quantity in the optimal control sequence to the current system, encapsulate its own state information, and transmit it to the subsequent unmanned platform; Step A5: Determine whether the end condition is met. If not, return to step A1. If so, end.

6. The longitudinal formation control method of multiple autonomous platforms under a dynamic communication topology according to claim 5 is characterized in that: The unmanned autonomous platform determines Q1, Q2, Q3, Q4, R, d, parameter.

Images