A hybrid coordinated controller design method based on artificial potential field and filtered backstepping
Through the hybrid coordination controller design based on artificial potential field and filtering inverse step method, the problem of limited stability and application range in multi-robot formation control is solved, and the stability and computing efficiency of the formation system are improved, avoiding the need for avoiding the System crash risk.
Patent Information
- Application Number
- CN202210730226.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-06-24
- Publication Date
- 2024-12-06
- Estimated Expiration
- 2042-06-24
AI Technical Summary
The existing multi-robot formation control methods are difficult to ensure stability, especially when the pilot fails or the environment changes, which can easily lead to system crashes, and the application scope of virtual structure methods is limited.
Using a hybrid coordination controller design method based on artificial potential field and filter inversion method, the relative displacement error and attitude direction angle error variable are defined by the directed topological structure, and the second-order filter is used to solve the input and virtual control signals. Constraint problems, and an auxiliary system was introduced, and the Lyapunov stability theory was used to prove that the error signal was asymptotically converged to zero.
The stability and quantitative derivation process of the formation system are realized, the risk of system collapse of the pilot method is avoided, the distributed control method is provided, the path planning is simplified, the calculation amount is reduced, and the potential field function is optimized to escape the minimum extreme point.
Smart Images

Figure CN115327896B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of coordinated control, and in particular to a hybrid coordinated controller design method based on artificial potential field and filtering backstepping method. Background Art
[0002] The problem of multi-robot coordinated control includes control problems such as multi-path planning, speed planning, formation maintenance, formation change, and collision avoidance. Among them, formation control has become one of the typical and representative research problems of multi-robot systems. Formation control is generally based on the research results of single robot path tracking control. When a single robot can move on the planned path, coordinated control should be used to keep the desired geometric formation between multiple robots. At the same time, it is necessary to adapt to changes in the surrounding environment to prevent collisions between robots and between robots and obstacles. At present, the commonly used formation control methods are mainly behavior-based methods, navigator-based methods, and virtual structure methods.
[0003] The behavior-based method is to decompose the formation control task into a series of basic behaviors and realize motion control through the integration of behaviors. This method usually uses empirical knowledge as the basis for controlling the formation, which is difficult to conduct quantitative analysis and does not have mathematically rigorous logical reasoning. Therefore, it is difficult to ensure the stability of the formation.
[0004] The navigator-based method is a control method in which the navigator plays a leading role. First, the controller is designed to make the navigator meet the desired motion state, and then the geometric position relationship between each follower and the navigator is controlled to achieve overall formation coordination control. In the case of navigator failure, if the design is not considered carefully and simply relies on it, the consequences will be very serious. This requires the use of strategies such as specifying alternate navigators and replacing navigators according to the situation in the design process to improve the performance of the system.
[0005] The virtual structure method is that the entire formation is treated as a single structure. In order to implement this method, it is first necessary to define the ideal dynamic characteristics of the virtual structure, then convert the motion of the virtual structure into the desired motion of each robot, and finally derive the trajectory tracking control method of the robot. The advantage of this method is that the behavior of the robot group (the behavior of the virtual structure) can be easily specified and the formation feedback can be performed. However, the requirement that the formation moves like a virtual structure limits its scope of application. Summary of the invention
[0006] In order to solve the technical problems existing in the background technology, the technical solution adopted by the present invention is:
[0007] A hybrid coordination controller design method based on artificial potential field and filtered backstepping method is proposed. For under-actuated robots performing six-degree-of-freedom motion underwater, communication information is exchanged between multiple robots according to a directed topological structure. The relative displacement error variables and two attitude direction angle error variables between robots are defined based on graph theory and artificial potential field theory. Then, a second-order formation dynamics model is established for the under-actuated robots to simplify the controller design process. The analytical structure of the controller is designed based on filtered backstepping technology, and a second-order filter is used to solve the typical constraint problems of input control signals and virtual control signals. At the same time, the filter can also avoid the problem of computational explosion caused by the derivation of virtual control signals in the backstepping process. Then, an auxiliary system is introduced to analyze the influence of system constraints on constraint control. Using Lyapunov stability theory, it is proved that all error signals in the closed-loop system converge to zero asymptotically.
[0008] Preferably, the method comprises the following steps:
[0009] Step 1: Select a communication topology with directed weighted edges based on graph theory knowledge; select an appropriate potential field function based on the expected formation structure between robots and artificial potential field theory, so as to derive the weighted values of the edges in the communication topology;
[0010] Step 2: Define the formation error variables, consider the relative distance maintenance and attitude control between robots, and give the distance error and two attitude error variables;
[0011] Step 3: Based on the formation error system established in step 2, the coordinated formation controller is designed using the filtered backstepping method.
[0012] Preferably, in step three, designing a coordinated formation controller using a filtering backstepping method comprises the following steps:
[0013] S1. In order to simplify the formation error dynamic system, the function f is designed i1 (q i ,r i ); Next, define the tracking error compensation signal e i9 =e i6 -s i1 and radial filtering velocity error e i10 =u i -u ir , design compensation signal s i1 The derivative structure of is used to compensate the error e i9 Next, define a new tracking error compensation signal e i11 =e i10 -s i2 , design compensation term s i2 The derivative structure and expected virtual control law of To stabilize ei9 Dynamic subsystem;
[0014] S2. In order to simplify the dynamic system of formation pitch attitude error, the function f is designed i2 (u i ,r i ); Next, define the tracking error compensation signal e i12 =e i7 -s i3 and the pitch attitude filtering error e i13 =q i -q ir , design compensation signal s i3 The derivative structure of is used to compensate the error e i12 ; Next, define a new tracking error compensation signal e i14 =e i13 -s i4 , design compensation term s i4 The derivative structure and expected virtual control law of To stabilize e i12 Dynamic subsystem;
[0015] S3. In order to simplify the dynamic system of the formation heading attitude error, the function f is designed i3 (u i ,q i ); Next, define the tracking error compensation signal e i15 =e i8 -s i5 and the heading attitude filtering error e i16 =r i -r ir , design compensation signal s i5 The derivative structure of is used to compensate the error e i15 ; Next, define a new tracking error compensation signal e i17 =e i16 -s i6 , design compensation term s i6 The derivative structure and expected virtual control law of To stabilize e i15 Dynamic subsystem;
[0016] S4. Design actual control force F 1i , control torque F 2i , F 3i , and the online estimate of the control compensation term And an adaptive learning algorithm for the estimated value is given, so that all error signals in the system converge to zero asymptotically.
[0017] The advantages of the present invention are: the analytical form of the controller can be obtained by adopting the coordinated formation control method provided by the present invention, which has a quantitative derivation process and rigorous mathematical logical reasoning, can ensure the stability of the formation system, and performs path planning based on the artificial potential field method, which is simple to implement, has a small amount of calculation, and a smooth path. The optimized potential field function can enable the system to escape from the minimum extreme point. It is also a distributed control method, which avoids the risk of system collapse in the navigation method. Artificial potential field path planning is combined with the filtered backstepping method to solve the formation coordination control problem of underwater under-actuated multi-robots. The invention adopts a second-order filter to solve the typical constraint problem of the input control signal and the virtual control signal. At the same time, the filter can also avoid the problem of computational explosion caused by the derivation of the virtual control signal in the backstepping process. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] The present invention and its features, configurations and advantages will become more apparent by reading the detailed description of non-limiting embodiments with reference to the following drawings. The same reference numerals indicate the same parts throughout the drawings. The drawings are not drawn to scale, but rather to illustrate the subject matter of the present invention.
[0019] Figure 1 is a design flow chart of the controller of the present invention;
[0020] Figure 2 is a directed weighted graph in the multi-robot system of the present invention;
[0021] Figure 3 This is a diagram of the multi-robot formation control system;
[0022] Figure 4 It is a system diagram of the second-order filter of the present invention. DETAILED DESCRIPTION
[0023] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.
[0024] like Figure 1-Figure 4As shown, the hybrid coordination controller design method based on artificial potential field and filtered backstepping provided by the present invention, for an underactuated robot performing six-degree-of-freedom motion underwater, communication information is exchanged between multiple robots according to a directed topological structure, and the relative displacement error variable and two attitude direction angle error variables between the robots are defined based on graph theory and artificial potential field theory. Then, a second-order formation dynamics model is established for the underactuated robot to simplify the controller design process, and the analytical structure of the controller is designed based on filtered backstepping technology. A second-order filter is used to solve the typical constraint problems of the input control signal and the virtual control signal. At the same time, the filter can also avoid the problem of computational explosion caused by the derivation of the virtual control signal during the backstepping process. Then, an auxiliary system is introduced to analyze the influence of system constraints on the constraint control, and Lyapunov stability theory is used to prove that all error signals in the closed-loop system converge to zero asymptotically. Specifically, the following steps are included:
[0025] Step 1: Select a communication topology with directed weighted edges based on graph theory knowledge; select an appropriate potential field function based on the expected formation structure between robots and artificial potential field theory, thereby deriving the weighted values of the edges in the communication topology;
[0026] The kinematic system of the i-th robot can be expressed as follows:
[0027]
[0028]
[0029]
[0030]
[0031]
[0032] {U} represents the inertial coordinate system that accommodates all robots at the same time. i} is the carrier coordinate system of the i-th robot itself. i Represents the center of mass of the robot. In order to simplify the mathematical model of robot motion, we usually use the carrier coordinate system {B i}The origin Q Bi Selected as the center of mass Q of the robot i The position of. i ,y i ,z i ] T is the center of mass Q in the inertial coordinate system {U} i The position vector of i ,ψ i ] T represents the center of mass Q in the inertial coordinate system {U} iThe direction angle vector. i ,v i ,w i ] T The robot is in the carrier coordinate system {B i} in the linear velocity vector. i ,r i ] T The robot is in the carrier coordinate system {B i} is the rotation angular velocity vector in .
[0033] For an underactuated robot, its dynamic system can be expressed as follows:
[0034]
[0035]
[0036]
[0037]
[0038]
[0039] in, d 11i =-X ui ,d 22i =-Y vi ,d 33i =-Z wi ,d 55i =-M qi ,d 66i =-N ri . m i and m (·) Represent the mass and additional mass of the i-th robot respectively. 1i ,F 2i , and F 3i Respectively represent the force and torque provided by the propeller. (·) ,Y (·) ,Z (·) ,M (·) , and N (·) Both represent the hydrodynamic coefficient. (·) is the inertia moment of the robot. ρ represents the density of seawater. g is the acceleration due to gravity. Represents the volume of seawater. Represents the longitudinal metacentric height of the robot. τ i =[τ 1i ,τ 2i ,τ 3i ,τ 5i ,τ6i ] T The representative system does not model dynamics and external environmental disturbances.
[0040] The position error and angular error between the i-th robot and the j-th robot are converted from the inertial coordinate system to the carrier coordinate system. This conversion system (3) is:
[0041] e i1 =(x i -x j )cosψ i cosθ i +(y i -y j )sinψ i cosθ i -(z i -z j )sinθ i
[0042] e i2 =-(x i -x j )sinψ i +(y i -y j )cosψ i
[0043] e i3 =(x i -x j )cosψ i sinθ i +(y i -y j )sinψ i sinθ i +(z i -z j )cosθ i
[0044] e i4 =θ i -θ j
[0045] e i5 =ψ i -ψ j (3)
[0046] By solving the instantaneous rate of change of the error variable in system (3) respectively, we can obtain the following error dynamic system in the carrier coordinate system:
[0047]
[0048]
[0049]
[0050]
[0051]
[0052] Represents a directed graph (such as Figure 1 As shown in Figure 2), this directed graph is the communication link network between multiple robots. Among them, υ represents the set of all communication nodes, that is, for a communication link network composed of n robots, υ = {1, 2, ..., n}. Represents the set of edges connecting different nodes. In the communication structure, if there is an edge between the i-th node and the j-th node, it means that the i-th node and the j-th node are connected. If there are r+1 connected nodes from node i to node j, the length of this connected path is recorded as r. If there is at least one connected path between any two nodes, the network is called is connected. In the communication topology network In this example, let A be the adjacency matrix, A = (a ij ) n×n is an n-order square matrix, where if the i-th node and the j-th node are connected, then a ij =1; if the i-th node and the j-th node are not connected, then a ij = 0. And it is stipulated that a ii =1,i=1,2,…,n. In the communication topology network In this example, D is the degree matrix, D = (d ij ) n×n is a diagonal matrix, where d ii =|N i |,|N i | is the set N i The potential of the set N i is the set of all nodes that have communication connections with the i-th node. If the communication topology network is connected, then the matrix D -1 (DA) is semi-positive definite, that is, for any non-zero n-dimensional real column vector x, there is x T D -1 (DA)x≥0. Communication topology network The Laplace operator of is defined as L = DA. From the definitions of matrix D and matrix A, it can be seen that matrix L is symmetric and satisfies L1 = 0, where 1 = [1] n×1 ,0=[0] n×1If the communication topology network is connected, then the matrix L has an eigenvalue of zero, its corresponding eigenvector is 1, and the rest of the eigenvalues of the matrix L are positive.
[0053] Will edge The weight of We can get the Laplacian matrix of a weighted graph:
[0054]
[0055]
[0056] Design the following potential field function for the edges in the multi-robot system communication topology:
[0057]
[0058] Among them, k ij ,σ,c ij ,λ ij are all positive constants, ||l ij ||It is a robotυ i and j The expected distance between ij is the elastic coefficient, which is used to determine the strength of the interaction between robots.
[0059] In formula (11), ||l ij || is a variable, so it is a check mark function. When defining c ij =2k ij +σ, then
[0060]
[0061] Established, that is,
[0062] minV ij =0 (12)
[0063] This ensures that the global minimum value of the potential field function is equal to zero.
[0064]
[0065] That is ||l ij ||=k ij +σ, the potential field function reaches the global minimum. ij When the distance between two robots is too far or too close, they will be attracted or repelled accordingly. Eventually, the distance reaches k. ij+σ, the potential field strength is zero, the attraction and repulsion disappear, and the target distance is reached. Since σ is a constant, the desired formation and control parameter k ij There is a relationship between them, that is, by designing the control parameter k corresponding to each edge in the communication topology ij The desired formation can be achieved by adjusting the value of
[0066] After defining the potential field function between the two robots, the potential energy of the entire multi-robot system can be obtained:
[0067]
[0068] By differentiating the potential energy of the system, we can obtain:
[0069]
[0070] According to formula (14), the following weight function is designed for each edge:
[0071]
[0072] Step 2: Define the formation error variables, consider the relative distance maintenance and attitude control between robots, and give the distance error and two attitude error variables;
[0073] For the i-th robot i , it has |N i | neighbors, so the following variables can be defined:
[0074]
[0075]
[0076]
[0077] Among them, L ij is the weighted Laplace matrix of the system communication topology graph The element value in .
[0078] The definition of the total error and error direction angle of the i-th robot is given as:
[0079]
[0080]
[0081]
[0082] in,
[0083] The derivative calculation results of the system (19)-(21) are given as follows:
[0084]
[0085]
[0086]
[0087] Control objective: Design a multi-robot formation structure. For the i-th robot, design the control force and torque signal [F 1i ,F 2i ,F 3i ] T The following conditions need to be met: the desired formation structure is maintained between the i-th robot and the j-th robot, that is, the formation tracking error variable converges to zero asymptotically. That is,
[0088]
[0089]
[0090]
[0091] Among them, ε ei6 ,ε ei7 , and ε ei8 are all arbitrarily small positive definite constants.
[0092] Step 3: Based on the formation error system established in step 2, the coordinated formation controller is designed using the filtered backstepping method.
[0093] The design structure of the second-order filter is given:
[0094]
[0095] Among them, ξ and ω n are the damping ratio coefficient and inertia frequency of the second-order filter respectively. is the expected virtual control signal. ir and are the virtual control signals of the filter (25) and their instantaneous change rates. To the output signal ui r The transfer function is:
[0096]
[0097] Given the structural form of the system (26), if a sufficiently large inertia frequency ω is chosen n , then the bandwidth of the filter transfer function G(s) will be much larger than the input signal Therefore, the error signal will gradually decrease, so the effectiveness of the filter is natural. Figure 3 Visible signal The acquisition is through the integration of forward information rather than differential operation. This method can greatly reduce the impact of high-frequency measurement noise on system feedback data, and also solve the problem of signal The problem of obtaining the instantaneous rate of change when it is not differentiable. However, the larger inertia frequency ω n It will increase the impact of high-frequency noise on the system, so we need to choose a suitable inertia frequency value to balance the error reduction effect and the impact of high-frequency noise.
[0098] In the same way, we can give the following filter models for the dive angular velocity and the yaw angular velocity:
[0099]
[0100]
[0101] in, and are all expected virtual control signals. ir ,r ir , and is the actual virtual control signal and its instantaneous rate of change calculated by the above filter. Based on the filter backstepping method, the nonlinear control law is derived in four steps:
[0102] S1. In order to simplify the error dynamic system (22) and simplify the control law derivation process, the following function is defined:
[0103]
[0104] Then, we get the system:
[0105]
[0106] Design control input signal u i , and apply it to the scalar system (30), so that the error system is stabilized at the origin, that is, e i6 = 0. The tracking error compensation signal is defined as e i9 =e i6 -s i1 , then its instantaneous rate of change is:
[0107]
[0108] where s i1 Used to offset the error signal e i9, and satisfy s i1 (0) = 0. Next, define a radial filtering velocity error:
[0109] e i10 =u i -u ir (32)
[0110] Among them, u i r is the virtual control input. The compensation signal s is designed as follows i1 The instantaneous rate of change:
[0111]
[0112] Applying the radial filter velocity error (32) and the compensation signal instantaneous rate of change (33) to the error system (31) simultaneously, the following system structure can be obtained:
[0113]
[0114] Let's define a new tracking error compensation signal:
[0115] e i11 =e i10 -s i2 (35)
[0116] where s i2 Used to offset the error signal e i11 , and satisfy s i2 (0) = 0. Design the desired virtual control law in the following form:
[0117]
[0118] By applying the tracking error compensation signal (35) and the desired virtual control law (36) to the error system (34) at the same time, the following system structure can be obtained:
[0119]
[0120] For system (37), choose a cost function, namely the Lyapunov function Its increment ratio limit is:
[0121]
[0122] By taking the derivative of the error signal of system (35), we can obtain:
[0123]
[0124] Design the following compensation structure:
[0125]
[0126] S2. In order to simplify the error dynamic system (23) and simplify the control law derivation process, the following function is defined:
[0127]
[0128] Then, we get the system:
[0129]
[0130] Design control input signal q i , and apply it to the scalar system (42) so that the error system stabilizes at the origin, that is, e i7 = 0. The tracking error compensation signal is defined as:
[0131] e i12 =e i7 -s i3 (43)
[0132] Then its instantaneous rate of change is:
[0133]
[0134] Among them, the introduction signal s i3 is to compensate for the error variable e i12 , and satisfy s i3 (0) = 0. Next, a filter tracking error variable is defined for the diving angular velocity:
[0135] e i13 =q i -q ir (45)
[0136] Among them, q ir is the virtual control input signal. Design compensation term s i3 The instantaneous rate of change is as follows:
[0137]
[0138] Among them, k i3 > 0. Next, the filter error variable (45) and the compensation system (46) are simultaneously applied to the error system (44), and the following system structure can be obtained:
[0139]
[0140] Based on system (47), a new tracking error compensation variable is defined:
[0141] e i14=e i13 -s i4 (48)
[0142] Among them, s i4 To offset the error e i14 , and satisfy s i4 (0) = 0. Design the desired virtual control law in the following form:
[0143]
[0144] Applying the tracking error compensation signal (48) and the desired virtual control law (49) to the error system (47) simultaneously, the following system structure can be obtained:
[0145]
[0146] For system (50), a cost function is selected, namely the Lyapunov function Its increment ratio limit is:
[0147]
[0148] Taking the derivative of the error signal of system (48), we can obtain:
[0149]
[0150] Design the following compensation structure:
[0151]
[0152] S3. In order to simplify the error dynamic system (24) and simplify the control law derivation process, the following function is defined:
[0153]
[0154] Then, we get the system:
[0155]
[0156] Design control input signal r i , and apply it to the scalar system (55), so that the error system is stabilized at the origin, that is, e i8 = 0. The tracking error compensation signal is defined as:
[0157] e i15 =e i8 -s i5 (56)
[0158] Then its instantaneous rate of change is:
[0159]
[0160] Among them, the introduction signal s i5 is to compensate for the error variable e i15 , and satisfy s i5 (0) = 0. Next, a filter tracking error variable is defined for the yaw angular velocity:
[0161] e i16 =r i -r ir (58)
[0162] Among them, r ir is the virtual control input signal. Design compensation term s i5 The instantaneous rate of change is as follows:
[0163]
[0164] Among them, k i5 > 0. Next, the filter error variable (58) and the compensation system (59) are simultaneously applied to the error system (57), and the following system structure can be obtained:
[0165]
[0166] Based on system (60), a new tracking error compensation variable is defined:
[0167] e i17 =e i16 -s i6 (61)
[0168] Among them, s i6 To offset the error e i17 , and satisfy s i6 (0) = 0. Design the desired virtual control law in the following form:
[0169]
[0170] By applying the tracking error compensation signal (61) and the desired virtual control law (62) to the error system (60) at the same time, the following system structure can be obtained:
[0171]
[0172] For system (63), choose a cost function, namely the Lyapunov function Its increment ratio limit is:
[0173]
[0174] By taking the derivative of the error signal of system (61), we can obtain:
[0175]
[0176] Design the following compensation structure:
[0177]
[0178] S4. For the actual control force and torque, the following structural form is designed:
[0179]
[0180]
[0181]
[0182] in, and They are the control compensation terms τ 1i ,τ 5i , and τ 6i The online estimated value of . Then, we define the estimation error as and The estimated value of the compensation term approaches the optimal value through adaptive learning adjustment, and its adaptive learning laws are:
[0183]
[0184]
[0185]
[0186] Among them, the adaptive learning coefficient satisfies k i7 >0,k i8 >0, and k i9 > 0. The adaptive learning law (70)-(72) is regarded as an exponential observer. Based on the subsequent system stability analysis, it can be seen that the observer (70)-(72) can converge the system interference error quickly at an exponential rate. The parameter m in the observer (70)-(72) 11i ,m 55i , and m 66i It will be affected by external disturbances of the system, so its value contains certain noise information, but these noisy parameter signals can be offset in stability analysis through control law design, thereby eliminating the influence of measurement noise on formation accuracy. Taking the adaptive learning law (70) as an example, is τ 1i So there will be an estimation error in the control system. At this time, the learning law (70) Can be used to offset the estimation error term The specific offset proof process is shown in equation (74). Although the parameters in the adaptive learning rate (70)-(72) are determined by the model, their control effect is not affected by model uncertainty. Therefore, the filter-based control algorithm makes the controlled system have closed-loop robustness.
[0187] For the coordinated formation control system of n robots, an overall cost function, namely the Lyapunov function, is selected as follows:
[0188]
[0189] The instantaneous rate of change of the function V can be calculated:
[0190]
[0191]
[0192] The designed system control force and torque (67)-(69), where the filter compensation term is (70)-(72), are applied to the system (74) to obtain the following results:
[0193]
[0194] remember is the error state vector. From equation (75), we can see that the instantaneous rate of change of the cost function is negative semidefinite. Therefore, the error state matrix E of the system is stable. Moreover, for all t ≥ 0, the error matrix E is bounded. Under the support of the Lasalle invariance principle, it can be proved that the error state matrix E converges asymptotically to zero. In the control law design process based on the filter backstepping method, if the inertia frequency ω of the second-order filter n If the design is large enough, the error between the virtual control law and the expected control law in each dimension and will converge to zero. The tracking error compensation signal e i9 ,e i12 , and e i15 From the definition of , we know that when time t→∞, the formation tracking error e i6 →0,e i7 →0,e i8 →0. That is, the formation error signal in the closed-loop system converges to zero exponentially.
[0195] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A hybrid coordinated controller design method based on artificial potential field and filter backstepping method, characterized by: For underactuated robots that move in six degrees of freedom underwater, communication information is exchanged between multiple robots according to a directed topological structure. The relative displacement error variables and two attitude direction angle error variables between robots are defined based on graph theory and artificial potential field theory. Then, a second-order formation dynamics model is established for the underactuated robots to simplify the controller design process. The analytical structure of the controller is designed based on the filtered backstepping technology. A second-order filter is used to solve the typical constraint problems of the input control signal and the virtual control signal. At the same time, the filter can also avoid the problem of computational explosion caused by the derivation of the virtual control signal during the backstepping process. Then, an auxiliary system is introduced to analyze the influence of system constraints on constraint control. Using Lyapunov stability theory, it is proved that all error signals in the closed-loop system converge to zero asymptotically. The method comprises the following steps: Step 1: Select a communication topology with directed weighted edges based on graph theory knowledge; select an appropriate potential field function based on the expected formation structure between robots and artificial potential field theory, so as to derive the weighted values of the edges in the communication topology; Step 2: Define the formation error variables, consider the relative distance maintenance and attitude control between robots, and give the distance error and two attitude error variables; Step 3: Based on the formation error system established in step 2, a coordinated formation controller is designed using the filter backstepping method; In step 3, the design of the coordinated formation controller using the filtered backstepping method includes the following steps: S1. In order to simplify the formation error dynamic system, the function is designed ; Next, define the tracking error compensation signal and radial filtered velocity error Design compensation signal The derivative structure of Next, define a new tracking error compensation signal , design compensation item The derivative structure and expected virtual control law of To stabilize Dynamic subsystem; S2. In order to simplify the dynamic system of formation pitch attitude error, the function is designed ; Next, define the tracking error compensation signal and pitch attitude filtering error , design compensation signal The derivative structure of ; Next, define a new tracking error compensation signal , design compensation item The derivative structure and expected virtual control law of To stabilize Dynamic subsystem; S3. In order to simplify the dynamic system of the formation heading attitude error, the function is designed ; Next, define the tracking error compensation signal and heading attitude filter error , design compensation signal The derivative structure of ; Next, define a new tracking error compensation signal , design compensation item The derivative structure and expected virtual control law of To stabilize Dynamic subsystem; S4. Design actual control , control torque , , and the online estimate of the control compensation term , , , and an adaptive learning algorithm for the estimated value is given, so that all error signals in the system converge to zero asymptotically.
Citation Information
Patent Citations
Designing method of water surface target rescuing tracking and observing controller based on filtering backstepping method
CN108052009A
Self-adaptive backstepping sliding mode multi-unmanned ship formation control method based on sampling communication
CN114047744A
