A predictive control method, device, and wheeled robot for a wheeled robot.

Abstract

This invention provides a predictive control method, device, and wheeled robot for a wheeled robot, relating to the field of automatic control technology. The predictive control method for the wheeled robot includes: acquiring the robot's operational information; performing trajectory planning based on position information and an environmental map to obtain reference trajectory information; obtaining a control increment sequence based on state information, reference trajectory information, and a pre-built predictive model; obtaining a target control quantity to compensate for the disturbance based on the control increment sequence, disturbance observations, and reference trajectory information; controlling the operation of the wheeled robot according to the target control quantity; and returning to the step of obtaining the control increment sequence when a preset trigger condition is met at the current sampling time, until the endpoint position is reached. This invention obtains independent control quantities for each wheel by compensating for independent disturbances in each wheel using disturbance observations, improving trajectory tracking accuracy. Optimization is only performed when the trigger condition is met, avoiding wasted computational power and improving operational efficiency.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, and more specifically, to a predictive control method, device, and wheeled robot for a wheeled robot. Background Technology

[0002] In recent years, robots have been widely used in fields such as field exploration, seabed exploration, extraterrestrial exploration, dangerous rescue, and household services, replacing humans. With the continuous development of autonomous driving technology, robots have also evolved from traditional remote control to automatic control that can achieve trajectory tracking.

[0003] Traditional robot control methods use PID controllers to perform trajectory tracking tasks. However, when using PID controllers to control robots, oscillations or untimely control are prone to occur, resulting in poor trajectory tracking accuracy. In recent years, MPC controllers have emerged. Current technologies using MPC controllers to perform trajectory tracking tasks typically treat the robot as a whole, predict multiple future control outputs based on the deviation between the robot's actual walking trajectory and the pre-planned trajectory, and then automatically execute the first of these to achieve trajectory tracking. However, in actual trajectory tracking control, this often requires a huge amount of computing power, resulting in low efficiency and relatively poor tracking accuracy. Summary of the Invention

[0004] The problem addressed by this invention is how to achieve high efficiency and accuracy in trajectory tracking for wheeled robots.

[0005] To address the above problems, this invention provides a predictive control method for a wheeled robot, comprising the following steps: The operation information of the wheeled robot is obtained, wherein the operation information includes an environmental map, location information, status information and disturbance observations, the location information includes a starting position and an ending position, and the disturbance observations include independent disturbances corresponding to each wheel of the wheeled robot; Based on the location information and the environment map, trajectory planning is performed to obtain reference trajectory information, wherein the reference trajectory information includes a reference trajectory with multiple nodes; Based on the state information, the reference trajectory information, and the pre-built prediction model, a control increment sequence is obtained, wherein the control increment sequence includes N control increments for correcting the state variables of the wheeled robot at N sampling times in the future. Based on the control increment sequence, the disturbance observations, and the reference trajectory information, a target control quantity for compensating for the disturbance is obtained, wherein the target control quantity includes independent control quantities corresponding to each of the wheels; The operation of the wheeled robot is controlled according to the target control quantity; When the wheeled robot meets the preset triggering condition at the current sampling time, it returns to the step of obtaining the control increment sequence based on the state information, the reference trajectory information and the pre-built prediction model, until it reaches the endpoint position. The triggering condition includes the perturbation observation value being greater than the preset perturbation threshold.

[0006] Optionally, the disturbance observations include instantaneous disturbance observations and cumulative disturbance observations, and the disturbance thresholds include instantaneous disturbance thresholds and cumulative disturbance thresholds; the step of returning to the step of obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model when the wheeled robot meets the preset triggering condition at the current sampling time, until reaching the endpoint position, includes: The triggering condition is met when the instantaneous disturbance observation value is greater than the instantaneous disturbance threshold, or when the cumulative disturbance observation value is greater than the cumulative disturbance threshold. The trigger condition is met when the number of sampling times from the last sampling time that met the trigger condition to the current sampling time is equal to N. Return to the step of obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, until the endpoint position is reached.

[0007] Optionally, the triggering condition satisfies: ; in, This indicates the sampling time at which the triggering condition will be met next. Indicates the current sampling time. This indicates the sampling time at which the triggering condition was last met. This represents the instantaneous disturbance observation value. This represents the instantaneous disturbance threshold. This indicates the number of sampling times elapsed from the last sampling time when the trigger condition was met to the current sampling time; or; ; in, This represents the cumulative disturbance observation value from the last sampling time when the trigger condition was met to the current sampling time. This represents the cumulative disturbance threshold.

[0008] Optionally, the wheeled robot includes four wheels and, before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, further includes: Construct a kinematic model of the wheeled robot, wherein the kinematic model satisfies: ; Among them, the state variables of the wheeled robot , This indicates the posture of the wheeled robot. , Indicates the first The coordinates of the aforementioned wheels, Indicates the first The turning angle of the aforementioned wheel, Indicates the first The speed of the aforementioned wheels and Representing trigonometric functions and ; Based on the kinematic model, the reference state equation of the wheeled robot is obtained, and the reference state equation satisfies: ; in, This represents the reference state variable of the wheeled robot. This represents the reference control variable of the wheeled robot; The reference state equation is linearized to obtain a first linear error model, which satisfies the following: ; in, , This represents the dynamic error of the state variable. This represents the dynamic error of the control variable, which is obtained based on the state variable and the kinematic model. for about Jacobian matrix, for about The Jacobian matrix; The first linear error model is discretized to obtain a discrete error model, wherein the discrete error model satisfies: ; in, This represents the error of the state variable of the wheeled robot at the next sampling time. This represents the error of the state variable of the wheeled robot at the current sampling time. and It is a matrix relating the speed and the turning angle of each of the aforementioned wheels.

[0009] Optionally, before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, the method further includes: The prediction model is obtained based on the discrete error model, wherein the prediction model is used to predict the future state variables of the wheeled robot based on the current state variables and the control variables; Based on the prediction model, a cost function is constructed; The constraints of the cost function are constructed, wherein the constraints include the target control quantity being greater than or equal to a preset minimum control quantity and less than or equal to a preset maximum control quantity.

[0010] Optionally, before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, the method further includes: Based on the first linear error model and the independent disturbances experienced by each wheel, a second linear error model is obtained, which satisfies the following: ; in, This represents the velocity and angular velocity disturbances experienced by each of the aforementioned wheels; The second linear error model is transformed into the following form: ; A disturbance observer is designed based on the second linear error model, and the disturbance observer satisfies: ; in, This represents the observed disturbance value. This represents the nonlinear gain of the disturbance observer. Indicates intermediate variables. For about Nonlinear functions.

[0011] Optionally, the state information includes the current state quantity of the wheeled robot at the current sampling time; the reference trajectory information includes reference state quantities corresponding to multiple nodes; obtaining the target control quantity to compensate for the disturbance based on the control increment sequence, the disturbance observation value, and the reference trajectory information includes: Based on the reference state quantity and the kinematic model, the reference control quantity corresponding to the current sampling time is obtained; The target control quantity is obtained based on the control increment sequence, the reference control quantity, and the disturbance observation value. The control increment sequence is obtained by optimizing the cost function based on the current state quantity, the reference state quantity, and the constraints. At the starting position and the sampling time satisfying the trigger condition, the target control increment satisfies: ; in, This represents the target control quantity corresponding to the current sampling time. This represents the reference control quantity corresponding to the current sampling time. This represents the control increment corresponding to the current sampling time. This represents the disturbance observation value corresponding to the current sampling time; At sampling times other than those meeting the triggering condition, the target control increment satisfies:

[0012] in, Indicates in The target control quantity corresponding to the sampling time. Indicates in The control increment corresponding to the sampling time.

[0013] Optionally, the step of performing trajectory planning based on the location information and the environmental map to obtain reference trajectory information includes: Create the environment map based on SLAM; Based on the heuristic algorithm and the environment map, the optimal path between the starting point and the ending point is obtained; The optimal path is smoothed to obtain the reference trajectory; The reference trajectory is divided to obtain a plurality of nodes evenly distributed on the reference trajectory; Based on the curvature and preset acceleration constraints at the node, the reference state variables corresponding to the node are planned to obtain the reference trajectory information.

[0014] This invention provides a reliable data foundation for generating target control quantities by acquiring operational information such as environmental maps, location information, state information, and disturbance observations. Based on the location information and environmental map, trajectory planning is performed to obtain reference trajectory information, providing a desired reference trajectory for subsequent trajectory tracking control. The reference trajectory obtained through trajectory planning ensures the wheeled robot's safe and stable arrival at the destination position, and also provides a reference basis for generating target control quantities. Based on the state information, reference trajectory information, and a pre-built prediction model, the control increment sequence corresponding to the next N sampling times is predicted, making it easier for the wheeled robot's actual walking trajectory to closely approximate the desired reference trajectory, thereby achieving high-precision trajectory tracking. Based on the control increment sequence, disturbance observations, and reference trajectory information, the following is obtained: The system achieves independent disturbance compensation for each wheel by obtaining the target control quantity for disturbance compensation. Based on the actual disturbance experienced by each wheel, it specifically enhances the anti-interference capability of each wheel, further improving trajectory tracking accuracy. Furthermore, it independently controls each wheel of the wheeled robot based on the target control quantity, improving the robot's flexibility and helping to avoid accidents such as rollovers and slippage, ensuring the stability and safety of the trajectory tracking process. When the wheeled robot meets the preset triggering conditions at the current sampling moment, combined with the independent disturbance experienced by each wheel, the target control quantity can be optimized before a certain trajectory deviation occurs, until the endpoint is reached. This significantly improves the wheeled robot's adaptability to complex terrain and road conditions while enhancing trajectory tracking accuracy and operational efficiency.

[0015] The present invention also provides a predictive control device for a wheeled robot, comprising: The acquisition module is used to acquire the operation information of the wheeled robot, wherein the operation information includes an environmental map, location information, status information and disturbance observations, the location information includes a starting position and an ending position, and the disturbance observations include independent disturbances corresponding to each wheel of the wheeled robot; A planning module is used to perform trajectory planning based on the location information and the environmental map to obtain reference trajectory information, wherein the reference trajectory information includes a reference trajectory with multiple nodes; The prediction module is used to obtain a control increment sequence based on the state information, the reference trajectory information and the pre-built prediction model, wherein the control increment sequence includes N control increments for correcting the state variables of the wheeled robot at N sampling times in the future. The compensation module is used to obtain a target control quantity for compensating the disturbance based on the control increment sequence, the disturbance observation value and the reference trajectory information, wherein the target control quantity includes the independent control quantity corresponding to each of the wheels; A control module is used to control the operation of the wheeled robot according to the target control quantity; The triggering module is used to return the operation of obtaining the control increment sequence based on the state information, the reference trajectory information and the pre-built prediction model when the wheeled robot meets the preset triggering conditions at the current sampling time, until the endpoint position is reached. The triggering conditions include the disturbance observation value being greater than the preset disturbance threshold.

[0016] The predictive control device for wheeled robots and the predictive control method for wheeled robots provided by this invention have essentially the same advantages as the prior art, and will not be repeated here.

[0017] The present invention also provides a wheeled robot, including a computer-readable storage medium storing a computer program and a processor, wherein the computer program is read and executed by the processor to implement the predictive control method for the wheeled robot as described above.

[0018] The wheeled robot and the predictive control method for wheeled robots provided by this invention have essentially the same advantages as existing technologies, and will not be elaborated further here. Attached Figure Description

[0019] Figure 1 This is a flowchart of a predictive control method for a wheeled robot according to an embodiment of the present invention; Figure 2 This is a kinematic model of a wheeled robot according to an embodiment of the present invention; Figure 3 This is another flowchart of a predictive control method for a wheeled robot according to an embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the accompanying drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0021] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0022] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; the term "optionally" means "optional embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first", "second", etc., mentioned in this invention are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0023] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0024] like Figure 1 As shown, an embodiment of the present invention provides a predictive control method for a wheeled robot, comprising the following steps: S1: Obtain the operation information of the wheeled robot, including environmental map, location information, status information and disturbance observations. The location information includes the starting position and the ending position, and the disturbance observations include the independent disturbances corresponding to each wheel of the wheeled robot.

[0025] Specifically, the state information referred to in this invention represents various information reflecting the operating state of the wheeled robot, such as the speed, angular velocity, and coordinates of each wheel, which can be obtained through speed detection devices, positioning devices, etc. The disturbance observation values ​​referred to in this invention include the independent disturbances corresponding to each wheel of the wheeled robot, which can be obtained through a disturbance observer. Obtaining the environmental map and position information of the wheeled robot facilitates subsequent scientific and reasonable trajectory planning between the starting position and the ending position, where the starting position can be the robot's current location. During the wheeled robot's movement, it will be subject to various disturbances due to poor road conditions, etc. Obtaining the independent disturbances corresponding to each wheel is beneficial for subsequent accurate compensation based on the actual disturbance situation, achieving high-precision trajectory tracking.

[0026] In this embodiment, acquiring operational information such as environmental maps, location information, status information, and disturbance observations provides a data foundation for the subsequent generation of target control quantities, which is beneficial to improving the accuracy of trajectory tracking.

[0027] S2: Based on location information and environmental map, perform trajectory planning to obtain reference trajectory information, which includes a reference trajectory with multiple nodes.

[0028] Specifically, a graph-based search algorithm can be used to find an optimal path between the starting and ending positions, such as Dijkstra's algorithm or A* algorithm. The obtained optimal path is then smoothed and divided into multiple nodes to obtain a reference trajectory with multiple nodes.

[0029] In this embodiment, trajectory planning is performed based on location information and environmental map to obtain reference trajectory information, which provides the desired reference trajectory for subsequent trajectory tracking control. The reference trajectory obtained through trajectory planning can ensure that the wheeled robot can safely and smoothly reach the destination position, and also provides a data basis for the subsequent generation of target control quantities.

[0030] S3: Based on the state information, reference trajectory information, and pre-built prediction model, obtain the control increment sequence, which includes N control increments used to correct the state variables of the wheeled robot at the next N sampling times.

[0031] Specifically, the prediction model describes the relationship between the future state and the current state of the wheeled robot, as well as the current control variables (i.e., the future state can be obtained based on the current state variables and the current control variables). The current state information contains the current state variables of the wheeled robot, and the reference trajectory information contains the expected future state variables of the wheeled robot (i.e., the reference state variables). Therefore, based on the state error between the current state variables and the reference state variables, the reference trajectory information, and the prediction model, N control variables that can correct the state variables corresponding to the next N sampling times can be solved, making the future state variables of the wheeled robot as close as possible to the reference state variables, thereby achieving the trajectory tracking requirement of the wheeled robot.

[0032] In this embodiment, based on the state information, reference trajectory information, and prediction model, N control increments that can correct the state variables of the wheeled robot at the next N sampling times are predicted. This helps the actual walking trajectory of the wheeled robot to be closer to the desired reference trajectory, thereby achieving high-precision trajectory tracking.

[0033] S4: Based on the control increment sequence, disturbance observations and reference trajectory information, the target control quantity for compensating the disturbance is obtained, where the target control quantity includes the independent control quantity corresponding to each wheel.

[0034] Specifically, the target control quantity obtained based on the control increment sequence, disturbance observations, and reference trajectory information is compatible with the disturbances encountered by the wheeled robot during actual walking, thus improving the robot's anti-interference capability. The disturbance observations include the independent disturbances experienced by each wheel, which can accurately compensate for the disturbances in the target control quantity, including the independent control quantities corresponding to each wheel.

[0035] In this embodiment, the target control quantity for compensating for disturbances is obtained based on the control increment sequence, disturbance observations, and reference trajectory information. This enables independent disturbance compensation for each wheel, and improves the anti-interference capability of each wheel in a targeted manner according to the actual disturbance situation experienced by each wheel, which is beneficial to further improve the trajectory tracking accuracy.

[0036] S5: Control the operation of the wheeled robot according to the target control variable.

[0037] Specifically, the target control quantity includes the independent control quantity of each wheel, and controlling the operation of the wheeled robot according to the target control quantity includes controlling the operation of each wheel according to the independent control quantity of each wheel.

[0038] In this embodiment, each wheel of the wheeled robot is independently controlled according to the target control quantity, which helps to improve the flexibility of the wheeled robot. Compared with robots that include passive wheels, the wheeled robot provided in this embodiment has a stronger ability to adapt to road conditions and overcome obstacles, which helps to avoid accidents such as rollover and slippage, and ensures the stability and safety of the wheeled robot's trajectory tracking process.

[0039] S6: When the wheeled robot meets the preset triggering conditions at the current sampling time, it returns to the steps of obtaining the control increment sequence based on the state information, reference trajectory information and the pre-built prediction model, until it reaches the endpoint position. The triggering conditions include the disturbance observation value being greater than the preset disturbance threshold.

[0040] Specifically, when the wheeled robot meets the preset triggering conditions at the current sampling time, it means that it is necessary to re-predict the control increment sequence based on the state information, reference trajectory information and prediction model corresponding to the current sampling time, in order to ensure the accuracy of trajectory tracking.

[0041] It should be understood that, unlike existing technologies that treat the wheels of a wheeled robot as a whole and perform rolling optimization at every sampling moment, this embodiment considers the independent disturbances experienced by each wheel to obtain the independent control quantity corresponding to each wheel. Furthermore, optimization is only performed once (i.e., the step of returning to obtain the control increment sequence) if the triggering condition is met at the current moment. This improves the wheeled robot's anti-interference capability and trajectory tracking accuracy while avoiding excessive computational waste. In addition, this embodiment uses disturbance observations containing independent disturbances of each wheel as the optimization trigger condition. The smaller the disturbance experienced by the wheeled robot, the more accurate the existing target control quantity can be in ensuring trajectory tracking. The trigger condition based on the independent disturbances of each wheel (i.e., disturbance observations) can more accurately and sensitively predict that the wheeled robot may produce a certain trajectory deviation. Thus, before a larger trajectory tracking error occurs, it can be optimized again by combining the independent disturbances corresponding to each wheel to ensure trajectory tracking accuracy. At the same time, since the optimized target control quantity compensates for the actual disturbances, even in scenarios with poor road conditions, it will not cause frequent fluctuations in the control system (i.e., frequent triggering), which greatly improves the wheeled robot's adaptability to complex terrain and road conditions.

[0042] In this embodiment, whether to perform an optimization is determined by whether the current sampling time meets the preset triggering conditions. When the triggering conditions are met, the target control quantity can be optimized before a certain trajectory deviation occurs, taking into account the independent disturbances experienced by each wheel. This ensures the accuracy of trajectory tracking while avoiding excessive computing power consumption, thus greatly improving operating efficiency.

[0043] In this embodiment, acquiring operational information such as environmental maps, location information, state information, and disturbance observations provides a reliable data foundation for the subsequent generation of target control quantities. Based on the location information and environmental map, trajectory planning is performed to obtain reference trajectory information, providing the desired reference trajectory for subsequent trajectory tracking control. The reference trajectory obtained through trajectory planning ensures the wheeled robot's safe and stable arrival at the destination position, and also provides a reference basis for the subsequent generation of target control quantities. Based on the state information, reference trajectory information, and a pre-built prediction model, the control increment sequence corresponding to the next N sampling times is predicted, making it easier for the wheeled robot's actual walking trajectory to be closer to the desired reference trajectory, thereby achieving high-precision trajectory tracking. Based on the control increment sequence, disturbance observations, and reference trajectory information... The system obtains the target control quantity to compensate for disturbances, enabling independent disturbance compensation for each wheel. Based on the actual disturbance experienced by each wheel, it specifically enhances the anti-interference capability of each wheel, further improving trajectory tracking accuracy. Then, based on the target control quantity, it independently controls each wheel of the wheeled robot, improving the robot's flexibility and helping to avoid accidents such as rollovers and slippage, ensuring the stability and safety of the wheeled robot's trajectory tracking process. When the wheeled robot meets the preset triggering conditions at the current sampling moment, combined with the independent disturbances experienced by each wheel, it can optimize the target control quantity before a certain trajectory deviation occurs, until the endpoint is reached. This significantly improves the wheeled robot's adaptability to complex terrain and road conditions while enhancing trajectory tracking accuracy and operational efficiency.

[0044] Optionally, the disturbance observations include instantaneous disturbance observations and cumulative disturbance observations, and the disturbance thresholds include instantaneous disturbance thresholds and cumulative disturbance thresholds; when the wheeled robot meets the preset triggering condition at the current sampling time, it returns to the step of predicting the control increment sequence corresponding to the next N sampling times until it reaches the endpoint, including: The trigger condition is met when the instantaneous disturbance observation value is greater than the instantaneous disturbance threshold, or when the cumulative disturbance observation value is greater than the cumulative disturbance threshold. The trigger condition is met when the number of sampling times from the last sampling time that met the trigger condition to the current sampling time is equal to N. Returning to the steps of the control increment sequence based on state information, reference trajectory information, and a pre-built prediction model, until the endpoint is reached.

[0045] Specifically, the instantaneous disturbance observation value represents the disturbance situation corresponding to the current sampling node. When the instantaneous disturbance observation value is greater than the preset instantaneous disturbance threshold, it is possible to sensitively detect that the wheeled robot may deviate from the reference trajectory. The cumulative disturbance observation value represents the cumulative disturbance experienced by the wheeled robot up to the current sampling time, which can reflect the actual road conditions as a whole. It can start accumulating from the sampling time when the trigger condition was last met, or it can start accumulating from the starting position. Preferably, it is selected to start accumulating from the sampling time when the trigger condition was last met. When the cumulative disturbance observation value is greater than the preset cumulative disturbance threshold, it is possible to detect that the wheeled robot will deviate from the reference trajectory more accurately and stably. At the same time, after the trigger condition is met, the cumulative disturbance observation value can be cleared to zero to simplify calculations. Since N control increment sequences were predicted when the trigger condition was met last time, at the current sampling time, the state of the wheeled robot can only be corrected for the next N sampling times based on the current state and control increment sequences (i.e., the predicted step size is N). When the number of sampling times from the last sampling time when the trigger condition was met to the current sampling time is equal to N, it means that the predicted step size has been used up. It is necessary to re-optimize based on the current state information and reference trajectory information to obtain the target control increment after error compensation, so as to ensure the accuracy of trajectory tracking.

[0046] In this embodiment, instantaneous disturbance observations can sensitively detect when a wheeled robot will deviate from the reference trajectory. Accumulated disturbance observations can accurately and smoothly detect potential deviations under current road conditions or their own disturbances. Since the wheeled robot experiences disturbances before generating trajectory tracking errors, this embodiment, based on the triggering conditions of disturbance observations, can predict trajectory tracking errors in advance. An optimization is performed before a significant trajectory deviation occurs, and compensation is applied to the target control quantity for independent disturbances received by each wheel, ensuring trajectory tracking accuracy. Furthermore, when the predicted step size is exhausted, another optimization and disturbance compensation are performed to ensure the reliability of trajectory tracking.

[0047] Optionally, the triggering condition is met:

[0048] in, This indicates the sampling time when the trigger condition will be met next. Indicates the current sampling time. This indicates the last sampling time when the trigger condition was met. Indicates the instantaneous disturbance observation value, Indicates the instantaneous disturbance threshold. This indicates the number of sampling times elapsed from the last sampling time when the trigger condition was met to the current sampling time; or;

[0049] in, This represents the cumulative disturbance observation value from the last sampling time when the trigger condition was met to the current sampling time. This represents the cumulative disturbance threshold.

[0050] Specifically, instantaneous disturbance observations can be acquired based on a disturbance observer. For example, at each sampling time, the disturbance observer acquires the independent velocity and angular velocity disturbances corresponding to each wheel. The cumulative disturbance observation represents the cumulative value of all instantaneous disturbance observations from the last sampling time that met the triggering condition to the current sampling time.

[0051] It should be understood that when a wheeled robot makes a prediction to obtain the target control quantity at the starting position, it can be considered as satisfying the initial triggering condition. When the wheeled robot starts from the starting position and the triggering condition is satisfied for the first time, the initial sampling time corresponding to the starting position can be used as the sampling time of the previous satisfaction of the triggering condition. The initial triggering condition can be expressed as:

[0052] in, The initial sampling time corresponding to the starting position can be denoted as the initial trigger time. Indicates the current sampling time.

[0053] In this embodiment, the constraints of instantaneous disturbance observations and the predicted step size (i.e., the constraints of the next N sampling times) constitute a complete set of triggering conditions, which can sensitively detect that the wheeled robot may deviate from the reference trajectory. The constraints of cumulative disturbance observations and the predicted step size also constitute a complete set of triggering conditions, which can accurately and smoothly detect that the wheeled robot will deviate from the reference trajectory to some extent. This can be selected according to needs. For example, when the disturbance observation value is consistently large, indicating consistently poor road conditions, using the cumulative disturbance observation value as the triggering condition can improve the stability of the wheeled robot's operation, thereby improving its environmental adaptability. When the disturbance observation value is consistently small, selecting the instantaneous disturbance observation value as the triggering condition can improve the wheeled robot's sensitivity and ensure the accuracy of its trajectory tracking.

[0054] Optionally, such as Figure 2 As shown, the wheeled robot contains four wheels and, before obtaining the control increment sequence based on state information, reference trajectory information, and a pre-built prediction model, also includes: Construct a kinematic model of the wheeled robot, wherein the kinematic model satisfies: ; Among them, the state variables of the wheeled robot , Indicates the posture of the wheeled robot. , ( ) indicates the first The coordinates of each wheel Indicates the first The turning angle of each wheel, Indicates the first The speed of each wheel and Representing trigonometric functions and ; Based on the kinematic model, the reference state equation of the wheeled robot is obtained, which satisfies: ; in, This represents the reference state variable of the wheeled robot. Represents the reference control variables for the wheeled robot; Linearizing the reference state equation yields the first linear error model, which satisfies the following: ; in, , Represents the dynamic error of the state variable. This represents the dynamic error of the control variable. for about Jacobian matrix, for about The Jacobian matrix; Discretizing the first linear error model yields a discrete error model, which satisfies the following: ; in, This represents the error of the state variable of the wheeled robot at the next sampling time. This represents the error of the state variable of the wheeled robot at the current sampling moment. and It is a matrix relating the speed and turning angle of each wheel.

[0055] Specifically, in this embodiment, the wheeled robot includes four wheels. Taking a four-wheel drive wheeled robot as an example, a global coordinate system is constructed. Robot coordinate system and the independent coordinate system of each wheel. , The velocity represents the coordinates of the wheeled robot, and the positions of the four wheels in the robot's coordinate system are respectively... , , , ,in, and Let represent the distances from the wheels to the center of the wheeled robot. Then, the kinematic model of the four-wheeled, four-wheeled robot satisfies:

[0056] in, The state variables of a wheeled robot are denoted as . , Represents the wheeled robot in the global coordinate system The posture of being down , Indicates the first The turning angle of each wheel, Indicates the first The speed of each wheel and Representing trigonometric functions and .

[0057] From the above kinematic model, we can see that the state variables of the wheeled robot are... Control variables In the trajectory planning process, given that the velocities at the starting and ending positions are 0, the reference state variables of the wheeled robot at each node on the reference trajectory can be obtained based on the curvature of the reference trajectory and the kinematic and dynamic constraints of the robot. These are denoted as... According to the reference state variable By combining the kinematic equations, the reference control input (i.e., the reference control variable) of the wheeled robot can be obtained. In the actual motion control of wheeled robots, various disturbances occur. It is necessary to adjust the target control variable of the wheeled robot based on the error between the actual state variable and the reference state variable, expressing the control variable (i.e., the control increment) in incremental form. Then, at the current sampling time k, the target control quantity can be expressed as:

[0058] Reference state variables of wheeled robots It can be obtained from the reference trajectory, Substituting these values ​​into the above formula yields the reference control variable. Then the reference state equation of the four-wheeled, four-wheel drive robot can be expressed as:

[0059] The reference state equations are linearized, and a Taylor expansion is performed at any point, retaining first-order terms and ignoring higher-order terms:

[0060] Subtracting the above equation from the reference state equation yields the first linear error model:

[0061] in, , This indicates the dynamic speed error of a wheeled robot. The dynamic error representing the state variables of a wheeled robot. This represents the dynamic error of the control variable. for about Jacobian matrix, for about The Jacobian matrix.

[0062] The first linear error model described above is discretized, with a sampling period of [period missing]. The discrete error model can be expressed as:

[0063] in, This represents the error of the state variable of the wheeled robot at the next sampling time. This represents the error of the state variable of the wheeled robot at the current sampling moment. , , It is the identity matrix. , , , , , , , , .

[0064] As can be seen from the discrete error model above, the state quantity at the next sampling moment can be obtained based on the current state quantity and the current control quantity.

[0065] In this embodiment, compared to the kinematic model in the prior art that treats each wheel as a whole, the kinematic model established in this embodiment, which includes the corresponding speed and rotation angle of each wheel, can accurately and independently describe the state of each wheel of the wheeled robot. The discrete error model obtained based on this kinematic model can provide a reliable model basis for the precise control and trajectory tracking of the wheeled robot.

[0066] Optionally, before obtaining the control increment sequence based on state information, reference trajectory information, and a pre-built prediction model, the following steps are also included: A prediction model is obtained based on a discrete error model, which is used to predict the future state variables of the wheeled robot based on the current state variables and control variables. Based on the prediction model, a cost function is constructed. Construct the constraints of the cost function, wherein the constraints include the target control quantity being greater than or equal to the preset minimum control quantity and less than or equal to the preset maximum control quantity.

[0067] Specifically, after obtaining the discrete error model, in order to predict the optimal control increment sequence for the next N sampling times, a new state vector is constructed, and the state variables and control variables of the wheeled robot are represented as follows:

[0068] Therefore, the above discrete error model can be transformed into:

[0069] The output equation of a wheeled robot can be expressed as:

[0070] in, , , .

[0071] Then, in the prediction time domain, the relationship between the predicted state variables and the incremental form of the control output (i.e., the control increment) can be expressed as:

[0072]

[0073] Therefore, based on the current state variables and control variables, the future state variables of the wheeled robot can be predicted. The output matrix of the wheeled robot (i.e., the prediction model) can be represented as:

[0074] in, , , , , In the above In the matrix, the first row contains one item, the second row contains two items, and so on.

[0075] In one embodiment, in order to obtain the optimal control increment sequence in the prediction time domain, a cost function needs to be designed based on the prediction model:

[0076] in, and These represent the weights of the state variables and the control variables, respectively.

[0077] In the trajectory tracking control of a wheeled robot, it is necessary to ensure its ability to track the reference trajectory, and at the same time, to ensure motion stability and reduce the impact of rapid changes in control increments on the wheeled robot, the above cost function can be transformed into:

[0078] in, As a relaxation factor, This represents the difference between the current state variable and the reference state variable, reflecting the wheeled robot's ability to track the reference trajectory. The control variable, expressed in incremental form, reflects the requirement for stable changes in the control increment.

[0079] In a preferred embodiment, to satisfy the kinematic and dynamic constraints of the wheeled robot itself, the following constraints are constructed for the cost function:

[0080] in, , express The target control quantity corresponding to the sampling time. This represents the minimum value of the target control quantity (i.e., the minimum control quantity). This represents the maximum value of the target control quantity (i.e., the maximum control quantity). The minimum and maximum control quantities can be obtained based on the dynamic constraints of the wheeled robot. For example, the maximum control quantity cannot exceed the control quantity corresponding to the maximum output power of the wheeled robot.

[0081] Transform the above equation into a constraint that controls the increment:

[0082] in, , .

[0083] Then, in the prediction time domain, the constraint of the target control variable can be expressed as:

[0084] In this embodiment, a prediction model for the wheeled robot is obtained based on a discrete error model, which achieves the goal of predicting future state variables based on current state variables and control variables. A cost function is designed based on the prediction model, and constraints are constructed considering the kinematic and dynamic constraints of the wheeled robot itself. This ensures the stability and reliability of the wheeled robot's motion while satisfying its trajectory tracking capability.

[0085] Optionally, before obtaining the control increment sequence based on state information, reference trajectory information, and a pre-built prediction model, the following steps are also included: Based on the first linear error model and the independent disturbances experienced by each wheel, a second linear error model is obtained, which satisfies the following: ; in, This represents the velocity and angular velocity disturbances experienced by each wheel; The second linear error model is transformed into the following form: ; A perturbation observer is designed based on the second linear error model, and the perturbation observer satisfies: ; in, Indicates the perturbation observation value, This represents the nonlinear gain of the perturbation observer. Indicates intermediate variables. For about Nonlinear functions.

[0086] Specifically, since the wheeled robot is subject to external disturbances during its movement, a second linear error model with disturbances is obtained by considering the actual disturbances experienced by the wheeled robot based on the first linear error model:

[0087] in, , representing the velocity and angular velocity perturbations corresponding to each wheel of the wheeled robot. The above equation can be transformed into the following form:

[0088] The estimated value of the disturbance experienced by the wheeled robot can be expressed as:

[0089] in, This represents the estimated value of the disturbance (i.e., the observed value of the disturbance). Let represent the nonlinear gain of the perturbation observer. Then, the error between the actual perturbation experienced by the wheeled robot and the estimated value of the perturbation can be expressed as:

[0090] In the actual operation of wheeled robots If the time derivative is zero, then according to the above formula, for Differentiation yields:

[0091] As can be seen from the above equation, selecting an appropriate observation gain function can ensure the asymptotic stability of the system, enabling the designed disturbance observer to effectively observe disturbances experienced by the wheeled robot. Since the disturbance observer needs to... The differential state is measured, and an intermediate variable is introduced to improve the above disturbance observer. The improved disturbance observer design is expressed as follows:

[0092] in, This indicates the introduced intermediate variable. It is about nonlinear functions, .

[0093] After introducing the above intermediate variables The formula can be converted to:

[0094] As can be seen from the above formula, by selecting the appropriate This allows the disturbance observer to reach an asymptotically stable state, with the error between the disturbance estimate and the actual disturbance approaching zero. In other words, the disturbance observation value can better reflect the actual disturbance experienced by the wheeled robot, thus improving the observation accuracy of the disturbance observer.

[0095] In this embodiment, a disturbance observer is constructed to observe the velocity and angular velocity disturbances of each wheel of the wheeled robot. By introducing intermediate variables, the disturbance observer reaches an asymptotically stable state, ensuring the accuracy and reliability of the disturbance observation values. This, in turn, ensures that the target control quantity after the disturbance is compensated can control the wheeled robot to achieve high-precision trajectory tracking.

[0096] Optionally, the state information includes the current state of the wheeled robot at the current sampling time; the reference trajectory information includes reference state quantities corresponding to multiple nodes; based on the control increment sequence, disturbance observations, and reference trajectory information, the target control quantity for compensating for the disturbance is obtained, including: Based on the reference state quantity and the kinematic model, the reference control quantity corresponding to the current sampling time is obtained; The target control quantity is obtained based on the control increment sequence, reference control quantity, and disturbance observations. The control increment sequence is obtained by optimizing the cost function based on the current state quantity, reference state quantity, and constraints. At the starting position and the sampling time satisfying the trigger condition, the target control increment satisfies: ; in, This represents the target control quantity corresponding to the current sampling time. This represents the reference control quantity corresponding to the current sampling time. This represents the control increment corresponding to the current sampling time. This represents the perturbation observation value corresponding to the current sampling time.

[0097] At sampling times other than those meeting the triggering condition, the target control increment satisfies:

[0098] in, Indicates in The target control quantity corresponding to the sampling time. Indicates in The control increment corresponding to the sampling time.

[0099] Specifically, by inputting the reference state variables into the aforementioned kinematic model, the corresponding reference control variables can be obtained. For example, the reference state variable corresponding to the node closest to the current sampling time interval can be selected and input into the kinematic model, or the reference state variable corresponding to the future node closest to the current sampling time interval can be selected and input into the kinematic model to obtain the reference control variables corresponding to the current sampling time. Then, the current state variable and reference state variable of the wheeled robot at the current sampling time are input into the aforementioned cost function, and the function is optimized according to the constraints to obtain a control increment sequence consisting of control increments corresponding to the next N sampling times. Controlling the incremental sequence Represented as:

[0100] At the starting position and the sampling time when the trigger condition is met, the target control quantity is obtained based on the first control increment in the control increment sequence, the reference control quantity, and the disturbance observation value:

[0101] in, This represents the target control quantity corresponding to the current sampling time, which is the control variable corresponding to the aforementioned kinematic model. It can be seen that, It contains the independent speed control and independent rotation angle control quantities corresponding to each wheel of the wheeled robot (i.e., the independent control quantities corresponding to each wheel). This represents the reference control quantity corresponding to the current sampling time. This represents the control increment corresponding to the current sampling time (i.e., the first control increment in the above control increment sequence). This represents the perturbation observation value corresponding to the current sampling time.

[0102] At other sampling times besides those where the triggering condition is met, the target control quantity satisfies:

[0103] in, Indicates in The target control quantity corresponding to the sampling time. Indicates in The control increment corresponding to the sampling time, for example, at the sampling time when the trigger condition was last met. The first sampling time experienced after that, i.e. At that time, the control increment contained in the target control quantity is the control increment sequence. .

[0104] As can be seen from the above formula, unlike the existing technology which performs rolling optimization at each sampling time, this embodiment only performs optimization again after the trigger condition is met. During the sampling time between the previous sampling time when the trigger condition was met and the next sampling time when the trigger condition is met, no optimization is performed. The control increment in the control increment sequence obtained at the previous sampling time when the trigger condition was met is still used, which avoids wasting computing power, reduces the number of updates of the model predictive controller, and greatly improves the operating efficiency of the wheeled robot.

[0105] In this embodiment, the target control quantity includes the reference control quantity, the control increment obtained from the solution optimization, and the disturbance observation value. This not only enables feedback adjustment of the wheeled robot's output based on state errors but also accurately compensates for the independent disturbances experienced by each wheel, facilitating high-precision trajectory tracking of the wheeled robot. Furthermore, optimization is performed only when the trigger condition is met. Compared to the existing method of performing rolling optimization at every sampling moment, this embodiment ensures trajectory tracking accuracy while avoiding wasted computational power, achieving a balance between high precision and high efficiency in the trajectory tracking process.

[0106] Optionally, such as Figure 3 As shown, trajectory planning is performed based on location information and an environmental map to obtain reference trajectory information, including: Creating environment maps based on SLAM; Based on heuristic algorithms and environmental maps, the optimal path between the starting point and the ending point is obtained; The optimal path is smoothed to obtain a reference trajectory; The reference trajectory is divided into multiple nodes that are evenly distributed on the reference trajectory. Based on the curvature and preset acceleration constraints at the nodes, the reference state quantities corresponding to the nodes are planned to obtain the reference trajectory information.

[0107] Specifically, operational information can also include environmental information. Wheeled robots can construct environmental maps using environmental and state information (such as the current posture of the wheeled robot) collected by sensors. For example, the steps for constructing an environmental map based on SLAM include: data processing: filtering, denoising, and feature extraction of environmental data collected by sensors (such as camera images, radar data, etc.) to provide effective information for subsequent environmental map construction; front-end matching: visual SLAM mapping can be performed based on the information collected by the camera to obtain a 3D local environmental map, and then laser SLAM mapping can be performed based on the information collected by LiDAR to obtain a laser local environmental map. Then, the 3D local environmental map and the laser local environmental map are fused to obtain a global grid map (i.e., the environmental map). Backend optimization: After obtaining the environmental map, filtering and nonlinear optimization methods can be used to optimize the current posture and environmental map matched by the front end, reducing the cumulative error of each stage. Preferably, after constructing the environmental map, loop closure detection can be performed based on the landmark prior information method, the bag-of-words model method, and an improved algorithm based on the bag-of-words model and probability. For example, the current posture of the wheeled robot and the current data of the sensors can be matched with historical data to determine whether the wheeled robot has passed the current position, reducing the cumulative error and constructing a globally consistent trajectory and map.

[0108] In one embodiment, based on the starting point, ending point, and environmental map, a heuristic algorithm is used to search for the optimal path between the starting point and ending point. The obtained optimal path is then smoothed, for example, using spline curve interpolation to reduce turning redundancy and improve path quality. The reference trajectory is then divided into multiple evenly distributed nodes. Calculate each node based on the reference trajectory Corresponding curvature :

[0109] Where y represents a node The ordinate of the coordinate can be used to obtain the radius of curvature for each node according to the above formula. .

[0110] To ensure the smooth operation of a wheeled robot without impact, its continuous speed and acceleration must be guaranteed. Therefore, design constraints to limit the acceleration increment:

[0111] in, Represents a node Tangential acceleration at the point, Represents a node Tangential acceleration at the point, This indicates the maximum increment of tangential acceleration. This indicates the distance between adjacent nodes.

[0112] In one embodiment, since the wheeled robot's initial velocity at the starting position is known to be zero, and its final velocity at the ending position is also known to be zero, the radius of curvature of the nodes on the reference trajectory is... Given the tangential acceleration constraints, the following equation can be used for iteration to obtain the position of the wheeled robot at each node on the reference trajectory. The corresponding time and speed information are used to obtain reference trajectory information.

[0113]

[0114]

[0115]

[0116] in, node radial acceleration at that point, Represents a node The square of the velocity, Represents the radius of curvature of the previous node. This represents the maximum value of the tangential acceleration. Indicates adjacent nodes and The Euclidean distance between them.

[0117] In this embodiment, creating an environment map based on SLAM is beneficial for wheeled robots to construct environmental maps when performing exploration tasks in unknown environments (such as extraterrestrial exploration), providing a reliable foundation for trajectory planning. Based on heuristic algorithms and the environment map, the optimal path between the starting and ending positions is obtained and smoothed. The resulting reference trajectory helps the wheeled robot reach the ending position smoothly with minimal cost. Based on the curvature corresponding to the nodes on the reference trajectory and preset acceleration constraints, the reference state variables corresponding to the nodes are planned, providing an effective reference data foundation for subsequent trajectory tracking of the wheeled robot.

[0118] Another embodiment of the present invention provides a predictive control device for a wheeled robot, comprising: The acquisition module is used to acquire the operation information of the wheeled robot. The operation information includes an environment map, location information, status information and disturbance observations. The location information includes the starting position and the ending position, and the disturbance observations include the independent disturbances corresponding to each wheel of the wheeled robot. The planning module is used to plan trajectories based on location information and environmental maps to obtain reference trajectory information, which includes a reference trajectory with multiple nodes. The prediction module is used to obtain the control increment sequence based on the state information, reference trajectory information and the pre-built prediction model. The control increment sequence includes N control increments for correcting the state variables of the wheeled robot at the next N sampling times. The compensation module is used to obtain the target control quantity for compensating the disturbance based on the control increment sequence, disturbance observations and reference trajectory information. The target control quantity includes the independent control quantity corresponding to each wheel. The control module is used to control the operation of the wheeled robot according to the target control quantity; The triggering module is used to return the operation of the control increment sequence based on the state information, reference trajectory information and pre-built prediction model when the wheeled robot meets the preset triggering conditions at the current sampling time, until the endpoint position is reached. The triggering conditions include the disturbance observation value being greater than the preset disturbance threshold.

[0119] In this embodiment, the acquisition module acquires operational information such as environmental maps, location information, state information, and disturbance observations to provide a reliable data foundation for the subsequent generation of target control quantities. The planning module performs trajectory planning based on the location information and environmental map to obtain reference trajectory information, providing the desired reference trajectory for subsequent trajectory tracking control. The reference trajectory obtained through trajectory planning can ensure the wheeled robot's safe and stable arrival at the destination position, and also provides a reference basis for the subsequent generation of target control quantities. The prediction module predicts the control increment sequence corresponding to the next N sampling times based on the state information, reference trajectory information, and a pre-built prediction model, making it easier for the wheeled robot's actual walking trajectory to be closer to the desired reference trajectory, thereby achieving high-precision trajectory tracking. The compensation module, based on the control increment sequence, disturbance observations, and reference... The trajectory information is used to obtain the target control quantity for compensating for disturbances, enabling independent disturbance compensation for each wheel. Based on the actual disturbance experienced by each wheel, the anti-interference capability of each wheel is specifically improved, further enhancing trajectory tracking accuracy. The control module then independently controls each wheel of the wheeled robot according to the target control quantity, improving the robot's flexibility and helping to avoid accidents such as tipping over and slipping, ensuring the stability and safety of the wheeled robot's trajectory tracking process. When the wheeled robot meets the preset triggering conditions at the current sampling moment, the triggering module, combined with the independent disturbances experienced by each wheel, can optimize the target control quantity before a certain trajectory deviation occurs, until the endpoint is reached. This not only improves trajectory tracking accuracy and operating efficiency but also greatly enhances the wheeled robot's adaptability to complex terrain and road conditions.

[0120] Another embodiment of the present invention provides a wheeled robot, including a computer-readable storage medium storing a computer program and a processor. The computer program is read and executed by the processor to implement the predictive control method of the wheeled robot as described above.

[0121] The technical effects achievable in this embodiment are basically the same as those achievable by the aforementioned predictive control method for wheeled robots, and will not be repeated here.

[0122] The present invention will now describe electronic devices that can serve as servers or clients of the present invention, which are examples of hardware devices that can be applied to various aspects of the present invention. Electronic devices are intended to represent various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. Electronic devices can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0123] Electronic devices include a computing unit that can perform various appropriate actions and processes based on a computer program stored in read-only memory (ROM) or loaded from a storage unit into random access memory (RAM). The RAM can also store various programs and data required for device operation. The computing unit, ROM, and RAM are interconnected via a bus. Input / output (I / O) interfaces are also connected to the bus.

[0124] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other.

[0125] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. In this application, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention according to actual needs. Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units can be implemented in hardware or as software functional units.

[0126] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.

Claims

1. A predictive control method for a wheeled robot, characterized in that, Includes the following steps: The operation information of the wheeled robot is obtained, wherein the operation information includes an environmental map, location information, status information and disturbance observations, the location information includes a starting position and an ending position, and the disturbance observations include independent disturbances corresponding to each wheel of the wheeled robot; Based on the location information and the environment map, trajectory planning is performed to obtain reference trajectory information, wherein the reference trajectory information includes a reference trajectory with multiple nodes; Based on the state information, the reference trajectory information, and the pre-built prediction model, a control increment sequence is obtained, wherein the control increment sequence includes N control increments for correcting the state variables of the wheeled robot at N sampling times in the future. Based on the control increment sequence, the disturbance observations, and the reference trajectory information, a target control quantity for compensating for the disturbance is obtained, wherein the target control quantity includes independent control quantities corresponding to each of the wheels; The operation of the wheeled robot is controlled according to the target control quantity; When the wheeled robot meets the preset triggering condition at the current sampling time, it returns to the step of obtaining the control increment sequence based on the state information, the reference trajectory information and the pre-built prediction model, until it reaches the endpoint position. The triggering condition includes the perturbation observation value being greater than the preset perturbation threshold. The state information includes the current state quantity of the wheeled robot at the current sampling time; the reference trajectory information includes reference state quantities corresponding to multiple nodes; obtaining the target control quantity to compensate for the disturbance based on the control increment sequence, the disturbance observation value, and the reference trajectory information includes: obtaining the reference control quantity corresponding to the current sampling time based on the reference state quantity and the kinematic model of the wheeled robot; and obtaining the target control quantity according to the control increment sequence, the reference control quantity, and the disturbance observation value. At the starting position and at the sampling time that satisfies the triggering condition, the target control quantity satisfies: ； in, This represents the target control quantity corresponding to the current sampling time. This represents the reference control quantity corresponding to the current sampling time. This represents the control increment corresponding to the current sampling time. This represents the disturbance observation value corresponding to the current sampling time; Before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, the method further includes: constructing a kinematic model of the wheeled robot; and obtaining a reference state equation of the wheeled robot based on the kinematic model, wherein the reference state equation satisfies: ； in, This represents the reference state variable of the wheeled robot. This represents the reference control variable of the wheeled robot; The reference state equation is linearized to obtain a first linear error model; the first linear error model satisfies: ； in, , This represents the dynamic error of the wheeled robot's state variables. This represents the dynamic error of the control variable, which is obtained based on the state variable and the kinematic model. for about Jacobian matrix, for about Jacobian matrix; Based on the first linear error model and the independent disturbances experienced by each wheel, a second linear error model is obtained, which satisfies the following: ； in, This represents the velocity and angular velocity disturbances experienced by each of the aforementioned wheels; The second linear error model is transformed into the following form: ； A disturbance observer is designed based on the second linear error model, and the disturbance observer satisfies: ； in, This represents the observed disturbance value. This represents the nonlinear gain of the disturbance observer. Indicates intermediate variables. For about Nonlinear functions.

2. The predictive control method for a wheeled robot according to claim 1, characterized in that, The disturbance observations include instantaneous disturbance observations and cumulative disturbance observations, and the disturbance thresholds include instantaneous disturbance thresholds and cumulative disturbance thresholds; the step of returning to the step of obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model when the wheeled robot meets the preset triggering condition at the current sampling time, until reaching the endpoint position, includes: The triggering condition is met when the instantaneous disturbance observation value is greater than the instantaneous disturbance threshold, or when the cumulative disturbance observation value is greater than the cumulative disturbance threshold. The trigger condition is met when the number of sampling times from the last sampling time that met the trigger condition to the current sampling time is equal to N. Return to the step of obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, until the endpoint position is reached.

3. The predictive control method for a wheeled robot according to claim 2, characterized in that, The triggering condition is satisfied: ； in, This indicates the sampling time at which the triggering condition will be met next. Indicates the current sampling time. This indicates the sampling time at which the triggering condition was last met. This represents the instantaneous disturbance observation value. This represents the instantaneous disturbance threshold. This indicates the number of sampling times elapsed from the last sampling time when the trigger condition was met to the current sampling time; or; ； in, This represents the cumulative disturbance observation value from the last sampling time when the trigger condition was met to the current sampling time. This represents the cumulative disturbance threshold.

4. The predictive control method for a wheeled robot according to claim 1, characterized in that, The wheeled robot includes four wheels and, before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, further includes: Construct a kinematic model of the wheeled robot, wherein the kinematic model satisfies: ； Wherein, the state variable , This indicates the posture of the wheeled robot. , Indicates the first The coordinates of the aforementioned wheels, Indicates the first The turning angle of the aforementioned wheel, Indicates the first The speed of the aforementioned wheels and Representing trigonometric functions and ; The first linear error model is discretized to obtain a discrete error model, wherein the discrete error model satisfies: ； in, This represents the error of the state variable of the wheeled robot at the next sampling time. This represents the error of the state variable of the wheeled robot at the current sampling time. and It is a matrix relating the speed and the turning angle of each of the aforementioned wheels.

5. The predictive control method for a wheeled robot according to claim 4, characterized in that, Before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, the method further includes: The prediction model is obtained based on the discrete error model, wherein the prediction model is used to predict the future state variables of the wheeled robot based on the current state variables and the control variables; Based on the prediction model, a cost function is constructed; The constraints of the cost function are constructed, wherein the constraints include the target control quantity being greater than or equal to a preset minimum control quantity and less than or equal to a preset maximum control quantity.

6. The predictive control method for a wheeled robot according to claim 5, characterized in that, The step of obtaining the target control quantity to compensate for the disturbance based on the control increment sequence, the disturbance observation, and the reference trajectory information includes: The control increment sequence is obtained by optimizing the cost function based on the current state quantity, the reference state quantity, and the constraint conditions. At sampling times other than those meeting the triggering condition, the target control increment satisfies: in, Indicates in The target control quantity corresponding to the sampling time. Indicates in The control increment corresponding to the sampling time.

7. The predictive control method for a wheeled robot according to any one of claims 1-6, characterized in that, The process of trajectory planning based on the location information and the environmental map to obtain reference trajectory information includes: Create the environment map based on SLAM; Based on the heuristic algorithm and the environment map, the optimal path between the starting point and the ending point is obtained; The optimal path is smoothed to obtain the reference trajectory; The reference trajectory is divided to obtain a plurality of nodes evenly distributed on the reference trajectory; Based on the curvature and preset acceleration constraints at the node, the reference state variables corresponding to the node are planned to obtain the reference trajectory information.

8. A predictive control device for a wheeled robot, characterized in that, include: The acquisition module is used to acquire the operation information of the wheeled robot, wherein the operation information includes an environmental map, location information, status information and disturbance observations, the location information includes a starting position and an ending position, and the disturbance observations include independent disturbances corresponding to each wheel of the wheeled robot; A planning module is used to perform trajectory planning based on the location information and the environmental map to obtain reference trajectory information, wherein the reference trajectory information includes a reference trajectory with multiple nodes; The prediction module is used to obtain a control increment sequence based on the state information, the reference trajectory information and the pre-built prediction model, wherein the control increment sequence includes N control increments for correcting the state variables of the wheeled robot at N sampling times in the future. The compensation module is used to obtain a target control quantity for compensating the disturbance based on the control increment sequence, the disturbance observation value and the reference trajectory information, wherein the target control quantity includes the independent control quantity corresponding to each of the wheels; A control module is used to control the operation of the wheeled robot according to the target control quantity; A triggering module is used to return the operation of obtaining the control increment sequence based on the state information, the reference trajectory information and the pre-built prediction model when the wheeled robot meets the preset triggering conditions at the current sampling time, until the endpoint position is reached. The triggering conditions include the disturbance observation value being greater than the preset disturbance threshold. The state information includes the current state quantity of the wheeled robot at the current sampling time; the reference trajectory information includes reference state quantities corresponding to multiple nodes; obtaining the target control quantity to compensate for the disturbance based on the control increment sequence, the disturbance observation value, and the reference trajectory information includes: obtaining the reference control quantity corresponding to the current sampling time based on the reference state quantity and the kinematic model of the wheeled robot; and obtaining the target control quantity according to the control increment sequence, the reference control quantity, and the disturbance observation value. At the starting position and at the sampling time that satisfies the triggering condition, the target control quantity satisfies: ； in, This represents the target control quantity corresponding to the current sampling time. This represents the reference control quantity corresponding to the current sampling time. This represents the control increment corresponding to the current sampling time. This represents the disturbance observation value corresponding to the current sampling time; Before obtaining the control increment sequence based on the state information, the reference trajectory information, and the pre-built prediction model, the method further includes: constructing a kinematic model of the wheeled robot; and obtaining a reference state equation of the wheeled robot based on the kinematic model, wherein the reference state equation satisfies: ； in, This represents the reference state variable of the wheeled robot. This represents the reference control variable of the wheeled robot; The reference state equation is linearized to obtain a first linear error model; the first linear error model satisfies: ； in, , This represents the dynamic error of the wheeled robot's state variables. This represents the dynamic error of the control variable, which is obtained based on the state variable and the kinematic model. for about Jacobian matrix, for about Jacobian matrix; Based on the first linear error model and the independent disturbances experienced by each wheel, a second linear error model is obtained, which satisfies the following: ； in, This represents the velocity and angular velocity disturbances experienced by each of the aforementioned wheels; The second linear error model is transformed into the following form: ； A disturbance observer is designed based on the second linear error model, and the disturbance observer satisfies: ； in, This represents the observed disturbance value. This represents the nonlinear gain of the disturbance observer. Indicates intermediate variables. For about Nonlinear functions.

9. A wheeled robot, characterized in that, The method includes a computer-readable storage medium storing a computer program and a processor, the computer program being read and executed by the processor to implement the predictive control method for a wheeled robot as described in any one of claims 1 to 7.

Images