AGV (Automatic Guided Vehicle) control system and method for emergency carrying of environmental chamber test vehicle

Abstract

The invention relates to an AGV (Automatic Guided Vehicle) control system for emergency carrying of a test vehicle in an environmental chamber, which comprises an emergency management module, an automatic unbinding module and a vehicle moving AGV module, and is characterized in that the automatic unbinding module is used for binding the test vehicle on a stand column and immediately unbinding after receiving an emergency carrying instruction, so that the test vehicle can be carried by the AGV; and the vehicle moving AGV module comprises an AGV control sub-module and an AGV execution sub-module. The invention further discloses an AGV control method for emergency carrying of the environmental chamber test vehicle. According to the AGV control system and method designed for the emergency carrying scene of the test vehicle in the environmental chamber, the vacancy of a current AGV emergency carrying control system is made up, rapid response and processing under the emergency situation can be effectively handled, and meanwhile the stability and safety of the carrying process are guaranteed; according to the invention, the emergency processing capability of the system is enhanced, the operation process of the AGV is optimized, the dependence on operators is reduced, and the possibility of human errors is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of AGV control, and in particular to an AGV control system and method for emergency transportation of an environmental chamber test vehicle. Background Art

[0002] With the continuous advancement of science and technology and the in-depth development of industrial automation, Automated Guided Vehicle (AGV), as an intelligent transportation equipment, is increasingly widely used in modern manufacturing, logistics warehousing, environmental chamber testing and other fields. AGV can realize unmanned handling operations, improve production efficiency, reduce labor costs, and ensure the safety of operations in complex or dangerous environments. Especially in the application scenarios of environmental chamber test vehicles, since test vehicles often need to be tested in a closed and strictly controlled environment, traditional manual handling methods are not only inefficient, but also pose safety risks. Therefore, it is particularly important to develop an AGV control system and method suitable for emergency handling of environmental chamber test vehicles.

[0003] At present, although there are many AGV control systems on the market, most of them are designed for conventional logistics handling and lack support for rapid response and handling capabilities in emergency situations. In the field of environmental chamber test vehicles, once an emergency occurs, such as fire, harmful gas leakage, etc., the test vehicle needs to be quickly transferred to a safe area to avoid casualties and property losses. However, existing AGV systems often cannot meet this demand for fast, accurate and stable handling. In addition, the special conditions in the environmental chamber, such as the control of temperature, humidity, pressure, etc., put higher requirements on the adaptability and reliability of the AGV system. These special conditions may affect the positioning accuracy, communication stability and planning and control of the handling path of the AGV, thereby affecting the efficiency and safety of the entire handling process. Summary of the invention

[0004] In order to solve the problem that the current AGV lacks support for rapid response and processing capabilities in emergency situations, the primary purpose of the present invention is to provide an AGV control system for emergency transportation of environmental chamber test vehicles that can effectively respond to and handle emergency situations, while ensuring the stability and safety of the transportation process, enhancing the system's emergency processing capabilities, optimizing the AGV's operating procedures, reducing dependence on operators, and reducing the possibility of human error.

[0005] To achieve the above purpose, the present invention adopts the following technical solution: an AGV control system for emergency transportation of an environmental chamber test vehicle, comprising:

[0006] Emergency management module, responsible for issuing emergency management instructions;

[0007] Automatic unbinding module, used to bind the test vehicle to the column, and immediately unbind it upon receiving an emergency transport command, so that the test vehicle can be transported by the AGV;

[0008] The vehicle transfer AGV module is used to complete the emergency transportation of the test vehicle;

[0009] The emergency management module comprises:

[0010] The monitoring and analysis unit monitors the battery pack cells of the test vehicle through CAN communication, monitors the battery pack cell temperature in real time and analyzes it, and issues an alarm when it determines that the status is abnormal;

[0011] The decision management unit immediately stops the test upon receiving the alarm information and asks the test personnel whether to perform the emergency handling operation. If the result of the inquiry is yes or there is no feedback within a period of time, an emergency handling instruction is sent to the vehicle moving AGV module and the automatic unbinding module;

[0012] The vehicle moving AGV module includes:

[0013] AGV control submodule, used to control the lifting and walking of AGV;

[0014] AGV execution submodule is used to perform emergency handling actions.

[0015] The AGV control submodule includes:

[0016] Lifting control unit, used to control the AGV to lift and unload the test vehicle;

[0017] The navigation and positioning unit is used to provide posture data for the AGV's automatic tracking and positioning control when it dives under the test vehicle;

[0018] Tracking control unit, used to control the AGV to travel along a specified path;

[0019] The AGV execution submodule includes:

[0020] A lifting actuator, which is composed of a front lifting actuator and a rear lifting actuator, wherein the front lifting actuator is a scissor-type lifting mechanism, and the rear lifting actuator is a height-adjustable rear wheel suspension;

[0021] The travel actuator is composed of a pair of front driving wheels and a pair of rear driven wheels. The front driving wheels are two independent steering wheels for realizing active driving and steering control; the suspension height of the rear driven wheels is adjustable;

[0022] Chassis, used for installing and connecting the lifting actuator and the traveling actuator;

[0023] The pair of front driving wheels are fixed below the front lifting actuator, and the pair of rear driven wheels are fixed below the rear lifting actuator.

[0024] The automatic unbinding module includes a motor, an electric push rod, a locking pin mechanism and a rope with a hook, and the locking pin mechanism includes a pin, a lock buckle and a base. The motor and the base are fixed on the column of the environmental chamber, and a first rotating shaft and a second rotating shaft are arranged on the base. The axes of the first rotating shaft and the second rotating shaft are perpendicular. The pin is hinged to the base through the first rotating shaft, and the lock buckle is hinged to the base through the second rotating shaft. Lock hooks are arranged at both ends of the rope with a hook, one end of the lock hook is buckled on the pin, and the other end of the lock hook is buckled at a designated position of the test vehicle. The pin has two states of inward locking and outward unlocking, and the lock buckle has two states of upward locking and downward unlocking. The output shaft of the motor is fixedly connected to one end of the electric push rod, and the motor drives the electric push rod to have two states of extended and retracted. When the electric push rod is in the retracted state, the pin is located in the groove provided in the base, and the end of the pin is against the recess provided in the lock buckle; when the electric push rod is in the extended state, the lock buckle is unlocked downward, and the pin is unlocked outward.

[0025] Another object of the present invention is to provide a control method for an AGV control system for emergency transport of an environmental chamber test vehicle, the method comprising the following steps in sequence:

[0026] (1) The monitoring and analysis unit in the emergency management module is responsible for monitoring and analyzing the status information of the test vehicle, and issues an alarm when an abnormal state is detected. After receiving the alarm, the decision management unit immediately stops the test and asks the test personnel whether to perform the emergency handling operation. If the result of the inquiry is yes or there is no feedback within a period of time, an unbinding instruction is sent to the automatic unbinding module, and an emergency handling instruction is sent to the vehicle moving AGV module;

[0027] (2) The automatic unbinding module unbinds the test vehicle according to the unbinding instruction;

[0028] (3) The AGV control submodule dispatches the AGV according to the emergency handling instruction, controls the AGV to dive under the test vehicle, support the chassis of the test vehicle, lift the test vehicle, and follow the specified path;

[0029] (4) When the AGV arrives at the transport destination, it unloads the test vehicle and sneaks out to escape; when the AGV escapes to a safe location, the emergency transport mission of the test vehicle is completed, and the status of the AGV is checked. If it is in good condition, it will be recycled.

[0030] In step (3), the tracking of the designated route specifically includes the following steps:

[0031] (3a) Establishing a neural network prediction model and setting the structure and parameters of the neural network prediction model: The neural network prediction model is used to predict the AGV posture in the future time domain based on the current and historical motion state information of the AGV; the motion state information includes the lateral speed V of the AGV y, yaw angular velocity w and the input sequence of the AGV in the future time domain; the input sequence of the AGV in the future time domain includes the longitudinal velocity V x and the steering wheel angle δ; the AGV posture includes the heading angle Yaw, the lateral coordinate X and the longitudinal coordinate Y in the vehicle coordinate system;

[0032] (3b) Data collection: Various sensors are deployed on the AGV to obtain the actual vehicle data of the AGV's position, speed and yaw rate under multiple working conditions, and these data are pre-processed to form a training data set for training the neural network prediction model; the various sensors include GPS global positioning system, IMU inertial sensor, camera and radar;

[0033] (3c) Using the training data set to train the neural network prediction model and verify the training effect;

[0034] (3d)Select is the state quantity, is the control quantity, Δu(t)=u(t)-u(t-1) is the control quantity increment, η(t)=[Yaw k , X k , Y k ] is the output variable; use f NN Represents the trained neural network prediction model, and the relationship between the state quantity, control quantity and output variable is obtained:

[0035] η(t)=f NN (ξ(t),u(t))

[0036] Where i = [tN h , t-(N h -1),…,t-1,t],N h represents the historical time domain step; d = [t]; j = [t+1, ..., t+N c -1, t+N c ],N c represents the control time domain step; k = [t+1, ..., t+(N p -1), t+N p ],N p Indicates the prediction time domain step;

[0037] According to the output variable η(t) and the reference value η in the prediction time domain r (t) Establish a multi-objective optimization function J(t):

[0038]

[0039] Where ε is the weight factor; Q and R are weight matrices;

[0040] Considering mechanical limitations, control smoothness, and handling stability, corresponding constraints are set respectively; thus, a trajectory tracking predictive control model based on neural network predictive control is constructed. The formula of the trajectory tracking predictive control model is:

[0041]

[0042]

[0043] In the formula, α is the limit value of the control amount, and β is the limit value of the control amount increment;

[0044] (3e) Combined with the state quantities corresponding to the real-time motion state of the AGV, the future motion posture of the AGV is predicted, and then the iterative optimization algorithm is used to solve the multi-objective optimization problem in combination with the constraints set by formula (2). The optimal control sequence of the longitudinal speed and steering wheel angle of the AGV is obtained by real-time rolling optimization. The first one of the optimal control sequence is taken as the real-time control quantity. The formula for iterative optimization using the gradient descent method is as follows:

[0045]

[0046] Among them, u * (t) represents the optimized control quantity at the current moment, u * (t-1) represents the control quantity after optimization at the last moment, l r Represents the learning rate.

[0047] It can be seen from the above technical scheme that the beneficial effects of the present invention are: First, the AGV control system and method designed for the emergency transportation scenario of the environmental chamber test vehicle of the present invention make up for the vacancy of the current AGV emergency transportation control system, and can effectively cope with the rapid response and processing in emergency situations, while ensuring the stability and safety of the transportation process; Second, the present invention proposes to use a neural network prediction model to build a trajectory tracking prediction control model, which simplifies the analysis of the AGV physical model and can effectively improve the trajectory tracking control effect of the AGV in emergency situations, which not only enhances the emergency handling capability of the system, but also optimizes the operation process of the AGV, reduces the dependence on operators, and reduces the possibility of human errors; Third, the present invention takes into account the convenience and flexibility of actual operation, so that it can maintain good adaptability and scalability in different emergency scenarios; Fourth, the present invention has significant advantages in improving emergency transportation efficiency and ensuring transportation safety, and provides an efficient and reliable solution for related fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 It is a block diagram of the system composition of the present invention;

[0049] Figure 2 is a flow chart of the method of the present invention;

[0050] Figure 3 This is a schematic diagram of the overall installation of the automatic unbinding module in the environmental chamber of the present invention;

[0051] Figure 4 It is a schematic diagram of partial installation of the automatic unbinding module on the column in the present invention;

[0052] Figure 5 This is a schematic diagram of the locked state of the automatic unbinding module in the present invention;

[0053] Figure 6 This is a schematic diagram of the unlocking state of the automatic unbinding module in the present invention;

[0054] Figure 7 It is a structural schematic diagram of the AGV execution submodule in the present invention;

[0055] Figure 8 It is a structural schematic diagram of the neural network prediction model in the present invention;

[0056] Fig. 9 It is the working flow chart of NN-MPC tracking control unit;

[0057] Fig.10 This is a schematic diagram of the AGV emergency handling path tracking scenario. DETAILED DESCRIPTION

[0058] like Figure 1 As shown, an AGV control system for emergency transport of an environmental chamber test vehicle includes:

[0059] Emergency management module, responsible for issuing emergency management instructions;

[0060] The automatic unbinding module 16 is used to bind the test vehicle to the column 7, and immediately unbind it upon receiving an emergency transport instruction, so that the test vehicle can be transported by the AGV 8. Figure 3 As shown;

[0061] The vehicle transfer AGV module is used to complete the emergency transportation of the test vehicle;

[0062] The emergency management module comprises:

[0063] The monitoring and analysis unit monitors the battery pack cells of the test vehicle through CAN communication, monitors the battery pack cell temperature in real time and analyzes it, and issues an alarm when it determines that the status is abnormal;

[0064] The decision management unit immediately stops the test upon receiving the alarm information and asks the test personnel whether to perform the emergency handling operation. If the result of the inquiry is yes or there is no feedback within a period of time, an emergency handling instruction is sent to the vehicle moving AGV 8 module and the automatic unbinding module 16;

[0065] The vehicle moving AGV module includes:

[0066] AGV control submodule, used to control the lifting and walking of AGV 8;

[0067] AGV execution submodule is used to perform emergency handling actions.

[0068] The AGV control submodule includes:

[0069] Lift control unit, used to control AGV 8 to lift and unload the test vehicle;

[0070] The navigation and positioning unit is used to provide posture data for the automatic tracking of AGV 8 and the positioning control when diving under the test vehicle;

[0071] A tracking control unit, used to control the AGV 8 to travel along a specified path;

[0072] like Figure 7 As shown, the AGV execution submodule includes:

[0073] The lifting actuator is composed of a front lifting actuator 11 and a rear lifting actuator 12, wherein the front lifting actuator 11 is a scissor-type lifting mechanism, and the rear lifting actuator 12 is a height-adjustable rear wheel suspension;

[0074] The walking actuator is composed of a pair of front driving wheels 9 and a pair of rear driven wheels 10. The pair of front driving wheels 9 are two independent steering wheels for realizing active driving and steering control; the suspension height of the pair of rear driven wheels 10 is adjustable;

[0075] The chassis 13 is used for installing and connecting the lifting actuator and the walking actuator;

[0076] The pair of front driving wheels 9 are fixed below the front lifting actuator 11 , and the pair of rear driven wheels 10 are fixed below the rear lifting actuator 12 .

[0077] like Figure 4 , Figure 5 , Figure 6As shown, the automatic untying module 16 includes a motor 15, an electric push rod 4, a locking pin mechanism 14 and a hook rope, the locking pin mechanism 14 includes a pin 2, a lock 3 and a base 1, the motor 15 and the base 1 are fixed on the column 7 of the environmental chamber, a first rotating shaft and a second rotating shaft are arranged on the base 1, the axes of the first rotating shaft and the second rotating shaft are perpendicular, the pin 2 is hinged to the base 1 through the first rotating shaft, and the lock 3 is hinged to the base 1 through the second rotating shaft; the hook rope includes a rope 6, both ends of which are provided with lock hooks 5, and one end of the lock hook 5 is buckled on the pin 2, and the other end of the lock hook is buckled at the specified position of the test vehicle; the pin 2 has two states of inward locking and outward unlocking, the lock catch 3 has two states of upward locking and downward unlocking, the output shaft of the motor 15 is fixedly connected to one end of the electric push rod 4, and the motor 15 drives the electric push rod 4 to have two states of extension and retraction. When the electric push rod 4 is in the retracted state, the pin 2 is located in the groove provided in the base 1, and the end of the pin 2 is against the recess provided in the lock catch 3; when the electric push rod 4 is in the extended state, the lock catch 3 is unlocked downward, and the pin 2 is unlocked outward.

[0078] The core component of the automatic untying module 16, namely the locking pin mechanism 14, is fixed on the column 7 of the environmental chamber. One end of the pin 2 rotates around the fixed axis. During normal operation, the other end of the pin 2 is locked by the lock buckle 3, and the lock hook 5 is buckled on the pin 2. When an emergency occurs, the lock buckle 3 that blocks the rotation of the pin 2 is opened by pushing the electric push rod 4, and then under the action of the tension of the rope 6, one end of the pin 2 automatically pops out to realize the automatic untying of the test vehicle.

[0079] like Figure 2 As shown, the method comprises the following steps in order:

[0080] (1) The monitoring and analysis unit in the emergency management module is responsible for monitoring and analyzing the status information of the test vehicle, and issues an alarm after detecting an abnormal state. After receiving the alarm, the decision management unit immediately stops the test and asks the test personnel whether to perform the emergency handling operation. If the result of the inquiry is yes or there is no feedback within a period of time, an unbinding instruction is sent to the automatic unbinding module 16, and an emergency handling instruction is sent to the vehicle moving AGV 8 module;

[0081] (2) the automatic unbinding module 16 unbinds the test vehicle according to the unbinding instruction;

[0082] (3) The AGV control submodule dispatches the AGV 8 according to the emergency handling instruction, controls the AGV 8 to drive under the test vehicle, supports the test vehicle chassis 13, lifts the test vehicle, and tracks the designated path;

[0083] (4) When AGV 8 arrives at the transport destination, it unloads the test vehicle and sneaks out to escape; when AGV 8 escapes to a safe location, the emergency transport mission of the test vehicle is completed, and the status of AGV 8 is checked. If it is in good condition, it is recycled.

[0084] In step (3), the tracking of the designated route specifically includes the following steps:

[0085] (3a) Figure 8 As shown, a neural network prediction model is established, and the structure and parameters of the neural network prediction model are set: the neural network prediction model is used to predict the AGV posture in the future time domain based on the current and historical motion state information of AGV 8; the motion state information includes the lateral speed V of AGV 8 y , yaw angular velocity w and the input sequence of the future time domain of AGV 8; the input sequence of the future time domain of AGV 8 includes the longitudinal velocity V x and the steering wheel angle δ; the AGV posture includes the heading angle Yaw, the lateral coordinate X and the longitudinal coordinate Y in the vehicle coordinate system;

[0086] (3b) Data collection: Various sensors are deployed on the AGV 8 to obtain the actual vehicle data of the position, speed and yaw rate of the AGV 8 under multiple working conditions, and these data are pre-processed to form a training data set for training the neural network prediction model; the various sensors include GPS global positioning system, IMU inertial sensor, camera and radar;

[0087] (3c) Using the training data set to train the neural network prediction model and verify the training effect;

[0088] (3d)Select is the state quantity, is the control quantity, Δu(t)=u(t)-u(t-1) is the control quantity increment, η(t)=[Yaw k , X k , Y k ] is the output variable; use f NN Represents the trained neural network prediction model, and the relationship between the state quantity, control quantity and output variable is obtained:

[0089] η(t)=f NN (ξ(t),u(t))

[0090] Where i = [tN h , t-(N h -1),…,t-1,t],N h represents the historical time domain step; d = [t]; j = [t+1, ..., t+N c -1, t+N c ],N c represents the control time domain step; k = [t+1, ..., t+(N p -1), t+N p ],N p Indicates the prediction time domain step;

[0091] According to the output variable η(t) and the reference value η in the prediction time domain r (t) Establish a multi-objective optimization function J(t):

[0092]

[0093] Where ε is the weight factor; Q and R are weight matrices;

[0094] Considering mechanical limitations, control smoothness, and handling stability, corresponding constraints are set respectively; thus, a trajectory tracking predictive control model based on neural network predictive control, namely, the NN-MPC tracking control unit, is constructed. The formula of the trajectory tracking predictive control model is:

[0095]

[0096]

[0097] In the formula, α is the limit value of the control amount, and β is the limit value of the control amount increment;

[0098] (3e) Fig. 9 As shown in the figure, the future motion posture of AGV 8 is predicted by combining the state quantities corresponding to the real-time motion state of AGV 8. Then, the iterative optimization algorithm is used to solve the multi-objective optimization problem by combining the constraints set by formula (2). The optimal control sequence of the longitudinal speed and steering wheel angle of AGV is obtained by real-time rolling optimization. The first one of the optimal control sequence is taken as the real-time control quantity. The formula for iterative optimization using the gradient descent method is as follows:

[0099]

[0100] Among them, u * (t) represents the optimized control quantity at the current moment, u * (t-1) represents the control quantity after optimization at the last moment, l r Represents the learning rate.

[0101] like Fig.10 As shown, in an emergency, the AGV enters the environmental chamber to lift and transport the test vehicle. Under the control of the NN-MPC tracking control unit, the AGV 8 travels along the desired trajectory, achieves stable and accurate path tracking, and transports the test vehicle to the designated location outdoors.

[0102] In summary, the AGV control system and method designed for the emergency transportation scenario of the environmental chamber test vehicle of the present invention fills the gap in the current AGV 8 emergency transportation control system, can effectively respond to and handle emergency situations quickly, and ensure the stability and safety of the transportation process; the present invention proposes to use a neural network prediction model to build a trajectory tracking prediction control model, which simplifies the analysis of the AGV physical model and can effectively improve the trajectory tracking control effect of AGV 8 in emergency situations, which not only enhances the emergency handling capability of the system, but also optimizes the operation process of AGV 8, reduces dependence on operators, and reduces the possibility of human errors; the present invention takes into account the convenience and flexibility of actual operation, so that it can maintain good adaptability and scalability in different emergency scenarios; the present invention has significant advantages in improving emergency transportation efficiency and ensuring transportation safety, and provides an efficient and reliable solution for related fields.

Claims

1. An AGV control system for emergency transport of environmental chamber test vehicles, characterized in that: include: Emergency management module, responsible for issuing emergency management instructions; Automatic unbinding module, used to bind the test vehicle to the column, and immediately unbind it upon receiving an emergency transport command, so that the test vehicle can be transported by the AGV; The vehicle transfer AGV module is used to complete the emergency transportation of the test vehicle; The emergency management module comprises: The monitoring and analysis unit monitors the battery pack cells of the test vehicle through CAN communication, monitors the battery pack cell temperature in real time and analyzes it, and issues an alarm when it determines that the status is abnormal; The decision management unit immediately stops the test upon receiving the alarm information and asks the test personnel whether to perform the emergency handling operation. If the result of the inquiry is yes or there is no feedback within a period of time, an emergency handling instruction is sent to the vehicle moving AGV module and the automatic unbinding module; The vehicle moving AGV module includes: AGV control submodule, used to control the lifting and walking of AGV; AGV execution submodule is used to perform emergency handling actions.

2. The AGV control system for emergency transport of environmental chamber test vehicles according to claim 1 is characterized in that: The AGV control submodule includes: Lifting control unit, used to control the AGV to lift and unload the test vehicle; The navigation and positioning unit is used to provide posture data for the AGV's automatic tracking and positioning control when it dives under the test vehicle; Tracking control unit, used to control the AGV to travel along a specified path; The AGV execution submodule includes: A lifting actuator, which is composed of a front lifting actuator and a rear lifting actuator, wherein the front lifting actuator is a scissor-type lifting mechanism, and the rear lifting actuator is a height-adjustable rear wheel suspension; The travel actuator is composed of a pair of front driving wheels and a pair of rear driven wheels. The front driving wheels are two independent steering wheels for realizing active driving and steering control; the suspension height of the rear driven wheels is adjustable; Chassis, used for installing and connecting the lifting actuator and the traveling actuator; The pair of front driving wheels are fixed below the front lifting actuator, and the pair of rear driven wheels are fixed below the rear lifting actuator.

3. The AGV control system for emergency transport of environmental chamber test vehicles according to claim 1 is characterized in that: The automatic unbinding module includes a motor, an electric push rod, a locking pin mechanism and a rope with a hook, and the locking pin mechanism includes a pin, a lock buckle and a base. The motor and the base are fixed on the column of the environmental chamber, and a first rotating shaft and a second rotating shaft are arranged on the base. The axes of the first rotating shaft and the second rotating shaft are perpendicular. The pin is hinged to the base through the first rotating shaft, and the lock buckle is hinged to the base through the second rotating shaft. Lock hooks are arranged at both ends of the rope with a hook, one end of the lock hook is buckled on the pin, and the other end of the lock hook is buckled at a designated position of the test vehicle. The pin has two states of inward locking and outward unlocking, and the lock buckle has two states of upward locking and downward unlocking. The output shaft of the motor is fixedly connected to one end of the electric push rod, and the motor drives the electric push rod to have two states of extended and retracted. When the electric push rod is in the retracted state, the pin is located in the groove provided in the base, and the end of the pin is against the recess provided in the lock buckle; when the electric push rod is in the extended state, the lock buckle is unlocked downward, and the pin is unlocked outward.

4. The control method of the AGV control system for emergency transport of an environmental chamber test vehicle according to any one of claims 1 to 3, characterized in that: The method comprises the following steps in order: (1) The monitoring and analysis unit in the emergency management module is responsible for monitoring and analyzing the status information of the test vehicle, and issues an alarm when an abnormal state is detected. After receiving the alarm, the decision management unit immediately stops the test and asks the test personnel whether to perform the emergency handling operation. If the result of the inquiry is yes or there is no feedback within a period of time, an unbinding instruction is sent to the automatic unbinding module, and an emergency handling instruction is sent to the vehicle moving AGV module; (2) The automatic unbinding module unbinds the test vehicle according to the unbinding instruction; (3) The AGV control submodule dispatches the AGV according to the emergency handling instruction, controls the AGV to dive under the test vehicle, support the chassis of the test vehicle, lift the test vehicle, and follow the specified path; (4) When the AGV arrives at the transport destination, it unloads the test vehicle and sneaks out to escape; when the AGV escapes to a safe location, the emergency transport mission of the test vehicle is completed, and the status of the AGV is checked. If it is in good condition, it will be recycled.

5. The control method according to claim 4, characterized in that: In step (3), the tracking of the designated route specifically includes the following steps: (3a) Establishing a neural network prediction model and setting the structure and parameters of the neural network prediction model: The neural network prediction model is used to predict the AGV posture in the future time domain based on the current and historical motion state information of the AGV; the motion state information includes the lateral speed V of the AGV y , yaw angular velocity w and the input sequence of the AGV in the future time domain; the input sequence of the AGV in the future time domain includes the longitudinal velocity V x and the steering wheel angle δ; the AGV posture includes the heading angle Yaw, the lateral coordinate X and the longitudinal coordinate Y in the vehicle coordinate system; (3b) Data collection: Various sensors are deployed on the AGV to obtain the actual vehicle data of the AGV's position, speed and yaw rate under multiple working conditions, and these data are pre-processed to form a training data set for training the neural network prediction model; the various sensors include GPS global positioning system, IMU inertial sensor, camera and radar; (3c) Using the training data set to train the neural network prediction model and verify the training effect; (3d)Select is the state quantity, is the control quantity, Δu(t)=u(t)-u(t-1) is the control quantity increment, η(t)=[Yaw k , X k , Y k ] is the output variable; use f NN Represents the trained neural network prediction model, and the relationship between the state quantity, control quantity and output variable is obtained: η(t)=f NN (ξ(t),u(t)) Where i = [tN h , t-(N h -1),…,t-1,t],N h represents the historical time domain step; d = [t]; j = [t+1, ..., t+N c -1, t+N c ],N c represents the control time domain step; k = [t+1, ..., t+(N p -1), t+N p ],N p Indicates the prediction time domain step; According to the output variable η(t) and the reference value η in the prediction time domain r (t) Establish a multi-objective optimization function J(t): Where ε is the weight factor; Q and R are weight matrices; Considering mechanical limitations, control smoothness, and handling stability, corresponding constraints are set respectively; thus, a trajectory tracking predictive control model based on neural network predictive control is constructed. The formula of the trajectory tracking predictive control model is: In the formula, α is the limit value of the control amount, and β is the limit value of the control amount increment; (3e) Combined with the state quantities corresponding to the real-time motion state of the AGV, the future motion posture of the AGV is predicted, and then the iterative optimization algorithm is used to solve the multi-objective optimization problem in combination with the constraints set by formula (2). The optimal control sequence of the longitudinal speed and steering wheel angle of the AGV is obtained by real-time rolling optimization. The first one of the optimal control sequence is taken as the real-time control quantity. The formula for iterative optimization using the gradient descent method is as follows: Among them, u * (t) represents the optimized control quantity at the current moment, u * (t-1) represents the control quantity after optimization at the last moment, l r Represents the learning rate.

Images