Anti-interference control system for trajectory tracking of wheeled robot

Through nonlinear dynamic modeling and perturbation estimation combined with sliding mode control and model prediction control, the problem of insufficient trajectory tracking accuracy and anti-interference ability of the wheeled robot trajectory tracking control system in complex dynamic environments is solved, and an efficient and robust trajectory tracking effect is achieved.

CN120523033APending Publication Date: 2025-08-22INNER MONGOLIA DATANG INT TUOKETUO POWER GENERATION
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202510640614.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2025-02-24
Filing Date
2025-05-19
Publication Date
2025-08-22

AI Technical Summary

Technical Problem

The existing wheeled robot trajectory tracking control system is difficult to achieve high-precision and high-rootability control in complex dynamic environments, lacks real-time estimation and compensation for external disturbances and modeling errors, has high computational complexity, is difficult to adapt to dynamic scenarios, has poor real-time trajectory planning, and it is difficult to balance anti-interference ability and trajectory tracking accuracy.

Method used

The nonlinear dynamic modeling module, disturbance estimation module, sliding mode control module, model prediction control module and fusion control module are adopted, combined with the extended state observer and dynamic window method, real-time estimation and compensation of external disturbances and modeling errors are achieved. Through the coordinated optimization of sliding mode control and model prediction control, the trajectory is dynamically adjusted to improve the trajectory tracking accuracy and system robustness.

Benefits of technology

It improves the trajectory tracking accuracy and anti-interference performance of wheeled robots in complex dynamic environments, reduces the computational complexity, realizes real-time trajectory adjustment and efficient operation in dynamic environments, and solves the shortcomings of the trajectory tracking system in the existing technology in complex environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120523033A_ABST
    Figure CN120523033A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of mobile robot control, and discloses a wheeled robot trajectory tracking anti-interference control system, which comprises a nonlinear dynamics modeling module used for establishing a nonlinear dynamics model of a wheeled robot so as to describe the motion characteristics of the robot under the influence of ground friction force, sliding force and modeling errors; the disturbance estimation module is used for estimating external disturbance and system modeling errors in real time through an extended state observer and providing disturbance compensation data for the controller; and the sliding mode control module is used for realizing rapid convergence of track deviation by constructing a sliding mode surface and enhancing the anti-interference capability of the system by dynamically adjusting the sliding mode gain. According to the invention, through nonlinear dynamic modeling, disturbance estimation, sliding mode control and model prediction control fusion and real-time trajectory optimization, high-precision trajectory tracking is realized, and the anti-interference capability and the dynamic environment adaptability are enhanced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of mobile robot control, in particular to a wheeled robot trajectory tracking anti-interference control system. Background Art

[0002] Wheeled robots, a common type of mobile robot, are widely used in a variety of fields, including industry, logistics, agriculture, and security inspection. In these applications, wheeled robots often need to perform precise trajectory tracking tasks to ensure that the robot completes its target mission along the set path. The trajectory tracking control system is a core component of wheeled robots. Its primary function is to calculate control inputs in real time based on the deviation between the target trajectory and the robot's current position, ensuring precise tracking of the target trajectory.

[0003] Existing wheeled robot trajectory tracking control systems primarily employ designs based on kinematic models or linear control algorithms. These control systems typically combine the global trajectory generated by a path planning module with methods such as PID control and incremental control to track the target trajectory. These control systems offer excellent trajectory tracking performance in static, regular environments and can meet the requirements of scenarios with relatively low dynamic complexity.

[0004] However, existing wheeled robot trajectory tracking control systems still have significant shortcomings in complex dynamic environments, making it difficult to meet the requirements of high-precision and robust control. First, existing technologies are often designed based on ideal kinematic models and fail to fully consider nonlinear factors such as ground friction and slip forces, resulting in insufficient trajectory tracking accuracy in complex environments. Second, most existing systems lack real-time estimation and compensation mechanisms for external disturbances and modeling errors, relying solely on fixed parameters or empirical compensation, making them difficult to adapt to dynamic scenarios with large variations in disturbance intensity. Furthermore, traditional control methods such as PID control, while simple in structure, lack the ability to respond quickly to large disturbances. While model predictive control offers high-precision optimization capabilities, it has weak adaptability to strong disturbances. Existing technologies cannot achieve a balance between anti-interference capability and trajectory tracking accuracy. Furthermore, existing trajectory tracking systems are mostly based on static path planning methods and lack the ability to adjust the trajectory in real time or dynamically replan, resulting in poor adaptability in dynamic environments. Finally, existing control systems have high computational complexity, making it difficult to achieve rapid response, especially in scenarios with long prediction horizons or high real-time requirements. Summary of the Invention

[0005] In response to the shortcomings of the existing technology, the present invention provides a wheeled robot trajectory tracking anti-interference control system, which solves the problems of the existing wheeled robot trajectory tracking control system in imperfect dynamic modeling, insufficient real-time disturbance estimation and compensation capabilities, a single control strategy that makes it difficult to balance anti-interference performance and trajectory tracking accuracy, insufficient trajectory optimization capabilities and low computational efficiency.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A wheeled robot trajectory tracking anti-interference control system, comprising:

[0007] Nonlinear dynamics modeling module, used to establish a nonlinear dynamics model of a wheeled robot to describe the robot's motion characteristics under the influence of ground friction, slip force, and modeling errors;

[0008] The disturbance estimation module is used to estimate external disturbances and system modeling errors in real time through an extended state observer, and to generate disturbance compensation data based on the dynamic model established by the nonlinear dynamic modeling module;

[0009] The sliding mode control module is used to achieve rapid convergence of trajectory deviation by constructing a sliding mode surface based on the disturbance compensation data provided by the disturbance estimation module, and to enhance the system's anti-interference ability by dynamically adjusting the sliding mode gain;

[0010] The model predictive control module is used to construct an optimization objective function based on the disturbance prediction data provided by the disturbance estimation module and optimize the control input to improve trajectory tracking accuracy and system robustness;

[0011] The fusion control module is used to combine the outputs of the sliding mode control module and the model predictive control module to achieve coordination between anti-interference ability and trajectory tracking accuracy by dynamically adjusting weights;

[0012] The trajectory optimization module is used to adjust the target trajectory based on the dynamic window method and provide the adjusted target trajectory to the fusion control module to ensure the motion stability of the robot in a dynamic environment.

[0013] Preferably, the nonlinear dynamics model established by the nonlinear dynamics modeling module includes:

[0014] Kinematic model, used to describe the relationship between the position, heading angle and velocity of the wheeled robot in the global coordinate system;

[0015] The dynamic model is used to describe the dynamic relationship between a wheeled robot under the action of ground friction, slip force and driving force, where the slip force is represented by the functional relationship between the slip coefficient and the speed.

[0016] Preferably, the disturbance estimation module is constructed based on an extended state observer, which is specifically:

[0017] State observation equation, used to observe the changing trend of robot state variables;

[0018] Disturbance estimation equations are used to estimate unknown disturbances and modeling errors as extended states in real time.

[0019] Preferably, the sliding mode control module realizes rapid convergence of trajectory deviation by designing a sliding mode surface, and the sliding mode surface is dynamically adjusted according to the deviation between the current position of the robot and the target trajectory to ensure rapid response of the system.

[0020] Preferably, the sliding mode control module realizes rapid convergence of trajectory deviation by designing a sliding mode surface, and the sliding mode surface is dynamically adjusted according to the deviation between the current position of the robot and the target trajectory to ensure rapid response of the system.

[0021] Preferably, the model predictive control module includes:

[0022] An optimization objective function construction unit is used to construct an optimization objective function according to trajectory tracking deviation, control input and disturbance prediction value;

[0023] The disturbance prediction unit is used to predict future disturbance changes using a recurrent neural network based on historical states, control inputs, and disturbance data;

[0024] The optimization solver is used to optimize the control input in real time through sparse matrix calculation method and event trigger mechanism.

[0025] Preferably, the fusion control module dynamically adjusts the output weights of the sliding mode control module and the model predictive control module according to external disturbances and trajectory deviations, wherein:

[0026] The weights are dynamically assigned based on the ratio of the disturbance estimate to the trajectory deviation, which is used to enhance the output weight of the sliding mode control module when the disturbance is large, and to enhance the output weight of the model predictive control module when the trajectory deviation is small.

[0027] Preferably, the trajectory optimization module adjusts the target trajectory based on a dynamic window method, wherein:

[0028] The dynamic window method combines the robot velocity constraint and the dynamic obstacle detection results to generate a new local target trajectory in real time;

[0029] The newly generated target trajectory is optimized by minimizing the objective function of local deviation and trajectory tracking deviation.

[0030] Preferably, the control system adopts a hierarchical control architecture, including:

[0031] The upper control module is used to generate global target trajectories and provide dynamic environment information;

[0032] The lower-level control module is used to realize trajectory tracking and anti-interference control based on the target trajectory and dynamic information provided by the upper layer, combined with the sliding mode control module and the model predictive control module.

[0033] Preferably, the control system is implemented through an embedded hardware platform and is modularly designed based on the robot operating system ROS. The perception module is used to obtain environmental information, the decision module is used for dynamic trajectory adjustment and controller output, and the execution module is used to implement motion control commands.

[0034] The present invention provides a wheeled robot trajectory tracking anti-interference control system. It has the following beneficial effects:

[0035] 1. This invention combines a nonlinear dynamics modeling module with an extended state observer (ESO) to accurately describe the true motion characteristics of a wheeled robot in a complex dynamic environment. It also estimates external disturbances and modeling errors in the system in real time, and utilizes disturbance compensation to improve the system's perception and response to unknown disturbances. This achieves the technical effect of enhancing the system's robustness and anti-interference performance. Compared to existing control schemes that rely on fixed parameter models or ignore disturbance compensation, it solves the problems of slow response to external disturbances and poor dynamic adaptability.

[0036] 2. This invention combines a sliding mode control module with a model predictive control (MPC) module to design a high-order sliding mode control law to reduce chattering. It also introduces a disturbance prediction mechanism based on an LSTM neural network to optimize control inputs, achieving comprehensive suppression and precise tracking of strong disturbances and trajectory deviations. The fusion control module dynamically adjusts the weight ratio of sliding mode control and model predictive control to enhance robustness under strong disturbance conditions and improve trajectory tracking accuracy under small disturbance conditions. Compared to existing solutions with a single control strategy, this solves the problem of finding a balance between anti-interference capability and trajectory tracking accuracy.

[0037] 3. This invention utilizes a trajectory optimization module and a dynamic window method to generate real-time trajectory candidate sets. Combined with cost function optimization and dynamic trajectory replanning, it ensures the feasibility and safety of the robot's trajectory in complex dynamic environments. A hierarchical optimization strategy significantly reduces the computational complexity of trajectory replanning. This achieves the technical effect of updating trajectories in real time and maintaining efficient system operation in dynamic environments. Compared to existing solutions based on global path planning, this approach addresses the issues of poor real-time trajectory planning and insufficient environmental adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a system structure diagram of the present invention;

[0039] Figure 2 This is a system structure diagram of the model predictive control module of the present invention;

[0040] Figure 3 This is a system structure diagram of the nonlinear dynamics modeling module of the present invention. DETAILED DESCRIPTION

[0041] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the specification of the present invention. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0042] Please see the attached Figure 1 -Attached Figure 3 , an embodiment of the present invention provides a wheeled robot trajectory tracking anti-interference control system, comprising:

[0043] Nonlinear dynamics modeling module, used to establish a nonlinear dynamics model of a wheeled robot to describe the robot's motion characteristics under the influence of ground friction, slip force, and modeling errors;

[0044] The design of the nonlinear dynamics modeling module specifically includes the following:

[0045] First, a kinematic model of the wheeled robot is established to describe the displacement and heading angle changes of the robot in the global coordinate system. Specifically, assuming that the robot has two differentially driven wheels, its kinematic equation is:

[0046]

[0047] in, and Respectively represent the position of the robot in the global coordinate system; represents the heading angle of the robot in the global coordinate system; v represents the linear velocity of the robot; δ represents the steering angle of the front wheel; L represents the wheelbase between the front and rear wheels of the robot. This kinematic model is mainly used to describe the ideal motion trajectory of a wheeled robot under interference-free conditions. This model can be used to generate the reference trajectory of the trajectory tracking module.

[0048] Secondly, a dynamic model of the wheeled robot is constructed to describe the dynamic characteristics of the force acting on the robot. The establishment of the dynamic model is based on Newton's laws of motion, which specifically includes the combined effects of driving force, friction force, and sliding force. The dynamic equation is:

[0049]

[0050] in, is the acceleration of the robot; m is the total mass of the robot; F driveThe driving force of the robot in the forward direction; F friction =c f v is the ground friction, where c f Indicates the friction coefficient; F slip =k slip v+∈ is the slip force, where k slip is the slip coefficient, ∈ represents the slip-related modeling error;

[0051] Generally speaking, the friction force F friction It mainly reflects the resistance between the robot and the ground, and its magnitude is proportional to the speed of the robot. As an option, the sliding force F slip Through the slip coefficient F slip The relationship between ∈ and linear velocity v is described, and an error term ∈ is added to reflect the nonlinear factors that cannot be accurately quantified in actual modeling;

[0052] As an extension of nonlinear dynamics modeling, this embodiment also introduces a systematic analysis of modeling errors, using a generalized disturbance term d(t) to supplement the model errors:

[0053]

[0054] Where d(t) represents the generalized disturbance term, which is used to reflect the additional impact of ground friction changes, load changes, and environmental disturbances on the system;

[0055] At the same time, in order to adapt to different types of robots or motion environments, the parameters of the above dynamic model can be adjusted according to actual needs, for example:

[0056] For robots of different masses, the system inertia characteristics can be adjusted by changing the value of m; for floors of different materials, the friction coefficient c can be changed. f Adjust friction characteristics; in scenarios with large tire slip, F can be further improved slip The value of , ensures that the model fully describes the slip force;

[0057] The disturbance estimation module is used to estimate external disturbances and system modeling errors in real time through an extended state observer, and to generate disturbance compensation data based on the dynamic model established by the nonlinear dynamic modeling module;

[0058] Generally, wheeled robots are subject to a variety of disturbances during operation, including changes in ground friction, external impacts, tire slip, and modeling errors. As an option, the present invention introduces an extended state observer to observe the disturbances as part of the system state in real time, and uses its output for subsequent control compensation.

[0059] The design of the disturbance estimation module specifically includes the following aspects:

[0060] First, an extended state observer is introduced to perform real-time disturbance estimation on the system dynamics model output by the nonlinear dynamics modeling module. The extended state observer expands the unknown disturbance into the system state variables, thus achieving online observation and estimation of the disturbance. In the design of the extended state observer, the dynamic characteristics of the robot motion are taken into account, and the following state observation equation is adopted:

[0061]

[0062] Where: z1 represents the observation position of the system; z2 represents the observation speed of the system; z3 represents the extended state, that is, the real-time estimated value of the disturbance; e = y-z1 is the observation error of the system, and y is the measured value; β1, β2, and β3 are the gain parameters of the extended state observer, which are used to adjust the convergence speed of the observation. By adjusting the parameters β1, β2, and β3, the dynamic performance of the observer can be optimized, enabling it to accurately estimate disturbances in different environments.

[0063] Secondly, the disturbance estimate of the extended state observer is calculated by the following formula:

[0064]

[0065] in, The real-time estimated disturbance value represents the generalized disturbance to which the system is subjected in the current state, including the influence of the external environment (such as the change in ground roughness) and the modeling error (such as the deviation caused by the simplified assumptions of the dynamic model). At the same time, the actual disturbance can be measured experimentally. Deviation;

[0066] In addition, the implementation of the extended state observer can also be combined with a variety of filtering algorithms to optimize the disturbance estimation results. A low-pass filter is added to the output end to filter out the influence of high-frequency noise. The design formula of the filter is:

[0067]

[0068] in: The filtered disturbance estimate is used to reduce high-frequency noise in the disturbance estimate, making it smoother and more stable. τ is the time constant of the filter; s is the complex frequency under the Laplace transform. By adding a low-pass filter, the impact of noise interference on the control system can be reduced while maintaining the sensitivity of the disturbance estimate.

[0069] In order to further improve the real-time performance and accuracy of the disturbance estimation module, this embodiment also introduces a joint estimation strategy of the disturbance observer and the extended state observer. Specifically, the disturbance observer is used to compensate for the problem that the extended state observer is insufficiently responsive to high-frequency disturbances. Its core formula is:

[0070]

[0071] in: The disturbance estimate of the output of the group disturbance observer; k d is the gain parameter of the disturbance observer; in the joint estimation strategy, the output of the disturbance observer is weightedly fused with the estimated value of the extended state observer to obtain a more accurate disturbance estimation result. The disturbance estimation value output by the extended state observer (ESO) is used to estimate external disturbances or modeling errors in real time; is the error term of the disturbance observer, which represents the estimated value d of the disturbance observer obs and the reference disturbance estimate Deviation between

[0072] Finally, the output of the extended state observer It is directly sent to the sliding mode control module and the model predictive control module to adjust the control gain and optimize the control input. In the sliding mode control module, Used to dynamically adjust the sliding mode gain; in the model predictive control module, It is used as the disturbance compensation term in the optimization objective function. This design ensures the close coupling between the disturbance estimation module and the control module, providing technical support for the system's anti-interference ability.

[0073] The sliding mode control module is used to achieve rapid convergence of trajectory deviation by constructing a sliding mode surface based on the disturbance compensation data provided by the disturbance estimation module, and to enhance the system's anti-interference ability by dynamically adjusting the sliding mode gain;

[0074] The sliding mode control module is one of the core modules in the wheeled robot's trajectory tracking and interference-resistant control system. Its function is to achieve rapid convergence of trajectory deviations by constructing a sliding mode surface, and to enhance the system's interference resistance by combining dynamic gain adjustment strategies. This module is based on the dynamic model provided by the nonlinear dynamics modeling module and utilizes real-time disturbance information provided by the disturbance estimation module to ensure that the controller can maintain trajectory tracking stability and accuracy in complex environments and under external disturbances.

[0075] The sliding mode control module of the present invention reduces the chattering problem of traditional sliding mode controllers by designing a high-order sliding mode control law. It also enables the controller to adapt to changes in external disturbances through a dynamic gain adjustment mechanism. Furthermore, the sliding mode control module is tightly integrated with the subsequent model predictive control module, achieving a coordinated optimization of anti-interference capability and trajectory tracking accuracy.

[0076] In this embodiment, the design of the sliding mode control module includes the following aspects:

[0077] First, to achieve fast convergence of trajectory tracking, a sliding surface is designed based on the nonlinear dynamic model of the wheeled robot and the target trajectory information. Specifically, it is assumed that the deviation between the current state of the robot and the target trajectory is:

[0078] e x =xx ref

[0079] e y =yy ref

[0080] Among them, e x and e y are the deviations between the robot’s current position and the target position in the x and y directions respectively; ref and y ref is the reference point of the target trajectory.

[0081] Based on the above deviations, the sliding surface is designed as:

[0082]

[0083] Where s is the sliding surface; is the derivative of the deviation; λ>0 is the sliding surface parameter, which is used to adjust the system convergence speed. As a possible implementation method, the value of λ can be dynamically adjusted according to the complexity of the robot's operating environment to balance the relationship between convergence speed and control smoothness;

[0084] Secondly, in order to achieve stable trajectory tracking control in complex environments, a high-order sliding mode control law is designed in this embodiment. The goal of the high-order sliding mode control law is to reduce the chattering phenomenon in traditional sliding mode control by introducing feedback of the sliding surface derivative, while enhancing the ability to suppress external disturbances. The specific control law is:

[0085]

[0086] Where u is the control input; sgn(s) is the sign function used to determine the adjustment direction of the system state; k1 and k2 are the sliding mode control gains;

[0087] In the above control law, the first term -k1sgn(s) realizes the basic anti-interference ability of traditional sliding mode control; the second term The feedback of the sliding surface derivative is introduced to reduce the high-frequency fluctuation of the control input;

[0088] As an option, in order to adapt to changes in external disturbances, a dynamic gain adjustment mechanism is also designed in this embodiment. Specifically, the sliding mode gain According to the disturbance information provided by the disturbance estimation module Dynamic adjustment, the adjustment formula is:

[0089]

[0090] Among them, k1 is the dynamic gain parameter of the sliding mode controller, which is used to adjust the control strength of the sliding mode controller, k 1,min is the minimum value of the sliding mode gain, which is used to ensure the basic performance of the controller when there is no disturbance; κ is the gain adjustment coefficient, which is used to adjust the influence of disturbance changes on the gain; as a possible implementation method, the value of κ can be calibrated offline or adjusted online according to the disturbance characteristics of the robot operating environment. is the absolute value of the disturbance estimate, which is used to quantify the intensity of the current disturbance and affects the dynamic adjustment of the sliding mode control gain;

[0091] The sliding mode control module in this embodiment also combines the output information of the disturbance estimation module To further optimize the control law, As a compensation term of the control input, the controller's ability to suppress unknown disturbances can be enhanced. The optimized control law is:

[0092]

[0093] Where γ is the disturbance compensation coefficient, which is used to adjust the weight of disturbance compensation; u is the control input of the sliding mode controller, which represents the control force or control command acting on the system; k1 is the gain parameter of the sliding mode controller, which is used to adjust the contribution strength of the sliding surface term to the control input; sgn(s) is a sign function, defined as Indicates the direction of the sliding surface s, k2 is the gain parameter of the sliding surface derivative feedback, which is used to adjust the contribution strength of the derivative term to the control input. is the derivative of the sliding surface, which represents the rate of change of the trajectory deviation s;

[0094] In some embodiments, in order to further improve the performance of the sliding mode controller, the sliding mode surface can be expanded. For example, in the multi-degree-of-freedom control scenario, a higher-dimensional sliding mode surface can be designed, specifically in the form of:

[0095] s=E+Λe

[0096] Where s is the high-dimensional sliding surface; E is the deviation matrix; Λ is the weight matrix. Through the design of the high-dimensional sliding surface, the collaborative optimization of multiple control objectives can be achieved, such as simultaneously controlling the position deviation and velocity deviation of the robot.

[0097] During use, the sliding mode control module can be used in conjunction with the model predictive control module to form a hybrid control strategy based on priority allocation. For example, in large disturbance scenarios, the sliding mode control module prioritizes outputting control instructions to quickly suppress disturbances; in small disturbance scenarios, it relies more on the model predictive control module to achieve refined optimization of trajectory tracking.

[0098] The model predictive control module is used to construct an optimization objective function based on the disturbance prediction data provided by the disturbance estimation module and optimize the control input to improve trajectory tracking accuracy and system robustness;

[0099] The model predictive control module is one of the key components of the wheeled robot trajectory tracking and anti-interference control system of the present invention. The main function of this module is to optimize the control input by predicting future state and disturbance changes, thereby achieving refined control of trajectory tracking, while enhancing the robustness and real-time performance of the system. This module is closely integrated with the nonlinear dynamics modeling module and the disturbance estimation module. By dynamically calculating the control input through optimizing the objective function, the wheeled robot can accurately follow the predetermined trajectory in a complex environment. The model predictive control module of the present invention specifically combines the disturbance prediction mechanism and the rapid optimization strategy. It can not only dynamically adjust the control parameters, but also respond to external disturbances in real time in a dynamic environment, thereby improving the control accuracy of the system.

[0100] In this embodiment, the design of the model predictive control module includes the following aspects:

[0101] First, for the nonlinear dynamic model of the wheeled robot, an optimization objective function is constructed to describe the trade-off relationship between trajectory deviation, control input, and disturbance compensation. Specifically, the optimization objective function is designed as follows:

[0102]

[0103] Where: J is the optimization objective function; N is the length of the prediction time domain; x k is the predicted state of the robot at step k; x ref,k is the reference state of step k; u k is the control input for step k; is the disturbance prediction value of the kth step; Q, R and P are weight matrices, corresponding to the weights of state deviation, control input and disturbance compensation respectively;

[0104] The value of the weight matrix needs to be adjusted according to the actual application scenario. For example, in scenarios with high requirements for trajectory tracking accuracy, the value of Q can be appropriately increased; and in scenarios with strict energy consumption restrictions, the value of R can be appropriately increased.

[0105] In order to achieve disturbance compensation, a data-driven disturbance prediction method is designed in this embodiment. Specifically, the disturbance prediction module uses a recurrent neural network (such as LSTM) to model historical disturbances and predict the changing trend of future disturbances. The disturbance prediction formula is:

[0106]

[0107] in: is the predicted value of the disturbance in the next step; f LSTm is the disturbance prediction function based on the long short-term memory network; x(t) is the current state; u(t) is the current control input; d(t) is the current disturbance value;

[0108] The disturbance prediction module, through online training on historical data, can adapt to disturbance changes in different environments. For example, when the robot is operating in a windy outdoor environment, the disturbance prediction module can identify the impact of wind on the system and adjust the control input in a timely manner, thereby enhancing the system's anti-interference ability.

[0109] As an option, in order to improve the computational efficiency of model predictive control, a fast optimization solution method is introduced in this embodiment. The core idea of ​​the optimization solution is to use the sparse matrix characteristics of the system to reduce the complexity of the optimization problem to a range that can be solved in real time. The specific formula is as follows:

[0110] Ax=b

[0111] Where: A is a sparse matrix, representing the constraints in the optimization problem; x is the optimization variable to be solved; b is the target value; sparse matrix technology can effectively reduce the amount of redundant calculations in the solution process, thereby improving optimization efficiency;

[0112] In addition, this embodiment also designs an event trigger mechanism to reduce unnecessary optimization calculations. When the trajectory deviation or disturbance change is lower than the set threshold, the control system will continue to use the control input at the previous moment without performing a new optimization calculation. The specific implementation of the event trigger mechanism is as follows:

[0113] If ||x k -x Ref,k ||<ε and Then u k =u k-1

[0114] Where: ε is the tolerance threshold of trajectory deviation; δ is the tolerance threshold of disturbance change; u k and u k-1 are the control inputs at the current and previous moments respectively. The event triggering mechanism can effectively reduce the computational burden of the control system while maintaining the dynamic response capability of the system;

[0115] Finally, the model predictive control module in this embodiment works closely with the sliding mode control module. In some embodiments, when the system is affected by a large disturbance, the sliding mode control module takes priority to quickly suppress the disturbance; when the trajectory deviation is small, the model predictive control module takes over the control task to optimize the accuracy of trajectory tracking. Through this collaborative strategy, a dynamic balance can be achieved between trajectory tracking performance and anti-interference ability.

[0116] The fusion control module is used to combine the outputs of the sliding mode control module and the model predictive control module to achieve coordination between anti-interference ability and trajectory tracking accuracy by dynamically adjusting weights;

[0117] This module combines the advantages of the sliding mode control module and the model predictive control module, dynamically adjusting their control weights to achieve comprehensive optimization of trajectory tracking accuracy and anti-interference capability. The output of the fusion control module serves as the final control instruction, directly driving the robot actuator, thereby ensuring the system's efficient and robust trajectory tracking in complex dynamic environments. The fusion control module of the present invention establishes a dynamic weight adjustment mechanism, enabling the system to automatically adjust the ratio of sliding mode control and model predictive control based on the actual environmental disturbance level and trajectory deviation, to achieve an optimal balance between the two performances.

[0118] In this embodiment, the specific design of the fusion control module includes the following aspects:

[0119] First, in order to achieve the coordinated operation of the sliding mode control module and the model predictive control module, a fusion control law is designed to integrate the control outputs of the two. Specifically, the expression of the fusion control law is:

[0120] u=αu SMC +(1-α)u MPC

[0121] Where: u is the output control quantity of the fusion control module; u SMC is the output control quantity of the sliding mode control module; u MPC is the output control quantity of the model predictive control module; α is the dynamic weight factor, and its value range is 0≤α≤1;

[0122] When the external disturbance is large or the system trajectory deviation is large, the weight factor α should be biased towards the sliding mode control module to quickly respond to the disturbance and reduce the trajectory deviation; when the disturbance is small and the trajectory deviation is small, the weight factor α should be biased towards the model predictive control module to fully utilize its optimized control performance;

[0123] Secondly, in order to dynamically adjust the weight factor α, this embodiment designs a weight adjustment mechanism based on the disturbance estimate and trajectory deviation. Specifically, the calculation formula of α is:

[0124]

[0125] in: Real-time disturbance estimation value provided to the disturbance estimation module; is the modulus of trajectory deviation, e x =xx ref和 e y =y are the deviations between the robot’s current position and the target trajectory. Specifically, when the disturbance When it is larger, α tends to be close to 1, at which point the sliding mode control module dominates and the control The anti-interference performance of the controller is enhanced; when the trajectory deviation e is large but the disturbance is small, α tends to approach 0. At this time, the model predictive control module plays a leading role to optimize the accuracy of trajectory tracking;

[0126] In one possible implementation, to avoid adverse effects of drastic changes in the weighting factors on control stability, a smoothing function is introduced into the dynamic adjustment of α in this embodiment. For example, the following smoothing adjustment formula can be used:

[0127]

[0128] Where: ∈>5 is a smoothing factor, which is used to limit the rapid change of α. By introducing the smoothing factor, the instability of the controller output caused by fluctuations in disturbances and trajectory deviations can be avoided in actual operation;

[0129] As an option, the fusion control module in this embodiment also supports limiting the upper and lower limits of the weight factors to further enhance the robustness of the system. Specifically, the weight factors can be limited to a certain range, for example:

[0130] α=max(min(α,α max ),α min )

[0131] Where: α max and α min are the maximum and minimum values ​​of the weight factor respectively. In an environment with strong external interference, α can also be maxSet it to a larger value to increase the weight of the sliding mode control module;

[0132] In this embodiment, the output u of the fusion control module directly drives the robot actuator. As a possible implementation method, when the control task involves multi-degree-of-freedom motion (for example, simultaneous control of the robot's position and angle), the fusion control law can be expanded to a multidimensional form. The specific formula is:

[0133] u=αu SMC +(1-α)u MPC

[0134] Among them: u is the vector form of multi-degree-of-freedom control quantity; u SMC and u MPC They are the output vectors of sliding mode control and model predictive control respectively. Through multi-dimensional fusion control, the coordinated control of the robot's position, speed and posture can be achieved, thereby improving the overall control performance of the system;

[0135] Finally, the fusion control module in this embodiment also supports linkage with the dynamic trajectory replanning module. When the robot needs to avoid obstacles or adjust the target trajectory in a dynamic environment, the fusion control module can dynamically adjust the weight factor α according to the newly generated reference trajectory, thereby achieving efficient tracking of the new trajectory.

[0136] The trajectory optimization module is used to adjust the target trajectory based on the dynamic window method and provide the adjusted target trajectory to the fusion control module to ensure the motion stability of the robot in a dynamic environment;

[0137] This module is closely connected with the aforementioned fusion control module, disturbance estimation module, and nonlinear dynamics modeling module. Based on the robot's real-time state information, environmental obstacle information, and target trajectory requirements, it dynamically generates or adjusts trajectory points to ensure that the robot can move smoothly in complex environments.

[0138] The trajectory optimization module of the present invention uses the dynamic window method (DWA) and the cost function minimization method, combined with the robot's kinematic and dynamic constraints, to generate dynamic trajectory points that meet the motion characteristics, and further provide an optimized reference trajectory for the subsequent control module;

[0139] In this embodiment, the specific design of the trajectory optimization module includes the following aspects:

[0140] First, a real-time feasible trajectory search space is constructed based on the dynamic window method (DWA). The basic idea of ​​the dynamic window method is to generate a set of feasible trajectory candidates within a limited time window by considering the robot's velocity constraints, acceleration constraints, and environmental obstacle constraints. Specifically, the dynamic window search space can be defined as:

[0141] V={v|v min ≤v≤v max ,a min ≤a≤a max}

[0142] Where: V is the current speed of the robot; v min and v msc are the minimum and maximum values ​​of the speed respectively; a min and a max are the minimum and maximum acceleration values, respectively. For robots with high inertia, the value ranges of v and a can be appropriately narrowed to reduce the computational complexity of trajectory planning.

[0143] Secondly, in order to ensure that the generated trajectory points can avoid dynamic obstacles in the environment, this embodiment designs a trajectory optimization method based on a cost function. The construction of the cost function takes into account the following aspects:

[0144] J trajectory =w1J goal +w2J collision +w3J smoothness

[0145] Among them: J trajectory is the total trajectory cost; J goal represents the distance cost between the trajectory point and the target position; J collision represents the collision risk cost between the trajectory point and the environmental obstacle; J smoothness represents the smoothness cost of the trajectory; w1, w2, and w3 are the weight coefficients of the cost function, which are used to adjust the influence of different factors on trajectory optimization. By optimizing the cost function, an optimal trajectory that takes into account goal achievement, obstacle avoidance, and trajectory smoothness can be selected;

[0146] The trajectory optimization module in this embodiment further incorporates the nonlinear dynamic constraints of the robot to ensure that the generated trajectory can actually be executed by the robot. Specifically, the constraints for trajectory optimization include:

[0147] Dynamic constraints: such as maximum speed, maximum acceleration, and maximum angular velocity; Kinematic constraints: such as the minimum turning radius that a wheeled robot can achieve;

[0148] For example, for the maximum speed v max The constraint can be expressed by the following inequality:

[0149] v≤v max

[0150] Among them, v is the current linear velocity of the robot, v max is the maximum allowable value of the robot's linear velocity, for the turning radius rmin The constraint can be expressed by the curvature κ as:

[0151]

[0152] Among them, κ is the trajectory curvature, which indicates the degree of curvature of the robot along the path, r min The minimum turning radius that the robot can achieve. By adding these constraints, we can effectively avoid the problem that the trajectory optimization results cannot be executed;

[0153] The trajectory optimization module in this embodiment also supports dynamic trajectory replanning. When the robot detects a dynamic obstacle, the module recalculates a set of feasible trajectories based on the dynamic window method and selects the trajectory with the lowest cost as the new target trajectory. Specifically:

[0154] Upper-layer optimization: Generate an initial reference path globally; Lower-layer optimization: Make local adjustments to the upper-layer path based on current environmental information. This hierarchical optimization strategy can significantly reduce the computational complexity of trajectory replanning.

[0155] Finally, the trajectory optimization module in this embodiment works in conjunction with the fusion control module to ensure that the optimized trajectory can be efficiently executed. For example, when the trajectory optimization module generates a new target trajectory, the fusion control module will readjust the control strategy based on the target trajectory to achieve smooth following of the new trajectory.

[0156] While 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 these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A wheeled robot trajectory tracking anti-interference control system, characterized in that: include: Nonlinear dynamics modeling module, used to establish a nonlinear dynamics model of a wheeled robot to describe the robot's motion characteristics under the influence of ground friction, slip force, and modeling errors; The disturbance estimation module is used to estimate external disturbances and system modeling errors in real time through an extended state observer, and to generate disturbance compensation data based on the dynamic model established by the nonlinear dynamic modeling module; The sliding mode control module is used to achieve rapid convergence of trajectory deviation by constructing a sliding mode surface based on the disturbance compensation data provided by the disturbance estimation module, and to enhance the system's anti-interference ability by dynamically adjusting the sliding mode gain; The model predictive control module is used to construct an optimization objective function based on the disturbance prediction data provided by the disturbance estimation module and optimize the control input to improve trajectory tracking accuracy and system robustness; The fusion control module is used to combine the outputs of the sliding mode control module and the model predictive control module to achieve coordination between anti-interference ability and trajectory tracking accuracy by dynamically adjusting weights; The trajectory optimization module is used to adjust the target trajectory based on the dynamic window method and provide the adjusted target trajectory to the fusion control module to ensure the motion stability of the robot in a dynamic environment.

2. A wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The nonlinear dynamics model established by the nonlinear dynamics modeling module includes: Kinematic model, used to describe the relationship between the position, heading angle and velocity of the wheeled robot in the global coordinate system; The dynamic model is used to describe the dynamic relationship between a wheeled robot under the action of ground friction, slip force and driving force, where the slip force is represented by the functional relationship between the slip coefficient and the speed.

3. A wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The disturbance estimation module is built based on the extended state observer, which is specifically: State observation equation, used to observe the changing trend of robot state variables; Disturbance estimation equations are used to estimate unknown disturbances and modeling errors as extended states in real time.

4. A wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The sliding mode control module achieves rapid convergence of trajectory deviation by designing a sliding surface, which is dynamically adjusted according to the deviation between the robot's current position and the target trajectory to ensure rapid response of the system.

5. The wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The sliding mode control module achieves rapid convergence of trajectory deviation by designing a sliding surface, which is dynamically adjusted according to the deviation between the robot's current position and the target trajectory to ensure rapid response of the system.

6. A wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The model predictive control module includes: An optimization objective function construction unit, used to construct an optimization objective function according to trajectory tracking deviation, control input and disturbance prediction value; The disturbance prediction unit is used to predict future disturbance changes using a recurrent neural network based on historical states, control inputs, and disturbance data; The optimization solver is used to optimize the control input in real time through sparse matrix calculation method and event trigger mechanism.

7. The wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The fusion control module dynamically adjusts the output weights of the sliding mode control module and the model predictive control module according to external disturbances and trajectory deviations, where: The weights are dynamically assigned based on the ratio of the disturbance estimate to the trajectory deviation, which is used to enhance the output weight of the sliding mode control module when the disturbance is large, and to enhance the output weight of the model predictive control module when the trajectory deviation is small.

8. The wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The trajectory optimization module adjusts the target trajectory based on the dynamic window method, where: The dynamic window method combines the robot velocity constraint and the dynamic obstacle detection results to generate a new local target trajectory in real time; The newly generated target trajectory is optimized by minimizing the objective function of local deviation and trajectory tracking deviation.

9. The wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The control system adopts a hierarchical control architecture, including: The upper control module is used to generate global target trajectories and provide dynamic environment information; The lower-level control module is used to realize trajectory tracking and anti-interference control based on the target trajectory and dynamic information provided by the upper layer, combined with the sliding mode control module and the model predictive control module.

10. The wheeled robot trajectory tracking anti-interference control system according to claim 1, characterized in that: The control system is implemented through an embedded hardware platform and is modularly designed based on the robot operating system ROS. The perception module is used to obtain environmental information, the decision module is used for dynamic trajectory adjustment and controller output, and the execution module is used to implement motion control commands.

Citation Information

Cited By

  • Automatic deviation correction method for inspection path of inspection robot

    CN120742905A

  • An automatic deviation correction method for a patrol robot patrol path

    CN120742905B

  • Adaptive intelligent control method for machine tool functional components

    CN122569194A