Intelligent vehicle group distributed control device and cooperative control method under communication failure

Abstract

The application discloses a kind of intelligent vehicle group distributed control device and collaborative control method under communication failure.Intelligent vehicle group constitutes distributed control system, and each intelligent vehicle is equipped with positioning module, signal receiving-transmitting device, collaborative controller, motor driver etc..Positioning module is used to obtain the absolute position of intelligent vehicle in environment, and the relative position between vehicles.Signal sending device broadcasts the position information of this intelligent vehicle, and signal receiving device receives the position information from other intelligent vehicles.Collaborative controller is used to realize collaborative control protocol, and uses prediction logic, to compensate the packet loss caused by communication network failure, and the delay generated in signal transmission process.Motor driver is used to convert control action into output torque of driving motor.Compared with existing achievements, the application can significantly overcome the packet loss and information lag problem caused by communication failure and communication delay, thereby improving the stability of the system and enhancing the implementation of the theory.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of multi-agent cooperative control, and particularly relates to an intelligent vehicle group distributed control device and a cooperative control method under communication failure. BACKGROUND

[0002] The research on multi-agent problem is derived from the cluster phenomenon in nature. When migrating, birds arrange into a fixed formation; when grazing, herds gather in a direction under the leadership of a "leader". Each individual contributes to the collective task while achieving its own individual task, and we call this behavior "cooperative" or "coordinated" multi-agent. If an intelligent vehicle is an intelligent agent, then a group of intelligent vehicles is called an intelligent vehicle group system.

[0003] Restricting the communication of individuals in the group so that each individual can only communicate and interact with its neighbors is a kind of distributed interaction idea. Intelligent vehicles communicate with each other to form a communication network with vehicles as nodes and transmission channels as topological edges. On the basis of the communication network, the control network with a control topological structure is formed according to the control relationship. Therefore, the intelligent vehicle group system can also be regarded as a distributed networked control system. Compared with point-to-point communication, transmitting information through a network can greatly improve the flexibility of data transmission, but it also brings problems such as uncertain transmission delay, packet loss, and vulnerability to attacks.

[0004] The use of prediction logic in the implementation of the control protocol can compensate for the uncertainty caused by packet loss and delay. When a fault occurs, the controller generates a control quantity while predicting the evolution of the entire system according to the data received at the previous time. The predicted value reproduces the control value under normal conditions, thereby solving the problem of uncertainty caused by communication failure.

[0005] Patent No. 202080070475.1, entitled "Multi-agent simulation", its features are "design method, system and device, including computer program encoded on computer storage medium, for generating simulation of environment in which multiple agents interact over multiple time steps"; a set of multi-agent simulation interaction platform is constructed, which provides convenience for theoretical analysis and practical application. Patent No. ZL201310713899.7, entitled "Multi-agent traffic signal control system", its features are "overcoming the problem that the existing technology intersection control mode cannot achieve the optimal in practical application, and lacks good coping ability for large changes"; a practical intersection control scheme is provided, which realizes the rationalization of traffic control and relieves the pressure of traffic intersection. Patent No. ZL201810286164.3, entitled "Multi-agent formation method based on behavior selection", its features are "setting a position-related detection benefit calculation method, setting a position-related detection benefit calculation method, determining the number of channels of the basal ganglia, establishing a basal ganglia channel model, initializing the relevant parameters to determine the number of channels of the basal ganglia, establishing a basal ganglia channel model, initializing the relevant parameters"; a dynamic agent formation method based on behavior selection is introduced, which effectively solves the shortcomings of existing formation detection methods such as weak robustness and weak reliability.

[0006] The above prior art studies different practical application problems from different angles. However, in the existing data, there is no application of prediction technology to intelligent vehicle group system, and there is no application case of using prediction logic to overcome the uncertainty problem caused by communication failure. For cooperative control, there are also few technical means that can directly apply theory to specific models. SUMMARY

[0007] The purpose of the present application is to provide an intelligent vehicle group distributed control device and a cooperative control method under communication failure. It can effectively solve the influence of packet loss and time delay in communication process on system performance, thereby enhancing the reliability of the system and providing protection for building more complex systems.

[0008] The application discloses a kind of intelligent vehicle group distributed control device under communication failure, including positioning module, signal receiving-transmitting module, cooperative controller, motor driver;Positioning module is installed in the position of intelligent vehicle top end signal interference less;The output end of positioning module is connected signal sending module;Signal sending module is used for signal processing and zero point transformation, own position information is directly transmitted to local controller on one hand, and is broadcasted to environment by signal sending device on the other hand;Signal receiving device completes the reception of signal and processes signal, and signal receiving device is directly connected to cooperative controller;Cooperative controller is single-chip microcomputer, corresponding control law is written according to control protocol, and the function of prediction compensation is realized simultaneously;Four wheels of intelligent vehicle are all drive wheels, and use mecanum wheel structure, and the motor corresponding to each wheel is controlled separately.

[0009] The cooperative control method of the intelligent vehicle group distributed control device under communication failure based on the application comprises the following steps:

[0010] Step S1: create basic conditions for cooperative control and prediction operation, and model intelligent vehicle;

[0011] Step S2: establish cooperative control protocol;

[0012] Step S3: obtain a prediction scheme based on the cooperative control protocol.

[0013] The intelligent vehicle group of the application constitutes a distributed control system, and each intelligent vehicle is equipped with a positioning module, a signal receiving-transmitting device, a cooperative controller and a motor driver. The positioning module is used to obtain the absolute position of the intelligent vehicle in the environment and the relative position between vehicles. The signal sending device broadcasts the position information of the intelligent vehicle, and the signal receiving device receives the position information from other intelligent vehicles. The cooperative controller is used to implement the cooperative control protocol and adopt prediction logic to compensate for packet loss caused by communication network failure and delay generated in the signal transmission process. The motor driver is used to convert the control effect into the output torque of the driving motor. Compared with the existing achievements, the application can significantly overcome the problems of packet loss and information lag caused by communication failure and communication delay, thereby improving the stability of the system and enhancing the implementation of the theory.

[0014] The intelligent vehicle group distributed control device and the cooperative control method under communication failure mentioned in the application can effectively overcome the uncertainty generated by packet loss and delay in the communication process of the system. The device is safe and reliable, the method is quick and effective, and is beneficial to popularization. The application has high popularization value in economy, safety, practicability and reliability. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 It is the structural schematic diagram of the distributed control device of the application.

[0016] Figure 2 is a model schematic diagram of the intelligent vehicle model I of the present application.

[0017] Figure 3 is a model schematic diagram of the intelligent vehicle model II of the present application.

[0018] Figure 4 is a cooperative controller program flow chart of the present application.

[0019] Figure 5 is a schematic diagram of three typical cooperative control targets. DETAILED DESCRIPTION

[0020] The implementation of the device and the method will be described in detail below in connection with the drawings.

[0021] 1. System structure

[0022] An intelligent vehicle group distributed control device under communication failure, comprising a positioning module, a signal receiving-transmitting module, a cooperative controller, and a motor driver. The positioning module is installed at the top of the intelligent vehicle where the signal interference is small. According to the distance between the vehicles, a GPS module can be considered to be used. If the distance between the vehicles is small, a camera can be installed in the moving environment to perform positioning through image algorithm. The output end of the positioning module is connected to the signal transmitting module. The signal transmitting module performs signal processing and zero point conversion. The self-position information is directly transmitted to the local controller on one hand, and is broadcasted to the environment by the signal transmitting device on the other hand. The signal receiving-transmitting process can be implemented by using a Zigbee module, or through WIFI and Bluetooth. The signal receiving device completes the signal receiving and performs filtering, amplification, A / D conversion and other processing on the signal. It is directly connected to the cooperative controller. The cooperative controller is implemented by using a single-chip microcomputer. According to the control protocol, the corresponding control law is written, and the function of prediction compensation is realized. The four wheels of the intelligent vehicle are driving wheels. In order to enhance the flexibility of movement, a Mecanum wheel structure is adopted.

[0023] 2. System modeling

[0024] In order to create basic conditions for cooperative control and prediction operation, the intelligent vehicle needs to be modeled. If the Mecanum wheel is used for each wheel of the intelligent vehicle, the kinematic model of the vehicle driving can be represented as follows:

[0025]

[0026] wherein V x and V y are the moving speeds of the intelligent vehicle platform in the horizontal and vertical directions respectively; ω0 is the rotation angular speed of the platform. P is the center point of the platform, and ω i(i = 1, 2, 3, 4) are the angular velocities of the four Mecanum wheels. 2L1 and 2L2 are the width and length of the platform, respectively. The wheel radius is R, and the angle between the roller axis and the wheel axis in space is a. The model I of the intelligent vehicle is shown in the accompanying drawings Figure 2 where O0, x0, y0 are the translated origin and the x and y axis directions of the moving coordinate system, respectively. If a = 45° is considered, the above model is transformed into:

[0027]

[0028] The above formula shows that the desired motion speed and rotational speed of the intelligent vehicle platform can be achieved by controlling the rotational speeds of the four Mecanum wheels. It can be found by observation that ω1 and ω2 change the same increment at the same time, and V x and ω0 can be changed alone without changing V y ; ω2 and ω4 change opposite increments at the same time, and V y and ω0 can be changed alone without changing V x .

[0029] The above is a model from the working mechanism of the Mecanum wheel, and the coordinate system used is a moving coordinate system, that is, the forward direction of the intelligent vehicle is always the y direction, and the vertical direction is the x direction. If a fixed reference coordinate system is used, the speed model of the intelligent vehicle can be described by the following expression:

[0030]

[0031] where v is the direction of the platform movement; ω is the angular velocity of the platform rotation; θ is the angle formed by the forward direction of the platform and the positive direction of the x axis, representing the rate of change. The model II of the intelligent vehicle is shown in the accompanying drawings Figure 3 .

[0032] The above is equivalent to constructing a speed change model of the intelligent vehicle platform, and the x direction speed and the y direction speed are independent and do not affect each other, so they can be analyzed alone. In system modeling, if the angle of the platform is not considered, the states of the two directions can be regarded as the same order but different dimensions, and the two dimensions can be regarded as two independent subsystems: x system and y system. Taking the x system as an example, the kinematic model of the intelligent vehicle can be constructed as follows:

[0033]

[0034] where x(t) is the position of the intelligent vehicle in the x direction; v(t) is the speed of the intelligent vehicle in the x direction; and respectively, u(t) is the control input signal in x direction. This is a double integrator model, which can be written in state space representation as:

[0035]

[0036] where G = [0 1; 0 0] is the system matrix, B = [0 1] is the input matrix T . Thus far, the model of the intelligent vehicle is constructed.

[0037] 3. Cooperative control protocol

[0038] The cooperative controller is a distributed controller, and each intelligent vehicle is equipped with a cooperative controller. The cooperative controller forms the control action according to the control protocol and completes the prediction logic. According to the characteristics of the distributed control, the control variable of each intelligent vehicle needs to be generated by means of the information of the intelligent vehicle itself and other intelligent vehicles. For different control protocols, the control objectives achieved are not the same. The specific form of the control protocol needs to be determined according to the control task. The algorithm flow chart of the cooperative controller is shown in FIG. 2. Figure 4 .

[0039] The control topology information of the communication network is saved in the controller in advance, and the information includes the adjacency matrix A and the number of each intelligent vehicle. The controller obtains the position information x i of the intelligent vehicle itself and the position information x j of the intelligent vehicles in the neighborhood relationship. By using the information, the control action is generated according to the control protocol U i = f(x i , x j , A).

[0040] 4. Prediction scheme

[0041] In order to achieve the control objective, the system is sampled at {0h, 1h, …, kh, …} respectively, and h is the sampling interval. In order to simplify the description, k is used instead of kh. In order to effectively solve the uncertainty influence of the packet loss and time delay in the communication process on the system, a prediction mechanism is introduced.

[0042] The cooperative controller first analyzes the data receiving process. If the waiting time exceeds h, it can be identified that the data is lost, and the packet loss prediction is made accordingly. The prediction process is mainly one-step prediction according to the last stored data on the controller, and then the stored data is updated. If continuous packet loss occurs, the packet loss prediction process will be performed multiple times. As follows:

[0043]

[0044] Where d is the total number of packet loss; G and B represent the system matrix and input matrix in the intelligent vehicle model respectively; x(·|k) and u(·|k) represent the prediction of the state and control input at the (·) sampling time at the kth sampling time respectively.

[0045] If the waiting time is less than h, the delay case can be handled. The prediction process is to obtain the timestamp information l k from the received data packet by analysis, and then subtract the current time t k to obtain the delay length τ k . Since the intelligent vehicle system contains many intelligent vehicle platforms, each prediction operation will consume some time, therefore, the step of prediction needs to be additionally increased by 1, so the final prediction step is τ k +1. Each prediction is performed, and the latest prediction value is stored for the next prediction.

[0046] The implementation of the device and the method will be described in detail below in combination with the drawings. Consistency control, formation control and containment control are the most common and classical control objectives in the field of multi-agent cooperation. The illustrations and differences of the three are shown in the attached Figure 5 .

[0047] The program logic is shown in the attached Figure 3 . The system continuously detects whether data is received, if the data is not received within the specified time h, the delay processing is performed; otherwise, the packet loss processing is performed. The content of the delay processing is to calculate the transmission delay τ k , and then predict the state and control input for τ k +1 steps; the packet loss processing is to perform one step of prediction through the last stored data. The last of the two processing needs to update the currently stored data.

[0048] Example 1: Formation control task under network attack.

[0049] This embodiment realizes the formation control task of the intelligent vehicle group system under the condition of considering the packet loss caused by denial of service (DoS) attack. It is assumed that the system contains 7 intelligent vehicles, so the control topology formed by the 7 intelligent vehicles must contain a directed spanning tree, otherwise the "island phenomenon" will occur and cooperation cannot be achieved.

[0050] The distributed control device of each intelligent vehicle is shown in Figure 1The positioning module scans the environment to determine the position of the platform, and then transmits the position of the platform to the sending device. The sending device processes and encodes the signal, and packages the signal with timestamp information and broadcasts the signal to the intelligent vehicle group and the local collaborative controller. After the signal receiving device receives the position signal sent by other intelligent vehicles, the signal receiving device decodes the signal, detects the integrity of the data, and sends the signal to the local collaborative controller. The controller uses a single-chip microcomputer and uses the following control protocol:

[0051]

[0052] In the formula, U Xi and U Yi are control actions in the x direction and the y direction, respectively. xi , x yi and v xi , v yi are positions and speeds in the x direction and the y direction, respectively. xi , f yi are the formation information of the intelligent vehicle itself, i.e., the position to which the intelligent vehicle is expected to move. xj , f yj are the formation information of the neighboring intelligent vehicles. ij k1 and k2 are coupling strengths, and a k is an element corresponding to the position in the adjacency matrix A. is a neighboring intelligent vehicle node adjacent to the intelligent vehicle platform node in the control topology.

[0053] The control action is sent to the motor driver of the intelligent vehicle, which is converted into the torque output of the motor, and finally controls the speed and position of the intelligent vehicle.

[0054] Suppose that the network is subjected to a DoS attack during the transmission of information in the communication network. The DoS attack hinders the transmission process, causing the data transmission to be interrupted, and finally causing the loss of data packets. The prediction mechanism of the present application can effectively compensate for the packet loss caused by the DoS attack. Assuming that the attack occurs at time l k , and lasts for 10 hours, then the controller will detect data every time, and the result will be judged as "packet loss". Therefore, the packet loss prediction strategy will also be repeated 10 times, and after each prediction, the latest prediction value will be taken as the latest state of the system, and the control amount will be calculated according to the latest state according to the control protocol, and then sent to the motor driver to convert into the driving amount (torque).

[0055] The coupling strength can be solved by solving a linear matrix inequality. By deriving the stability theorem of the system, and selecting an optimal solution that satisfies the performance index in the feasible region described by the theorem using the LMI method. In addition to solving the coupling strength, other parameters that need to be designed can also be solved.

[0056] Example 2: Containment control with time delay

[0057] This example realizes the containment control. There are multiple leader vehicles and several follower vehicles in the system. Containment control means that all follower vehicles are finally in the convex hull formed by the leader vehicles. Containment control has a wide application prospect in the military field. For example, a cluster of unmanned armored vehicles with different equipment, a cruise vehicle equipped with radar as a leader, and a missile vehicle as a follower. In this way, the measurement range of the radar can be maximized, and the follower can be maximized. The protection is provided to ensure the safety of the vehicle fleet.

[0058] The system structure of this example is similar to that of example 1, but the difference is that: a) the content of the control protocol, the control protocol no longer contains formation information; b) the system contains multiple leader vehicles. The following control protocol can be used to achieve the containment control goal.

[0059]

[0060] where the definitions of the parameters are the same as those involved in the control protocol of example 1. In addition, the information transmission network will produce a certain delay. Therefore, the prediction method can be used to infer the information under normal transmission according to the received information, so as to overcome the influence of delay on the system.

[0061] Example 3: Consensus control task under network attack

[0062] Consensus control is a typical application of networked control. Consensus means that the output or state of each subsystem in the network reaches consensus. Consensus is usually divided into leaderless consensus and leader-follower consensus. The difference is that the final consensus state of the former is related to the initial value of all subsystems, while the final state of the latter is the same as the state of the leader agent. The most typical consensus protocol is as follows:

[0063]

[0064] where U Xio and U Yio are the control actions in the x and y directions, respectively; n is the order of the intelligent vehicle system; x xli , x xlj and x yli , x ylj are the lth order states in the x and y directions, respectively. d xl and d yl are the lth order states of the virtual leader in the x and y directions, respectively. δ j={0,1} represents whether the intelligent agent has a direct connection with the virtual leader in the communication topology, and o represents the oth-order control input of the system.

[0065] The system structure of the embodiment is similar to that of the first embodiment, and the difference is that: a) the control protocol adopted by the system is different, and the virtual leader role is introduced; b) there is at most only one subsystem with the leader identity in the system.

[0066] The distributed cooperative device and method under communication failure in the application can effectively overcome the uncertainty of the system under the conditions of packet loss and time delay in the communication process. The device is safe and reliable, the method is fast and effective, and is beneficial to popularization. The application has high popularization value in economy, safety, practicability and reliability.

Claims

1. A collaborative control method for a distributed control device for intelligent vehicle groups under communication failure, characterized in that... Includes the following steps: Step S1: Create basic conditions for coordinated control and predictive operation by modeling the intelligent vehicle; Step S2: Establish a collaborative control protocol; Step S3: Obtain the prediction scheme based on the cooperative control protocol; the specific process is as follows: Step S31: Determine the model for each intelligent vehicle that satisfies the conditions for predicting the system evolution; Step S32: If packet loss occurs during the communication process, then a single-step prediction is made based on the data stored at the previous moment before the packet loss occurred; if continuous packet loss occurs, then the prediction operation will be performed in multiple steps. Step S33: If a time delay occurs during the communication process, then based on the timestamp information contained in the received data packet... k , and the current time t k The time extension τ can be obtained by subtracting the time extension. k Then the prediction step size should be τ. k +1; To achieve the control objective, the system is sampled at times {0h, 1h, ..., kh, ...}, where h is the sampling interval. For simplicity, k is used instead of kh. The collaborative controller first analyzes the data reception process. If the waiting time exceeds h, it determines that data loss has occurred and makes a packet loss prediction accordingly. The prediction process mainly involves making a prediction based on the data previously stored by the controller, and then updating the stored data. If continuous packet loss occurs, the packet loss prediction process will be performed multiple times: Where d is the total number of packet loss events; G and B represent the system matrix and input matrix in the intelligent vehicle model, respectively; x(·|k) and u(·|k) represent the prediction of the state and control input at sampling time (·) at sampling time k, respectively; If the waiting time is less than h, it is handled as a delay; the prediction process involves obtaining the timestamp information l by parsing the received data packets. k and the current time t k By subtracting, we can obtain the time extension τ. k Since the intelligent vehicle system contains several intelligent vehicle platforms, each prediction calculation takes time. Therefore, the prediction step size needs to be increased by 1, so the final prediction step size is τ. k +1; Each time a prediction is made, the latest predicted value must be stored for the next prediction.

2. The collaborative control method of the intelligent vehicle group distributed control device under communication failure as described in claim 1, characterized in that... The modeling of the intelligent vehicle in step S1 above is specifically carried out as follows: Modeling is based on the working mechanism of the Mecanum wheel, using a moving coordinate system, where the y-axis is always the direction of the intelligent vehicle's forward movement, and the x-axis is perpendicular to it. Each intelligent vehicle uses a Mecanum wheel, and the kinematic model describing the vehicle's drive is as follows: Among them, V x and V y ω0 represents the lateral and longitudinal movement speeds of the intelligent vehicle platform, respectively; ω0 is the rotational angular velocity of the intelligent vehicle platform; P is the center point of the intelligent vehicle platform; ω0... i Let be the rotational angular velocity of the four Mecanum wheels, where i = 1, 2, 3, 4; 2L1 and 2L2 are the width and length of the intelligent vehicle platform, respectively; the wheel radius is R; and the angle between the roller axle on the Mecanum wheel and the wheel axle in space is α. If α = 45°, then the above model is transformed into: The above formula shows that the desired speed and rotation speed of the intelligent vehicle platform can be achieved by controlling the rotation speed of the four Mecanum wheels; A velocity variation model of the intelligent vehicle platform is constructed. The x-direction velocity and y-direction velocity of the intelligent vehicle platform are independent and do not affect each other, so they are analyzed separately. Using a fixed reference coordinate system, the velocity model of the intelligent vehicle is characterized by the following expression: Where v is the direction of movement of the intelligent vehicle platform; ω is the angular velocity of rotation of the intelligent vehicle platform; and θ is the angle between the forward direction of the intelligent vehicle platform and the positive x-axis. This represents its rate of change.

3. The collaborative control method of the intelligent vehicle group distributed control device under communication failure as described in claim 2, characterized in that... The establishment of the collaborative control protocol in step S2 above is specifically as follows: Step S21: Construct a communication network and assign a number to each intelligent vehicle platform. Preload the control topology reflecting the intelligent vehicle system, including the adjacency matrix A: (a ij =1,a ii =0), node V: {v1,v2,…,v n } and topological edge ε: (v i ,v j The system provides information on the intelligent vehicles in the system and assigns leader or follower attributes to each vehicle based on its position in the topology and its actual function. Step S22: The controller acquires its own position information x fed back by the signal transmitting device. i ; Step S23: Obtain the location information x of intelligent vehicles that are neighbors of this intelligent vehicle in the control topology. j ; Step S24: Based on different collaborative objectives, adopt the corresponding control protocol U. i =f(x) i ,x j A) Generation control function.

4. A distributed control device for intelligent vehicle groups implementing the cooperative control method under communication failure as described in claim 1, characterized in that... Includes a positioning module, a signal transceiver module, a coordination controller, and a motor driver; The positioning module is installed at the top of the intelligent vehicle in a location with minimal signal interference. The output of the positioning module is connected to the signal transmitting module. The signal transmitting module performs signal processing and zero-point transformation, transmitting its own position information directly to the local controller and broadcasting it to the environment via the signal transmitting device. The signal receiving device receives and processes the signal, and is directly connected to the collaborative controller. The collaborative controller is a microcontroller that implements corresponding control laws according to the control protocol, while also performing predictive compensation. All four wheels of the intelligent vehicle are drive wheels, employing a Mecanum wheel structure, with each wheel's corresponding motor controlled independently.

Images