A posture self-balancing control method for a single-ball balancing robot
The self-balancing control method of single ball balanced robot designed by the Euler-Lagrangian equation and the root trajectory method solves the problem of dynamic instability of traditional wheeled mobile robots, and realizes the self-balancing of single ball balanced robots and Simplified structure and suitable for narrow environments.
Patent Information
- Application Number
- CN202210400777.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-17
- Publication Date
- 2024-11-15
- Estimated Expiration
- 2042-04-17
AI Technical Summary
Traditional wheeled mobile robots are dynamically unstable when the center of gravity is too high or accelerated or decelerated, and cannot be effectively applied in narrow spaces. Single-ball balanced robots can maintain dynamic balance at high center of gravity, but the control method is complex and difficult to achieve in industrialization.
The dynamic model of a single-sphere balanced robot is established by using the Euler-Lagrangian equation method, and the controller is designed through the root trajectory method to realize the self-balancing control of the robot's posture, simplify the structure and control method, and enhance the anti-interference.
The self-balancing attitude control of a single-ball balanced robot is realized, which reduces structural complexity and control difficulty, enhances the stability and flexibility of the robot, is suitable for narrow environments, and expands the application range.
Smart Images

Figure CN114859951B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a single-ball balancing robot control method, which specifically includes motion analysis, establishing a dynamics model and designing a control strategy, and belongs to the field of mechanical intelligent control. Background Art
[0002] The existing wheeled mobile robots have simple structures, control algorithms and driving methods, but when the center of gravity is too high or the acceleration is too large, the robot is very easy to flip over. In order for the robot to move more stably, the center of gravity of its structure must be low enough, and the mobile chassis needs to occupy a wide enough area. This limits the application of ordinary wheeled mobile robots, and they cannot complete tasks that require fast turning or narrow working spaces. The single-ball robot can maintain dynamic balance with a high center of gravity, and its body can be slimmer and its movements can be more flexible to adapt to narrow working environments. Therefore, it is of great significance to study the control of single-ball balancing robots.
[0003] Miao Wennan, Zhou Zheng, Xuan Jiyi, Chen Dongshui, Cao Naochang. Design and simulation of ball balancing robot controller based on fuzzy PID [J]. Sensors and Microsystems, 2021. The design of ball balancing robot controller based on fuzzy PID proposed in the paper has a complex control structure and requires a large number of parameters to be adjusted. The single ball balancing robot posture self-balancing control method proposed in the present invention refers to the two-dimensional inverted pendulum design, and the robot is projected onto each orthogonal plane for separate control. It has a clear structure, a simple control method, and is easy to implement in industry, so that the robot has stronger anti-interference ability. Summary of the invention
[0004] Technical issues to be solved
[0005] In order to overcome the defects of the traditional wheeled mobile robot, which is dynamically unstable and inflexible, the present invention proposes a flexible single-ball balancing robot (such as Figure 1 The control method of posture self-balancing (as shown in the figure) is based on establishing a reasonable dynamic model of the robot. First, the controllability and observability of the system state are determined, and then the root locus method is used to design a controller according to certain performance indicators to achieve the robot's posture self-balancing control.
[0006] Technical Solution
[0007] A single ball balancing robot posture self-balancing control method, characterized by the following steps:
[0008] Step 1: Use the Euler-Lagrange equation method to establish the robot's dynamic model
[0009] The robot's motion is projected onto the XOZ and YOZ planes to establish dynamic models. The Lagrangian dynamic differential equations of the XOZ plane projection model are:
[0010]
[0011]
[0012] In the formula, is the inertial force matrix generated by acceleration, is the centrifugal force matrix, G(q xz ) is the gravitational matrix, τ sys,xz is the torque acting on the system;
[0013] Step 2: Controllability analysis of the self-balancing posture of the single-ball balancing robot
[0014] The state space equation of the robot system in the XOZ plane is:
[0015]
[0016] y X =C X x X
[0017] In the formula, x X ,u X are the state variables and input torque of the system respectively, A X , B X , C X are the system matrix, input matrix and output matrix respectively; the state variable x X and the input torque q of the system xz They are
[0018] u X =τ sys,xz
[0019]
[0020]
[0021] in, is the linearized inertia force matrix;
[0022] Determine whether the system state is fully controllable. If Then the system state is completely controllable, otherwise it is not completely controllable;
[0023] Step 3: Design a single ball balancing robot posture self-balancing controller
[0024] The differential equations of the system are linearized within the maximum tilting angle range of the system, and the following approximation is taken
[0025] cos(θ)≈1
[0026] sin(θ)≈θ
[0027]
[0028]
[0029] The Laplace transform of the linearized system differential equations in the XOZ plane yields
[0030]
[0031]
[0032] In the formula, M′ and G′ are the linearized inertial force matrix and gravitational matrix respectively, and g is the gravitational acceleration;
[0033] After rearrangement, the open-loop transfer function is obtained
[0034]
[0035] For the divergent open-loop transfer function G(s), a series correction network is designed using the root locus method according to certain performance indicators to correct the system so that the system can converge quickly to achieve control of the robot's posture.
[0036] The robot's posture self-balancing control is achieved through the linear superposition and synthesis of posture self-balancing controllers in the XOZ and YOZ planes.
[0037] In step 2, n=4.
[0038] The maximum tilting angle θ of the system is set to 10°.
[0039] The specific design of performance indicators in step 3 is:
[0040] Maximum overshoot: σ p %≤10%
[0041] Adjustment time: t s =0.5s.
[0042] Beneficial Effects
[0043] The present invention proposes a control method for the posture self-balancing of a single-ball balancing robot, adopts the Euler-Lagrange equation method to establish a dynamic model, constructs a group of dynamic differential equations, and obtains the relationship between the input torque and the posture; according to the method of modern control theory, the controllability and observability of the robot system are pre-analyzed, and finally an advance correction series network is designed to realize the posture self-balancing control of the single-ball balancing robot.
[0044] In addition, in order to solve the problem that some parameters of the robot cannot be accurately measured, the present invention utilizes an auxiliary software modeling method to establish a three-dimensional model of the robot, and obtains parameters such as the moment of inertia on a computer, thus solving the problem that parameters cannot be measured, reducing workload, and expanding the scope of practical engineering applications. BRIEF DESCRIPTION OF THE DRAWINGS
[0045] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.
[0046] Figure 1 Single ball balancing robot structure diagram;
[0047] Figure 2 Robot model diagram in XOZ plane;
[0048] Figure 3 Flowchart for the design of the posture self-balancing control method for a single-ball balancing robot. DETAILED DESCRIPTION
[0049] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0050] In order to realize the posture self-balancing of a single-ball balancing robot, the present invention proposes a posture self-balancing controller design method based on an advance correction network to realize the posture self-balancing control of the robot, including the following parts:
[0051] First, the robot dynamics model is established. Through reasonable assumptions and abstract design, the differential equations required for the control algorithm are obtained to express the relationship between the input torque of the system and the robot's balance angle in the time domain. The design of the robot dynamics differential equation provides the basis for the judgment of controllability and the design of the controller in the subsequent steps.
[0052] The robot state space model can be further obtained through the differential equation of the robot dynamics model to determine the controllability of the robot system. The controllability of the system indicates that there is an appropriate control rate to stabilize the system within a finite time, that is, the appropriate input torque can keep the robot's balance angle stable. If the system is controllable, the controller can be further designed to ensure the stability of the robot system.
[0053] Finally, the transfer function between the input torque and the robot's balance posture is obtained according to the differential equation of the dynamic model to describe the system. The nonlinear system described by the transfer function is linearized and the controller is designed using the root locus method to achieve the self-balancing control of the robot's posture.
[0054] The specific steps are as follows:
[0055] Step 1: Use the Euler-Lagrange equation method to establish the robot's dynamic model;
[0056] The single-ball balancing robot is a multi-rigid body structure with complex internal interactions, and its posture is time-varying. It has 5 degrees of freedom in three-dimensional space. To directly solve the motion energy function of the robot in three-dimensional space, it is necessary to solve the high-dimensional dynamic equations, which is a complicated process. Therefore, the method proposed in the present invention projects the robot's motion to three planes, XOY, XOZ and YOZ, to establish dynamic models respectively. For the control of the robot's posture self-balancing, only the XOZ and YOZ planes need to be considered, and the projections under the two planes are exactly the same. Taking the design of the control method under the XOZ plane as an example, the YOZ plane will not be repeated.
[0057] The Euler-Lagrange equation method to establish a dynamic model mainly consists of the following four steps:
[0058] (1) Solve the kinetic energy and potential energy of the robot power system as a whole in the coordinate system (the robot's center of gravity is the coordinate origin, the vertical direction upward is the Z axis, and the robot's forward direction is the Y axis, establishing a right-handed coordinate system):
[0059] The robot model in the XOZ plane is as follows Figure 1 As shown in the figure, the robot body is simplified into a rigid link, and the three omnidirectional drive wheels and motors are abstractly simplified into a virtual drive wheel. w1 , r w2 are the radii of the spherical wheel and the virtual driving wheel respectively; θ w1,xz ,θ w2,xz ,θ y are the rotation angles of the spherical wheel, virtual driving wheel and robot body around the y-axis; I w1 , I w2 , I R are the moments of inertia of the spherical wheel, virtual driving wheel and robot body respectively; l is the distance between the center of gravity of the spherical wheel and the center of gravity of the robot; the masses of the robot body, spherical wheel and virtual wheel are m respectively R 、m w1 、m w2 Respectively.
[0060] In the XOZ plane, the coordinates of the spherical wheel, virtual drive wheel, and robot body can be expressed as:
[0061]
[0062]
[0063]
[0064] In the formula, They are the coordinates of the spherical wheel, virtual driving wheel and robot body in the XOZ plane.
[0065] The total kinetic energy of the robot system in the XOZ plane is
[0066] K xz =K R,xz +K w1,xz +K w2,xz
[0067] In the formula, K R,xz , K w1,xz , K w2,xz They represent the kinetic energy of the robot body, spherical wheel and virtual driving wheel in the XOZ plane respectively.
[0068] The total potential energy of the robot system in the XOZ plane is
[0069] P xz =P R,xz +P w1,xz +P w2,xz
[0070] Where P R,xz , P w1,xz , P w2,xz They represent the potential energy of the robot body, spherical wheel and virtual driving wheel in the XOZ plane respectively.
[0071] (2) Solve all external torques in the robot power system:
[0072] The robot is only subjected to the torque τ of the virtual driving wheel in the XOZ plane y,xz The torque in the coordinate system XOZ is
[0073]
[0074] In the formula, τ sys,xz , They are the torques on the overall system, spherical wheel and robot body respectively.
[0075] (3) Find the Lagrange equation Where K and P are the total kinetic energy and total potential energy of the robot system respectively:
[0076] The Lagrangian function of the system in the XOZ plane is:
[0077]
[0078] (4) Solving the Lagrangian kinematic equations
[0079]
[0080]
[0081] Substituting the obtained results into the Lagrangian dynamic differential equations describing the robot projection model on the XOZ plane, we can obtain
[0082]
[0083]
[0084] In the formula, is the inertial force matrix generated by acceleration, is the centrifugal force matrix, G(q xz ) is the gravitational matrix, τ sys,xz is the torque on the system. The Lagrangian dynamic differential equations in the YOZ plane are established in the same way.
[0085] Step 2: Controllability analysis of the self-balancing posture of the single-ball balancing robot;
[0086] The state space equation of the robot system in the XOZ plane is:
[0087]
[0088] y X =C X x X
[0089] In the formula, x X ,u X are the state variables and input torque of the system respectively, A X , B X , C X are the system matrix, input matrix and output matrix respectively. The state variable x X and the input torque q of the system xz They are
[0090] u X =τ sys,xz
[0091]
[0092]
[0093] in, is the linearized inertia force matrix.
[0094] Determine whether the system state is fully controllable. If Then the system state is completely controllable, otherwise it is not completely controllable.
[0095] The design of the robot state space model in the YOZ plane is similar.
[0096] The state space equation of the robot system in the YOZ plane is:
[0097]
[0098] y Y =C Y x Y
[0099] Where x Y ,u Y are the state variables of the system in the YOZ plane and the input torque of the system, A Y , B Y , C Y are the system matrix, input matrix and output matrix respectively. The state variable x Y and the input torque q of the system yz They are
[0100] u=τ sys,yz
[0101]
[0102]
[0103] in is the linearized inertia force matrix.
[0104] Determine whether the system state is fully controllable. If Then the system state is completely controllable, otherwise it is not completely controllable.
[0105] Step 3: Design of posture self-balancing controller for single ball balancing robot;
[0106] When the single ball balancing robot posture system is fully controllable, the robot state self-balancing controller is designed. The maximum tilting angle θ of the system is set to 10°, and the differential equations of the system are linearized within this angle range, taking the following approximation:
[0107] cos(θ)≈1
[0108] sin(θ)≈θ
[0109]
[0110]
[0111] The Laplace transform of the linearized system differential equations in the XOZ plane yields
[0112]
[0113]
[0114] Where M′ and G′ are the linearized inertial force matrix and gravitational matrix, respectively, and g is the gravitational acceleration.
[0115] After rearrangement, the open-loop transfer function is obtained
[0116]
[0117] For the divergent open-loop transfer function G(s), the root locus method is used to design a series correction network according to certain performance indicators to correct the system so that the system can converge quickly to achieve control of the robot's posture. The specific performance indicators of the design are:
[0118] Maximum overshoot: σ p %≤10%
[0119] Adjustment time: t s =0.5s (2% error band)
[0120] The design of the machine controller under the YOZ plane is similar.
[0121] The specific implementation of the present invention is described below in conjunction with the actual parameters of the robot:
[0122] Execute step 1: Use the Euler-Lagrange equation method to establish the robot's dynamic model;
[0123] (1) Solve the kinetic energy and potential energy of the robot's overall power system in the coordinate system
[0124] The total kinetic energy of the robot in the XOZ plane is K xz =K R,xz +K w1,xz +K w2,xz , where the robot body kinetic energy is
[0125]
[0126]
[0127] The kinetic energy of the spherical wheel is
[0128]
[0129]
[0130] The virtual wheel kinetic energy is
[0131]
[0132]
[0133] The total potential energy of the robot in the XOZ plane is P xz =P R,xz +P w1,xz +P w2,xz , where the potential energy of the robot body is:
[0134] P R,xz =m R gl cosθ y
[0135] Spherical wheel potential energy P w1,xz is 0
[0136] The virtual wheel potential energy is
[0137] P w2,xz =m w2 gl cosθ y
[0138] (2) Solve all external torques in the robot power system:
[0139] The robot is only subject to the torque τ of the driving wheel in the XOZ plane y,xz The torque in the coordinate system XOZ is
[0140]
[0141] (3) Find the Lagrange equation Where K and P are the total kinetic energy and total potential energy of the robot system respectively:
[0142] The Lagrangian function of the system in the XOZ plane is:
[0143]
[0144] Where K R,xz , K w1,xz , K w2,xz , P R,xz , P w1,xz , P w2,xz Already calculated in step (1).
[0145] (4) Solving the Lagrangian kinematic equations
[0146] According to the Euler-Lagrange equation method, the dynamic differential equations describing the robot's projection model on the XOZ plane are:
[0147]
[0148]
[0149]
[0150]
[0151]
[0152] Execution step 2: Analysis of the controllability and observability of the posture self-balancing of the single-ball balancing robot
[0153] According to the state space model of the system, substituting the actual parameters, the state space equation of the robot in the XOZ plane can be calculated as follows:
[0154]
[0155]
[0156] n is the number of state variable components.
[0157] The state variables of the system are The number of its components n=4.
[0158] The basis for judging the controllability of the system state is
[0159]
[0160] This shows that the system state is completely controllable and a controller can be designed to make the system stable.
[0161] The observability of the system is determined based on
[0162]
[0163] This shows that the robot is completely controllable and observable in the XOZ plane, indicating that a state feedback controller can be designed to keep the robot dynamically stable. The controllability and observability analysis of the robot in the YOZ plane is similar.
[0164] Execution step 3: Design of the posture self-balancing controller of the single ball balancing robot
[0165] In the XOZ plane, the inertial force matrix and gravitational matrix of the linearized system are
[0166]
[0167]
[0168] Its open-loop transfer function can be expressed as
[0169]
[0170] Substitute actual parameters
[0171]
[0172] Through Matlab and other computer-aided software simulation, it can be known that the system diverges. The root locus method is used to design an advance corrector to correct the system. After the series correction network is connected, the system must meet the following performance indicators
[0173] Maximum overshoot: σ p %≤10%
[0174] Adjustment time: t s =0.5s (2% error band)
[0175] Calculation based on corrected performance indicators
[0176]
[0177]
[0178] The damping coefficient ξ can be calculated to be 0.512. Taking ξ as 0.61, ω can be calculated n It is 13.1.
[0179] With ξ=cosβ, we can get β=46.37°
[0180] The dominant pole of the corrected system is
[0181]
[0182] The lead phase angle of the lead correction network is
[0183]
[0184]
[0185]
[0186]
[0187]
[0188]
[0189] Therefore, the correction network transfer function is
[0190]
[0191] Where K c is the compensation gain value to be determined, which is used to compensate for the assignment condition when the new root trajectory passes through the dominant pole. The corrected open-loop transfer function is
[0192]
[0193] According to the amplitude condition |G c (s)G(s)H(s)|=1,s=-7.99+10.38j,K can be calculated c =1272.8. The system's lead correction controller G can be obtained c (s) and the corrected system transfer function G c (s)G(s) are
[0194]
[0195]
[0196] Through Matlab and other computer-aided software simulation, it can be seen that the system after the series correction network has converged stably and achieved the control of the attitude angle. The design of the robot attitude self-balancing controller in the YOZ plane is the same.
[0197] The robot's posture self-balancing control is achieved through the linear superposition and synthesis of posture self-balancing controllers in the XOZ and YOZ planes.
[0198] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.
Claims
1. A single ball balancing robot posture self-balancing control method, characterized in that Here are the steps: Step 1: Use the Euler-Lagrange equation method to establish the robot's dynamic model The robot's motion is projected onto the XOZ and YOZ planes to establish dynamic models. The Lagrangian dynamic differential equations of the XOZ plane projection model are: In the formula, is the inertial force matrix generated by acceleration, is the centrifugal force matrix, G(q xz ) is the gravitational matrix, τ sys,xz is the torque acting on the system; Step 2: Controllability analysis of the self-balancing posture of the single-ball balancing robot The state space equation of the robot system in the XOZ plane is: y X =C X x X In the formula, x X ,u X are the state variables and input torque of the system respectively, A X , B X , C X are the system matrix, input matrix and output matrix respectively; the state variable x X and the input torque q of the system xz They are you X =t sys,xz in, is the linearized inertia force matrix; Determine whether the system state is fully controllable. If Then the system state is completely controllable, otherwise it is not completely controllable; Step 3: Design a single ball balancing robot posture self-balancing controller The differential equations of the system are linearized within the maximum tilting angle range of the system, and the following approximation is taken cos(θ)≈1 sin(θ)≈θ The Laplace transform of the linearized system differential equations in the XOZ plane yields In the formula, M′ and G′ are the linearized inertial force matrix and gravitational matrix respectively, and g is the gravitational acceleration; After rearrangement, the open-loop transfer function is obtained For the divergent open-loop transfer function G(s), a series correction network is designed using the root locus method according to certain performance indicators to correct the system so that the system can converge quickly to achieve control of the robot's posture. The robot's posture self-balancing control is achieved through the linear superposition and synthesis of posture self-balancing controllers in the XOZ and YOZ planes.
2. The method for controlling the posture self-balancing of a single-ball balancing robot according to claim 1, characterized in that In step 2, n=4.
3. The method for controlling the posture self-balancing of a single-ball balancing robot according to claim 1, characterized in that The maximum tilting angle θ of the system is set to 10°.
4. The method for controlling the posture self-balancing of a single-ball balancing robot according to claim 1, characterized in that The specific design of performance indicators in step 3 is: Maximum overshoot: σ p %≤10% Adjustment time: t s =0.5s.
