A cluster control method based on perception control and perception tolerance

Abstract

The present invention discloses a cluster control method based on perception control and perception tolerance, aiming to overcome the problem of poor cluster performance in existing cluster intelligent control, which comprises the following steps: configuring cluster information of a group, initializing the number of iterations, configuring individual state information, substituting the state information into a path planning method to obtain path trajectory data; driving individual movement and obtaining a current state target value and an end state target value according to the path trajectory data; calculating the current perception information amount of the individual; updating individual perception attributes; predicting state errors and error correction amounts; updating the end state; updating the path trajectory data, and increasing the number of iterations by one; judging whether the number of iterations is equal to the number of termination iterations, if so, terminating the iteration, otherwise continuing the iteration.

Description

Technical Field

[0001] The present invention belongs to the field of cluster intelligent control, and in particular refers to a cluster control method based on perception control and perception tolerance. Background Art

[0002] In the field of mechanical manufacturing and automation, robots and drones have important research and application significance. With the rise of artificial intelligence and digital communication technology, intelligent robots or aircraft technology has gradually become a multidisciplinary and multi-field research topic. The principle of strength in numbers also applies to intelligent machinery. Generally speaking, robot clusters have higher execution efficiency, which means that more tasks can be completed in the same time, or it takes less time to perform the same task; driving a group requires cluster control technology. Compared with the drive of a single individual, it needs to consider the communication transmission and arrangement structure between cluster individuals.

[0003] Due to the noise and various engineering errors in the real environment, the controlled individuals are often unable to accurately reach the artificially set target state. At this time, additional control methods are needed to deal with the interference factors in the environment. The most common and widely used one is the PID control method, which adjusts the input value according to historical data and the occurrence rate of differences to make the system more accurate and stable. PID can effectively deal with the control errors caused by environmental noise, but it does not perform well when facing the problem of cluster congestion caused by the increase in the number of individuals. How to solve the problem of complex environmental interference and the decline in cluster performance indicators caused by the increase in the number of individuals by building a method model has become the focus of cluster intelligent control research. Summary of the invention

[0004] In order to overcome the deficiencies and existing problems in the prior art, the present invention provides a cluster control method based on perception control and perception tolerance.

[0005] To achieve the above object, the present invention adopts the following technical solution:

[0006] A cluster control method based on perception control and perception tolerance includes the following steps:

[0007] Step 1: The processor configures the cluster information of the group, wherein the cluster information includes the number of individuals, the number of termination iterations, the maximum error of the terminal state and the individual perception attributes, and the individual perception attributes include the limit acceleration, the perception range, the perception tolerance and the iteration step of the path planning method;

[0008] Step 2: The processor initializes the number of iterations and configures individual state information, where the state information includes an initial state and an end state;

[0009] Step 3: The processor substitutes the state information into the path planning method to calculate the path trajectory data, wherein the path trajectory data includes the speed information and position information of the individual in each iteration step, and the path trajectory method can be a path planning method based on the Dubins curve;

[0010] Step 4: The processor drives the individual to move according to the path trajectory data, and obtains the individual's current state target value and terminal state target value according to the path trajectory data index;

[0011] Step 5: The sensor senses the state of the individual where the sensor is located and the states of other individuals within the sensor's sensing range, and calculates the current sensed information amount of the individual based on the state of the sensor and the states of other individuals;

[0012] Step 6: The sensor inputs its own state and the amount of perceived information into the current state control unit, and inputs the amount of perceived information, the end state target value and the perceived tolerance into the perception attribute control unit;

[0013] Step 7: The perception attribute control unit updates the individual perception attribute according to the amount of perception information, the end state target value and the perception tolerance;

[0014] Step 8: The current state control unit predicts the state error and the error correction amount;

[0015] Step 9: The perception attribute control unit inputs the updated individual perception attribute to the sensor and the end state control unit, the current state control unit inputs the state error and the error correction amount to the end state control unit, and the end state control unit updates the end state according to the state error, the error correction amount and the updated end state;

[0016] Step 10: The processor substitutes the updated individual perception attributes and the end point state into the path planning method to update the path trajectory data, and the number of iterations increases by one;

[0017] Step 11: The processor determines whether the number of iterations is equal to the termination number of iterations. If so, the iteration is terminated. If not, steps 4 to 11 are iterated.

[0018] Preferably, the step 5 specifically includes:

[0019] Step 51: The sensor senses the state of the individual where the sensor is located and the states of other individuals within the sensor sensing range;

[0020] Step 52: The sensor calculates the Euclidean distance between the individual and other individuals based on its own state and the states of other individuals;

[0021] Step 53: The sensor calculates the perception angles occupied by other individuals according to the following formula:

[0022]

[0023] Where φ is the perception angle occupied by other individuals, BL is the individual’s own diameter, DisSefOth is the Euclidean distance;

[0024] The perception angles of all other individuals are added together to obtain the overall perception angle, and the overall perception angle is divided by 2π to obtain the amount of perceived information.

[0025] Preferably, the step 7 specifically includes:

[0026] Step 71: The perception attribute control unit compares the perceived information amount and the perception tolerance to obtain the difference between the two, and determines whether the difference between the two is greater than 0. If so, the difference between the two is retained. If not, it is determined whether the absolute value of the difference between the two is less than η times the perception tolerance, where η is an interval parameter for controlling the individual to reversely adjust the individual perception attribute, η∈(0,1). If so, the difference between the two is retained. If not, the difference between the two is set to 0.

[0027] Step 72: The difference between the two is multiplied and then input into the following formula to update the individual perception attribute:

[0028] PecpAtrbNew=PecpAtrbCurr+α·PecpDiff,

[0029] Where PecpAtrbNew is the updated individual perception attribute, PecpAtrbCurr is the current individual perception attribute, and α is the proportional coefficient;

[0030] Step 73: Predict whether the current individual and other individuals will collide based on the path trajectory data and other individual states. If so, set the collision signal to 1; otherwise, set the collision signal to 0.

[0031] Preferably, the step 8 specifically includes:

[0032] Step 81: Subtract the current state target value from the own state to obtain the state error;

[0033] Step 82: Determine whether the number of state errors in the state error container is equal to the length of the error prediction sequence. If so, obtain the error correction amount according to the error prediction model, input the error correction amount into the terminal state control unit, and initialize the state error container. If not, input the state error into the state error container, set the error correction amount to 0 and input it into the terminal state control unit.

[0034] Preferably, the step 9 specifically includes:

[0035] Step 91: Input the collision signal to the end state control unit to determine whether the collision signal is equal to 1, if so, change the end state, if not, do not change the end state;

[0036] Step 92: Compare the modulus of the own state and the modulus of the target value of the terminal state to obtain the difference between the two, and determine whether the difference between the two is less than or equal to the maximum error of the terminal state. If so, change the terminal state, otherwise do not change the terminal state;

[0037] Step 93: Input the error correction amount to the endpoint state control unit and update the endpoint state using the following formula:

[0038] StateEndNew=StateEnd+β·ErrCorrVal,

[0039] In the formula, StateEndNew is the updated end state, StateEnd is the current end state, and β is the error amplification coefficient;

[0040] Step 94: Input the updated limit acceleration and iteration step size to the terminal state control unit, and the terminal state unit substitutes the updated limit acceleration, the updated iteration step size and the updated terminal state into the path trajectory method.

[0041] Compared with the prior art, the present invention has the following outstanding and beneficial technical effects:

[0042] In the present invention, the linkage control of the movement of the group is realized from the perspective of perception control and perception tolerance, so that the group behavior, information interaction and task intention are organically combined, which can significantly improve the efficiency and performance of each individual in the group in completing the task collaboratively, and can flexibly solve complex environments and emergencies, and will not cause cluster congestion problems when the number of individuals increases. Therefore, the present invention has the advantages of high work efficiency, strong robustness, high reliability and excellent cluster performance.

[0043] In the present invention, the cluster control method based on perception control and perception tolerance helps the cluster to effectively cope with complex environmental interference to improve the stability and accuracy of control and solve the problem that the survival rate of the cluster in a closed space decreases rapidly as the number of individuals increases. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the steps of the present invention;

[0045] Figure 2 It is a schematic diagram of the method framework of the present invention; DETAILED DESCRIPTION

[0046] To facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0047] It should be noted that the cluster control method based on perception control and perception tolerance is applied to a swarm robot system to solve the problem of achieving the desired swarm motion by controlling the motion of individuals. The swarm robot system includes a group, which is composed of multiple individuals, and an individual refers to a single robot. A robot refers to an intelligent machine that can work semi-autonomously or fully autonomously.

[0048] like Figure 1 As shown, a cluster control method based on perception control and sensing tolerance includes the following steps:

[0049] Step 1: The processor configures the cluster information of the group, where the cluster information includes the number of individuals (N), the number of termination iterations (TermStepCnt), the maximum error of the terminal state (TermStepCnt) and the individual perception attributes (PecpAtrbCurr), and the individual perception attributes include the maximum acceleration (MaxAcc), the perception range (DetRange), the perception tolerance (PecpLmtTagt) and the iteration step size (IterStepSize);

[0050] In the above steps, the sensing range is the detection area of ​​the sensor. If the individual is within the sensing range, the sensor can detect the state of the individual. If the individual is outside the sensing range, the sensor does not detect the state of the individual.

[0051] Step 2: The processor initializes the number of iterations (AgentIterCnt) and configures the state information of the individual, where the state information includes the initial state (StateInit) and the end state (StateEnd);

[0052] In the above steps, the initial state includes the position and velocity of the individual at the beginning, and the terminal state includes the position and velocity of the individual at the end. The parameter type of the initial state is the same as the parameter type of the terminal state. The number of iterations at the beginning is 0.

[0053] Step 3: The processor substitutes the state information into the path planning method to calculate the path trajectory data (PathTrajSet), wherein the path trajectory data includes the speed information and position information of the individual in each iteration step;

[0054] In the above steps, the parameter type of the initial state, the parameter type of the end state and the parameter type of the input path planning method are the same. The path trajectory method can be pre-set in the processor. The path planning method is unlimited but needs to meet the following conditions:

[0055] (1) The path planning method complies with the laws of kinematics. The path trajectory data is calculated by substituting the initial state and the terminal state into the path planning method. When the path trajectory data is used to drive the individual movement, the individual's movement state presents a continuous change;

[0056] (2) The path trajectory data includes the individual's limit acceleration and iteration step length, where the iteration step length is the number of iterations from step 4 to step 11 within one second;

[0057] (3) The path trajectory data needs to ensure that the mapping relationship between the input (i.e., the initial state and the terminal state) and the output (i.e., the path trajectory data) is unique, that is, the same input can only calculate the same output.

[0058] Step 4: The processor drives the individual to move according to the path trajectory data, and obtains the individual's current state target value (StateCurrTagt) and end state target value (StateEndTagt) according to the path trajectory data index;

[0059] In the above steps, the individual may include at least one of a motor and a servo, and the processor drives the individual to move according to the path trajectory data, that is, the processor controls the motor operation and the servo steering according to the path trajectory data. Since the path trajectory data is a discrete trajectory sequence under an ideal state, the individual's current state target value and the terminal state target value are included in the path trajectory data, and the current state target value and the terminal state target value are obtained by indexing for subsequent iteration. The initial state is the data obtained by indexing in the path trajectory data when the index position is 0, the individual's current state target value is the data obtained by indexing in the path trajectory data when the index position is 1, and the terminal state target value is the data obtained by indexing in the path trajectory data when the index position is the last bit.

[0060] Step 5: The sensor senses the state of the individual where the sensor is located (StateCurrSef) and the states of other individuals within the sensor's sensing range (StateCurrOth), and calculates the current perception information amount of the individual (PecpInfoAmt) based on the state of the sensor and the states of other individuals;

[0061] In the above steps, the sensor is embedded in the individual, and the sensor detects the data of the corresponding individual as the individual's own state. Other individuals are individuals within the sensing range except the individual where the sensor is located. The number of sensors is the same as the number of individuals, and the sensors correspond to individuals one by one, and the sensors are embedded in the corresponding individuals.

[0062] Step 6: The sensor inputs its own state and the amount of perceived information into the current state control unit (StateCurrECU), and inputs the amount of perceived information, the end state target value and the perceived tolerance into the perceived attribute control unit (PecpAtrbECU);

[0063] In the above steps, the current state control unit and the perception attribute control unit are part of the processor.

[0064] Step 7: The perception attribute control unit updates the individual perception attribute according to the amount of perception information, the end state target value and the perception tolerance;

[0065] In the above steps, the perception attribute control unit realizes the iterative adjustment of individual perception attributes according to the amount of perception information, the terminal state target value and the perception tolerance, thereby realizing that the perception attribute control unit can iteratively obtain more realistic individual perception attributes during the movement of the individual.

[0066] Step 8: The current state control unit predicts the state error (StateErr) and the error correction amount (ErrCorrVal);

[0067] In the above steps, the current state control unit realizes iterative detection of state error and error correction amount, thereby realizing that the current state control unit can iteratively detect more realistic state error and error correction amount during the movement of the individual.

[0068] Step 9: The perception attribute control unit inputs the updated individual perception attribute to the sensor and the end state control unit (StateEndTagtECU), the current state control unit inputs the state error and the error correction amount to the end state control unit, and the end state control unit updates the end state according to the state error, the error correction amount and the updated end state;

[0069] In the above steps, the terminal state control unit is a part of the processor. The updated individual perception attributes include the updated limit acceleration, perception range and iteration step. The perception attribute control unit inputs the updated perception range to the sensor, and the perception attribute control unit inputs the updated perception range and iteration step to the terminal state control unit. The state error, error correction amount, updated perception range and iteration step are used to update the terminal state, so as to iteratively realize the individual terminal state to compensate for the environmental error, thereby improving the accuracy of the individual motion state.

[0070] Step 10: The processor substitutes the updated individual perception attributes and the end point state into the path planning method to update the path trajectory data, and the number of iterations increases by one;

[0071] In the above steps, the number of iterations plus one is:

[0072] AgentIterCntNew=AgentIterCnt+1,

[0073] Where AgentIterCntNew is the number of output iterations, and AgentIterCnt is the number of input iterations.

[0074] Step 11: The processor determines whether the number of iterations is equal to the termination number of iterations. If so, the iteration is terminated. If not, iterates from step 4 to step 11.

[0075] In the above steps, steps 4 to 11 are the perception control process, which implements negative feedback control in an iterative manner to facilitate more accurate cluster control.

[0076] The step 5 specifically includes:

[0077] Step 51: The sensor senses the state of the individual where the sensor is located and the states of other individuals within the sensing range.

[0078] Step 52: The sensor calculates the Euclidean distance (DisSefOth) between the individual and other individuals according to its own state and the states of other individuals;

[0079] In the above steps, the self-state includes the location of the current individual (LocSef), the other individual states include the locations of other individuals (AgentOth), and the Euclidean distance is the distance between the location of the individual and the location of other individuals. The calculation formula of the Euclidean distance is:

[0080]

[0081] In the above formula, LocSef.x is the X-axis coordinate of the individual's position, LocSef.y is the Y-axis coordinate of the individual's position, AgentOth.x is the X-axis coordinate of other individuals, and AgentOth.y is the Y-axis coordinate of other individuals.

[0082] Step 53: The sensor calculates the perception angle (φ) occupied by other individuals according to the following formula:

[0083]

[0084] Where φ is the perception angle occupied by other individuals, BL is the individual diameter, and DisSefOth is the Euclidean distance.

[0085] Step 54: The sensor accumulates all other individual perception angles to obtain a total perception angle (PecpAngle), and divides the total perception angle by 2π to obtain the perception information amount;

[0086] In the above steps, the calculation formula for the amount of perceived information is:

[0087]

[0088] The step 7 specifically includes:

[0089] Step 71: The perception attribute control unit compares the perception information amount and the perception tolerance to obtain the difference between the two (PecpDiff), and determines whether the difference between the two is greater than 0. If so, the difference between the two is retained. If not, it is determined whether the absolute value of the difference between the two is less than η times the perception tolerance, where η is an interval parameter for controlling the individual to reversely adjust the individual perception attribute, η∈(0,1). If so, the difference between the two is retained. If not, the difference between the two is set to 0.

[0090] In the above steps, the calculation formula for the difference between the two is:

[0091] PecpDiff=PecpInfoAmt-PecpLmtTagt,

[0092] η can be pre-set in the perception attribute control unit of the processor.

[0093] Step 72: The perception attribute control unit amplifies the difference between the two by K times and then inputs it into the following formula to update the individual perception attribute:

[0094] PecpAtrbNew=PecpAtrbCurr+α·PecpDiff,

[0095] Where PecpAtrbNew is the updated individual perception attribute, PecpAtrbCurr is the current individual perception attribute, and α is the proportional coefficient;

[0096] In the above steps, the perception attribute control unit includes a limit acceleration control unit (MaxAccECU), a perception range control unit (DectRangeECU), and an iteration step control unit (IterStepSizeECU). Since the individual perception attribute includes the limit acceleration, the perception range, and the iteration step, the limit acceleration control unit updates the limit acceleration according to the above formula, the perception range control unit updates the perception range according to the above formula, and the iteration step control unit updates the iteration step according to the above formula. The limit acceleration control unit, the perception range control unit, and the iteration step control unit can respectively use different proportional coefficients α to update the individual perception attribute. The proportional coefficient α can be pre-set in the limit acceleration control unit, the perception range control unit, and the iteration step control unit, respectively. The perception attribute control unit inputs the updated perception range to the sensor, thereby adjusting the perception range of the sensor. The perception attribute control unit inputs the updated limit acceleration and iteration step to the terminal state control unit for subsequent adjustment of the terminal state.

[0097] Step 73: The processor predicts whether the current individual and other individuals will collide based on the path trajectory data and other individual states, and if so, sets the collision signal to 1, otherwise, sets the collision signal to 0;

[0098] In the above steps, the path trajectory data includes the path of the current individual, and the states of other individuals include the speed directions of other individuals. The processor determines whether the path of the current individual and the speed directions of other individuals intersect. If so, it is determined that the current individual and the other individuals will collide. Otherwise, it is determined that the current individual and the other individuals will not collide.

[0099] The step 8 specifically includes:

[0100] Step 81: Subtract the current state target value from the own state to obtain the state error (StateErr);

[0101] Step 82: Determine whether the number of state errors in the state error container is equal to the error prediction sequence length (ErrPredLength); if so, obtain an error correction amount according to the error prediction model, input the error correction amount to the terminal state control unit, and initialize the state error container; if not, input the state error to the state error container, set the error correction amount to 0, and input it to the terminal state control unit;

[0102] In the above steps, the length of the error prediction sequence is related to the error prediction model. The state error container is a part of the processor, and the error prediction model can be a Kalman filter model.

[0103] The step 9 specifically includes:

[0104] Step 91: The processor inputs the collision signal to the end state control unit to determine whether the collision signal is equal to 1, and if so, changes the end state, and if not, does not change the end state;

[0105] In the above steps, the problem of avoiding collision between the current individual and others is achieved.

[0106] Step 92: The processor compares the modulus of its own state with the modulus of the target value of the end state to obtain the difference between the two (CurrEndDiff), and determines whether the difference between the two is less than or equal to the maximum error of the end state. If so, the end state is changed, otherwise, the end state is not changed;

[0107] In the above steps, the difference between the modulus of the own state and the modulus of the target value of the terminal state is obtained by comparing the modulus of the own state and the modulus of the target value of the terminal state:

[0108] The difference between the two = the modulus of the own state - the modulus of the target value of the end state.

[0109] Step 93: The processor inputs the error correction amount to the endpoint state control unit and updates the endpoint state using the following formula:

[0110] StateEndNew=StateEnd+β·ErrCorrVal,

[0111] In the formula, StateEndNew is the updated end state, StateEnd is the current end state, and β is the error amplification coefficient;

[0112] In the above steps, β is pre-set in the processor, and the size of β can be adjusted by actual testing to make the deviation between the individual actual trajectory and the ideal trajectory smaller.

[0113] Step 94: The processor inputs the updated limit acceleration and iteration step size to the terminal state control unit, and the terminal state unit substitutes the updated limit acceleration, the updated iteration step size and the updated terminal state into the path trajectory method.

[0114] like Figure 2 FIG. 1 is a schematic diagram of a method framework provided by the present invention. The present invention is a control method based on perception control and perception tolerance, and includes the following five parts:

[0115] (1) Perception area: The individual obtains its own state and the states of other individuals within the perception range through sensors. With the help of the states of other individuals and its own state, the perception capacity of the current individual can be generated through relevant methods. The individual's own state and perception capacity are used as input signals of the processor. The individual's own state includes the current state and the terminal state. The terminal state will only be perceived and transmitted at the end of a single iteration.

[0116] (2) Target state area: As the reference signal input of the ECU area, the perception tolerance and the initial terminal target state need to be set in advance. The current state target value and the terminal state target value can be generated by the path planning method;

[0117] (3) ECU area: namely, the processor. The decision control and error correction part in the invention relies on the basic control unit and the hierarchical architecture. The perception attribute control unit is the highest layer. The output signal is not only used to control the change amplitude of the output value of the limit acceleration control unit, the perception range control unit and the iteration step control unit, but also to realize the decision change generated when the individual predicts a collision; the current state control unit compares and analyzes the current state value of the individual with the state target value to evaluate the error and other noise in the environment, and outputs a signal to the end state control unit for adjusting the end point information; the end state control unit receives the error analysis reference signal input by the current state control unit on the one hand, thereby adjusting the end point information to counteract environmental factors, and on the other hand, receives the decision conversion reference signal input by the highest layer perception attribute control unit to realize obstacle avoidance operation between individuals in a closed environment;

[0118] (4) Path planning method area: used to help individuals plan future paths and guide individuals to perform exercise behaviors;

[0119] (5) External area: includes environmental feedback functions, sensors, and action generator hardware facilities.

[0120] The above embodiments are only preferred embodiments of the present invention and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A cluster control method based on perception control and perception tolerance, characterized in that: The steps include: Step 1: The processor configures the cluster information of the group, wherein the cluster information includes the number of individuals, the number of termination iterations, the maximum error of the terminal state and the individual perception attributes, and the individual perception attributes include the limit acceleration, the perception range, the perception tolerance and the iteration step of the path planning method; Step 2: The processor initializes the number of iterations and configures individual state information, where the state information includes an initial state and an end state; Step 3: The processor substitutes the state information into the path planning method to calculate the path trajectory data, wherein the path trajectory data includes the speed information and position information of the individual in each iteration step; Step 4: The processor drives the individual to move according to the path trajectory data, and obtains the individual's current state target value and terminal state target value according to the path trajectory data index; Step 5: The sensor senses the individual's own state and the states of other individuals within the sensing range, and calculates the individual's current sensed information amount based on the individual's own state and the states of other individuals. Step 5 specifically includes: Step 51: The sensor senses the current individual's own state and the states of other individuals within the sensing range; Step 52: Calculate the Euclidean distance between the individual and other individuals based on the individual state and the states of other individuals; Step 53: Calculate the perception angles occupied by other individuals according to the following formula: Where f is the perception angle occupied by other individuals, BL is the individual’s own diameter, DisSefOth is the Euclidean distance; Add up all the perception angles of other individuals to get the overall perception angle, and divide the overall perception angle by 2π to get the amount of perceived information; Step 6: The sensor inputs its own state and the amount of perceived information into the current state control unit, and inputs the amount of perceived information, the end state target value and the perceived tolerance into the perceived attribute control unit. The current state control unit and the perceived attribute control unit are part of the processor. Step 7: The perception attribute control unit updates the individual perception attribute according to the amount of perception information, the terminal state target value and the perception tolerance. Step 7 specifically includes: Step 71: The perception attribute control unit compares the perceived information amount and the perception tolerance to obtain the difference between the two, and determines whether the difference between the two is greater than 0. If so, the difference between the two is retained. If not, it is determined whether the absolute value of the difference between the two is less than η times the perception tolerance, where η is an interval parameter for controlling the individual to reversely adjust the individual perception attribute, η∈(0,1). If so, the difference between the two is retained. If not, the difference between the two is set to 0. Step 72: The difference between the two is multiplied and then input into the following formula to update the individual perception attribute: PecpAtrbNew=PecpAtrbCurr+α·PecpDiff, Where PecpAtrbNew is the updated individual perception attribute, PecpAtrbCurr is the current individual perception attribute, and α is the proportional coefficient; Step 73: Predict whether the current individual and other individuals will collide based on the path trajectory data and other individual states, and if so, set the collision signal to 1; otherwise, set the collision signal to 0; Step 8: The current state control unit predicts the state error and the error correction amount. Step 8 specifically includes: Step 81: Subtract the current state target value from the own state to obtain the state error; Step 82: Determine whether the number of state errors in the state error container is equal to the error prediction sequence length. If so, obtain the error correction amount according to the error prediction model, input the error correction amount to the terminal state control unit, and initialize the state error container. If not, input the state error to the state error container, set the terminal state target value to 0, and input the error correction amount to the terminal state target value. Step 9: The perception attribute control unit inputs the updated individual perception attribute to the sensor and the end state control unit, the current state control unit inputs the state error and the error correction amount to the end state control unit, and the end state control unit updates the end state according to the state error, the error correction amount and the updated end state. Step 9 specifically includes: Step 91: Input the collision signal to the end state control unit to determine whether the collision signal is equal to 1, if so, change the end state, if not, do not change the end state; Step 92: Compare the modulus of the own state and the modulus of the target value of the terminal state to obtain the difference between the two, and determine whether the difference between the two is less than or equal to the maximum error of the terminal state. If so, change the terminal state, otherwise do not change the terminal state; Step 93: Input the error correction amount to the endpoint state control unit and update the endpoint state using the following formula: StateEndNew=StateEnd+β·ErrCorrVal, In the formula, StateEndNew is the updated end state, StateEndNew is the current end state, and β is the error amplification coefficient; Step 94: inputting the updated limit acceleration and iteration step into the terminal state control unit, and the terminal state unit substitutes the updated limit acceleration, the updated iteration step and the updated terminal state into the path trajectory method; Step 10: The processor substitutes the updated individual perception attributes and the end point state into the path planning method to update the path trajectory data, and the number of iterations increases by one; Step 11: The processor determines whether the number of iterations is equal to the termination number of iterations. If so, the iteration is terminated. If not, steps 4 to 11 are iterated.

Images