Robot distributed cooperative control method, system, equipment and medium

Abstract

The invention provides a robot distributed cooperative control method, system and device and a medium, and the method comprises the steps: taking each wheel of a robot as an independent intelligent agent, and building a distributed model prediction controller of each intelligent agent with the trajectory tracking and collision avoidance of the intelligent agents and the motion matching between the intelligent agents as control targets; the characteristics of independent driving and independent steering of the robot can be brought into full play, so that the robot can move in environments such as narrow environments or rugged terrains; in addition, aiming at the influence of possible unknown disturbances such as slipping and different friction factors of each wheel, estimation is carried out based on an extended state observer, and the motion control quantity of the robot is compensated so as to ensure the normal motion process. According to the scheme of the invention, the control precision of the robot can be improved, and the anti-interference performance can be improved especially in the face of complex system dynamics and uncertainty.

Description

Technical Field

[0001] The present invention belongs to the technical field related to robot control, and in particular relates to a robot distributed collaborative control method, system, equipment and medium. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] The six-wheel independently driven and independently steered robot is a cutting-edge mobile robot equipped with six sets of independently controlled wheels arranged side by side, allowing each wheel to be manipulated independently, thereby achieving fine motion control and steering. The independent drive and independent steering design gives the robot high maneuverability in changing terrains, allowing the robot to travel stably in narrow spaces and complex terrains. The notable features of the six-wheel independently driven and independently steered robot include: 1. Independent operation: The independent control capability of the wheels enables the robot to move while maintaining its direction, or rotate on the spot; 2. Terrain adaptability: The robot can travel on a variety of terrain conditions, including sand, mud, snow, and even remain stable on slopes; 3. Obstacle avoidance performance: Equipped with advanced sensors such as lidar, cameras, and ultrasonic sensors, the robot can perceive the surrounding environment and achieve effective obstacle avoidance and path planning.

[0004] During actual missions, existing technologies for six-wheel independently driven and independently steered robots are insufficiently adaptable to complex terrains such as slopes and roads with uneven friction coefficients. The middle wheels are prone to spinning, slipping, or even rolling over, leading to energy loss and mechanical damage. Furthermore, existing control methods are complex in multi-wheel coordinated control, making it difficult to accurately adjust the driving force and speed of each wheel in real time. This results in insufficient trajectory tracking accuracy and poor lateral stability for the robot in complex terrain. Furthermore, existing technologies lack control accuracy and real-time performance when dealing with frequent and unknown external interference, making it difficult to meet mission requirements in complex environments. Therefore, solving the control problem of six-wheel independently driven and independently steered robots has great application value. Summary of the Invention

[0005] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a robot distributed collaborative control method, system, equipment and medium, which can improve the control accuracy of the robot and improve the anti-interference ability, especially when facing complex system dynamics and uncertainties.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: In a first aspect, the present invention provides a distributed collaborative control method for a robot, wherein the robot is a six-wheel independently driven and independently steered robot, and the distributed collaborative control method for the robot comprises: Each wheel of the robot is considered an independent intelligent agent. With the control objectives of trajectory tracking, collision avoidance, and motion matching between agents, a distributed model predictive controller and corresponding constraints are constructed for each agent. The optimal control input of each agent in the prediction time domain is obtained. The extended state observer constructed based on the discrete error model of the intelligent agent and the external unknown disturbance is called to estimate the unknown disturbance in real time and generate the compensation control quantity. Through iterative optimization, the current optimal control input is combined with the generated compensation control quantity to generate a composite control quantity to control the robot.

[0007] In a second aspect, the present invention provides a robot distributed collaborative control system, wherein the robot is a six-wheel independently driven and independently steered robot, and the robot distributed collaborative control system comprises: The model predictive control module is configured to treat each wheel of the robot as an independent agent, construct a distributed model predictive controller and corresponding constraints for each agent with the control objectives of trajectory tracking, collision avoidance, and motion matching between agents, and solve for the optimal control input for each agent within the prediction time domain; The control module is configured to: call an extended state observer constructed based on the discrete error model of the intelligent agent and external unknown disturbances, estimate the unknown disturbances in real time and generate compensating control quantities, and through iterative optimization, combine the current optimal control input with the generated compensating control quantities to generate a composite control quantity to control the robot.

[0008] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein the computer instructions, when executed by the processor, perform the method described in the first aspect.

[0009] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0010] One or more of the above technical solutions have the following beneficial effects: In this invention, each wheel of the robot is treated as an independent intelligent agent. A distributed model predictive controller is constructed for each agent, with trajectory tracking, collision avoidance, and inter-agent motion matching as control objectives. This fully utilizes the robot's independent drive and steering characteristics, enabling it to maneuver in confined environments or on rugged terrain. Furthermore, an extended state observer is used to estimate the effects of unknown disturbances, such as slippage and varying friction coefficients between wheels, and compensate for the robot's motion control variables to ensure proper motion. This solution, particularly when faced with complex system dynamics and uncertainties, can improve the robot's control accuracy and interference resistance.

[0011] Advantages of additional aspects of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0013] Figure 1(a) is a schematic diagram of centralized model predictive control; Figure 1(b) is a schematic diagram of distributed model predictive control; Figure 2 Schematic diagram of motion control under the distributed framework of intelligent agents in embodiment 1 of the present invention; Figure 3 This is a block diagram of distributed collaborative control in Example 1 of the present invention; Figure 4 Schematic diagram of intelligent body motion matching in different situations in embodiment 1 of the present invention; FIG5 (a) is a schematic diagram of the movement of the vehicle body correction robot based on distributed control in the first embodiment of the present invention; FIG5( b ) is a schematic diagram of the motion of a robot in a single-wheel fault state based on distributed control in the first embodiment of the present invention; Figure 6 This is a control block diagram of a distributed system based on an ESO feedforward controller in the first embodiment of the present invention. DETAILED DESCRIPTION

[0014] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0015] It should be noted that the terms used herein are for describing particular embodiments only and are not intended to limit the exemplary embodiments according to the present invention.

[0016] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0017] Example 1 The six-wheeled independently driven and independently steered robot features independent control of both drive and steering on each wheel. This feature allows the robot to maneuver in confined environments or on unusual terrain, free from the constraints of traditional steering models. This allows for the robot to combine six-wheel active steering with a short turning radius with specialized motion models such as pan and tilt for motion control.

[0018] The control of a six-wheeled independently driven and independently steered robot can be divided into centralized and distributed control. The biggest difference between centralized and distributed control lies in the different degrees of freedom used to control the robot. Although centralized control methods have a single-wheel position deviation constraint, the control objective is to treat the robot as a center-of-mass control model, with only two degrees of freedom: drive and steering. Distributed control targets a single agent, namely, single-wheel drive and steering, and controls 12 degrees of freedom (six drive and six steering). Therefore, distributed control significantly enhances the robot's flexibility. The optimization process of centralized control continuously corrects the deviation input feedback, correcting the deviation to zero at a certain point in the future time domain. The overall process is a gradual optimization process. Distributed control, on the other hand, provides more precise control of the robot. When the robot deviates from its current position, it can correct the deviation through tilting, translation, or single-agent steering movements. The error is corrected at the current moment, resulting in a direct deviation optimization process, as shown in Figures 1(a) and 1(b). This method is more suitable for narrow and special environments. Therefore, this embodiment studies distributed collaborative control of robots.

[0019] This embodiment discloses a distributed collaborative control method for a robot, wherein the robot is a six-wheel independently driven and independently steered robot. The distributed collaborative control method for the robot includes: Step 101: Treat each wheel of the robot as an independent agent. Build a distributed model predictive controller for each agent and its corresponding constraints with the control objectives of trajectory tracking, collision avoidance, and motion matching between agents. Obtain the optimal control input for each agent within the prediction time domain. Step 102: Call the extended state observer constructed based on the discrete error model of the intelligent agent and the external unknown disturbance to estimate the unknown disturbance in real time and generate the compensation control quantity. Through iterative optimization, the current optimal control input is combined with the generated compensation control quantity to generate a composite control quantity to control the robot.

[0020] This embodiment treats each wheel of the robot as an independent intelligent agent, constructing a distributed model predictive controller for each agent with trajectory tracking, collision avoidance, and inter-agent motion matching as control objectives. This fully utilizes the robot's independent drive and steering characteristics, enabling the robot to move in confined environments or on rugged terrain. Furthermore, the extended state observer estimates the effects of unknown disturbances, such as slippage and varying friction coefficients between wheels, and compensates for the robot's motion control variables to ensure normal motion. This solution, particularly when faced with complex system dynamics and uncertainties, can improve adaptability to complex terrain, enhance the robot's motion flexibility and real-time performance, and enhance its lateral stability and anti-interference capabilities.

[0021] Based on the distributed control method, this embodiment will regard the six-wheeled robot as a structure composed of multi-agent components connected by rigid connections. The collaborative control method after the multi-agent components are rigidly connected will be studied. The influence of disturbances such as ground slippage and unevenness will be considered. A feedforward controller based on the extended state observer will be designed to estimate and compensate for the disturbance. Finally, it is summarized as a distributed control problem with synchronous tracking and anti-interference characteristics to be solved, providing algorithmic theoretical support for simulation and prototype experiments.

[0022] Combined with the distributed collaborative strategy, the control method of multi-agent collaborative motion is studied: the reference input trajectory and the actual state trajectory are introduced to obtain the target optimization problem that can be calculated and processed in a distributed manner. By using the specific form of the system model and the distributed target optimization problem to formulate the objective function, each objective function is summarized as a performance function. By including the compatibility synchronization constraint and the hard connection constraint as coupling terms in the quadratic optimization problem of the performance function, the distributed model predictive control method is used to design a distributed controller for the agent. The differential matrix of the computational optimization problem and the iterative optimization of the constraint terms are used to converge the target problem to the optimal solution. Through the designed distributed controller of the agent, any state in the distributed optimization problem solution set satisfies the coordination constraints and obstacle avoidance constraints. Based on the above-mentioned distributed algorithm with multi-agent collaborative motion as the control target, the collaborative control method is verified in the application scenario of trajectory tracking and obstacle avoidance functions. The design structure of the distributed collaborative control scheme is as follows. Figure 2 shown.

[0023] Distributed collaborative control of six-wheeled robots involves the robots collaborating with each other and performing real-time coordinated actions during motion control tasks. This requires the design of appropriate collaborative control strategies to improve the efficiency and flexibility of overall task execution. Key challenges in designing distributed collaborative control strategies for robots include distributed decision-making coordination and system coordination constraints.

[0024] The following is a detailed description of the robot distributed collaborative control method proposed in this embodiment: Step 101: Treat each wheel of the robot as an independent intelligent agent, and construct a distributed model predictive controller and corresponding constraints for each intelligent agent with the control objectives of trajectory tracking, collision avoidance, and motion matching between intelligent agents, and solve to obtain the optimal control input of each intelligent agent in the prediction time domain.

[0025] The collaborative strategy of multiple intelligent agents is analyzed through the problem of system collaborative constraints, which mainly includes system geometric constraints, namely reference trajectory constraints, and trajectory parameter synchronization constraints, namely trajectory parameter coordination.

[0026] Combined with the relevant definitions of robot state quantities, the reference trajectory state of the robot's center of mass for , agent i The distributed reference trajectory state is: , and represents the position of the robot's center of mass at time k and the position of agent i, where the robot's left front wheel, right front wheel, left middle wheel, right middle wheel, left rear wheel, and right rear wheel are i The values are assigned as 1, 2, 3, 4, 5, and 6. The relative positions of the agents are restricted by the robot's connection frame. Therefore, each agent generates its own reference trajectory based on the robot's center of mass reference trajectory during movement. The following reference trajectories of the agents are derived through physical constraints.

[0027] Agent 1 and Agent 2 in discrete time domain k The reference trajectory state at time : (1) Agent 3 and Agent 4 in discrete time domain k The reference trajectory state at time : (2) The reference trajectory states of Agent 5 and Agent 6 at time k in the discrete time domain are: : (3) Among them, the reference heading angles of each agent are , represents the length of the robot, Indicates the width of the robot.

[0028] In addition to the geometric constraint problem in the multi-agent collaborative matching process, it is also necessary to consider the parameter coordination problem in the time domain. This embodiment adopts a synchronous control method to achieve parameter coordination of multi-agents. In the synchronous control method, each agent reference trajectory variable is obtained through The parameters are advanced in the time domain, where Indicates the index of the trajectory point.

[0029] Defining an Agent i The reference trajectory variables are: (4) The iterative update of trajectory variables is as follows: (5) (6) in, is the correlation function of the trajectory error between the robot and the agent, as the index of the i-th robot trajectory point at time k The judgment conditions for iterative updates, Represents the total number of trajectory points, and is the weight coefficient of the correlation function, It is l The correlation function threshold of the trajectory points is and represents the position and expected position of the robot's center of mass at time k, and They represent the pose and expected pose of agent i at time k respectively.

[0030] Through the analysis of the synchronization control method, it can be concluded that under the control of the iterative update rules of formula (5) and formula (6), each agent i Reference trajectory variables Updates are performed along the same iterative time step, and the trajectory parameters of different agents are synchronized, i.e. , the center of mass of each agent and robot will be at the reference position and maintain the desired pose.

[0031] The collaborative control of multiple agents is similar to the motion coordination of a mixed swarm of fixed-connected components. The robot as a whole is similar to the component swarm, and each agent is similar to a module in the swarm. The main source of the idea is the emergence of typical swarm behavior in nature that is realized by individual actions. Inspired by the emergence of swarm behavior, we can summarize the three behavioral principles of modules in the coordinated motion of components: 1. Trajectory tracking: Each agent follows a trajectory. The trajectory of the agent is calculated from the trajectory of the component center and the relative position of the agent in the formation relative to the component center. 2. Collision avoidance: The distance between each agent and obstacles is greater than the safe distance, including static and dynamic obstacles; 3. Motion matching: The motion state of an agent tends to match the motion state of an agent with high risk avoidance urgency.

[0032] Based on the above analysis, after resolving the system constraints of the agents, the distributed decision-making problem of the system must be addressed. The main objective of this embodiment is to design a distributed composite controller for the agents to ensure the stability of the closed-loop system's input state and synchronize all trajectory parameters. The composite controller consists of a distributed model predictive controller (DMPC) and a local compensation controller. Based on a nominal system without external disturbances, the trajectory tracking controller is designed using the DMPC method. A local compensation controller for unknown disturbances is designed using the extended state observer principle, and control variable compensation is used to mitigate the impact on the system. Furthermore, an urgency strategy is defined based on the distance between the agent and the obstacle. Using a shared communication network, the agent with the highest urgency transmits its urgency to other agents within the component. After comparing the urgency of other agents with its own, the agent adds matching constraints to its predictive controller to align its motion with that of the agent with the highest urgency. This solves high-dimensional nonlinear problems and achieves coordinated control and collaborative motion among the agents.

[0033] The multi-agent control problem involves the trajectory tracking of each agent and the collaborative matching and coordination of agents. The most important thing is to ensure that each agent tracks its reference trajectory and moves forward according to the given trajectory and posture; secondly, by ensuring that the trajectory iteratively updates the parameters The synchronization of ensures the consistency of reference trajectory of each agent, so as to achieve collaborative matching of agents, such as Figure 3 shown.

[0034] The proposed multi-agent system constraints will be combined to design a DMPC-based distributed cooperative controller with trajectory tracking, collision avoidance and motion matching as control objectives.

[0035] by Figure 3 For example, in the absence of external interference, the agent i exist The predictive control model for future states in the discrete time domain is: (7) in, is the time step, is the prediction time domain, represents the agent at time t i The n-1th state quantity, that is, the posture information. In the robot steering model: , It is i The heading angle of the agent relative to the robot, is the steering angular velocity of agent i, represents the nth control quantity of agent i at time t, Indicates the length of the robot.

[0036] Then formula (7) can be written as: (8) in, represents the agent at time t i The nth state quantity, Represents the heading angle of agent i.

[0037] Affected by its own mechanical structure, the intelligent body i The driving linear velocity and steering angular velocity are constrained, so they are expressed as upper and lower limit constraints of the control quantity. Medium Agent i The control quantity constraint is: (9) in, Indicates the minimum value of the control quantity, Indicates the maximum value of the control quantity, represents the n-th control quantity of agent i at time t.

[0038] Establish hard constraints on the predicted states between agents to describe the hard connection relationship between agents. Based on the synchronization control method, all agent controllers are optimized synchronously in each time step, and the future state is exchanged between each agent controller. k The predicted state information for each time step.

[0039] definition In the prediction time domain Medium Agent i The difference in control volume between the agents and the others, and the distance constraint is relaxed to Therefore, the coordination matching expression between agents is: (10) (11) in, express The 2-norm of represents the nth state of agent i at time t, represents the agent at time t j The nth state quantity, represents the physical constraint distance between agent i and agent j, is a very small positive number.

[0040] like Figure 2 As shown, the agent i The overall objective function form is designed as follows: (12) in, is the expected control quantity of agent i. In the objective function, 、 、 Item represents an agent i The error term of the state trajectory in the prediction time domain, that is, minimizing the deviation between the agent trajectory tracking term and the reference trajectory; The relaxation factor is the weight coefficient of each objective function. The system can adjust the influence of each objective function on the system motion by adjusting the size of the weight coefficient. In general, the relaxation factor takes a very large positive integer, and its weight coefficient is The value is small, just to meet the system convergence speed requirements; if there are obstacles, the weight coefficient Dynamic adjustment to prioritize obstacle avoidance requirements.

[0041] Combined with formula (8), the agent is defined within the prediction time domain i Tracking error for: (13) in, represents the agent at time t i The nth state quantity, represents the agent at time t i The expected posture of It is i The heading angle of the agent relative to the robot, is the steering angular velocity of agent i.

[0042] Therefore, the system tracking deviation objective function at time t is: (14) In order to meet the requirement of reducing system energy consumption, the control goal of the system is to achieve the reference state while minimizing the control amount at discrete moments in the time domain. Based on this, the objective function of the multi-agent distributed collaborative control is: (15) The superscript T represents transposition. Represents the n+1th control quantity of agent i at time t.

[0043] Similarly, the system optimization function needs to set a relaxation factor to smooth the optimization target of the system and improve the convergence speed. The penalty function is: (16) in, represents the nth relaxation factor at time t.

[0044] Combining Equations (14), (15) and (16) we can get the agent iThe overall objective function is: (17) in, As the weight coefficient of each objective function, the system can adjust the influence of each objective function on the system movement by adjusting the size of the weight coefficient.

[0045] The following is an analysis of the agent's obstacle avoidance and coefficient adjustment: According to equations (1), (2) and (3), the reference trajectory of each agent in normal circumstances is obtained by distributed measurement of the reference trajectory of the robot's center of mass, as follows: Figure 4 As shown in the normal situation. However, when a single agent detects an obstacle and finds that the distance from an agent to the obstacle is less than the safety threshold and needs to avoid the obstacle, the system needs to prioritize the calculation of the obstacle avoidance agent control input, while other agents need to perform motion constraint matching to solve the coordination problem. In order to solve the motion constraint matching problem, the agent anti-collision constraint function is designed. Indicates the possibility of collision between the agent and the obstacle in the time domain. Represents the obstacle state quantity and the agent's anti-collision constraint function for: (18) in, Represents the nth state quantity of agent i at time t.

[0046] Establishing urgency index based on anti-collision function To avoid collision behavior, the urgency is calculated as: (19) Each agent compares the urgency through a shared communication network, and the agent with the highest urgency is used as the target for motion matching with other agents.

[0047] by Figure 4 For example, under normal circumstances, the reference position of each agent can be calculated by equations (1), (2) and (3). However, when performing motion matching in the obstacle avoidance situation, the reference position calculation needs to be calculated based on the reference trajectory of the agent with the highest urgency, such as Figure 4 Obstacle avoidance situation.

[0048] Taking the left rear wheel reference trajectory of the robot as an example, under normal circumstances, the reference trajectory of this wheel changes with the right front wheel when the urgency is the highest and the reference trajectory of this wheel changes as shown in formula (20): (20) Therefore, the reference trajectory of each agent in the obstacle avoidance situation can be represented by a mapping set: Represents the relative position parameter function, then the agent iThe trajectory mapping relationship relative to the reference can be expressed as: (twenty one) in, is the predicted state sequence of the agent with the highest urgency. When performing motion matching coordination, the local reference trajectory of each agent is given by gather get.

[0049] After establishing the mapping relationship set, each agent compares the urgency level through the communication network. If the current agent's urgency is not the highest, it needs to match the movement with the agent with the highest urgency to design the obstacle avoidance relaxation factor. replace , and add it to the optimization function constraint set: (twenty two) Therefore, the penalty function of the obstacle avoidance relaxation factor at time t is designed as: (twenty three) When the system needs to perform an obstacle avoidance task, matching the motion of other agents with the motion of the agent with the highest urgency is the most important optimization goal of the current system. At this time, the obstacle avoidance factor should be given the highest weight. At the same time, in order to reduce the complexity of the system, the relaxation factor of the control convergence speed term is set to zero during the motion matching process, and only the relaxation factor objective function term is retained to avoid the system optimization function from having the desired feasible solution collide with the obstacle. Therefore, the optimization function of the distributed coordination controller single agent obtained by combining the objective function of formula (17) with the agent motion matching is: (twenty four) According to the above analysis, the constraints of the optimization function are: (25) Among them, the coefficients of each control objective function are: (26) In summary, combining equations (24) and (25), the final optimization goal of the system is yes: (27) Similarly, the nonlinear programming problem of formula (27) and the constraints of formula (25) are solved by the IPOPT solver using the interior point method. In addition, distributed cooperative control can enable the robot to perform body corrections when deviations occur during movement and ensure the normal performance of the robot's motion control tasks when a single wheel fails, as shown in Figure 5 (a)-Figure 5 (b), where, 、 、 、 、 and They represent the speeds of the left front wheel, left middle wheel, left rear wheel, right front wheel, right middle wheel and right rear wheel at the kth moment respectively, Indicates the speed of the entire vehicle.

[0050] Step 102: Call the extended state observer constructed based on the discrete error model of the intelligent agent and the external unknown disturbance to estimate the unknown disturbance in real time and generate the compensation control quantity. Through iterative optimization, the current optimal control input is combined with the generated compensation control quantity to generate a composite control quantity to control the robot.

[0051] Unknown disturbances during robot motion usually have an adverse effect on the control performance of the motion system. In real environments, disturbances that are difficult to measure directly, such as slippage or terrain undulations, can interfere with the robot's modeling in an ideal environment.

[0052] The Extended State Observer (ESO) is an algorithm for estimating the state of a dynamic system. It extends the traditional state observer to handle more complex system models. This paper proposes a feedforward compensation controller for estimating unknown disturbances using the ESO observer designed using the pole placement method to ensure the input state stability of the closed-loop system. Combined with the synchronization of all trajectory parameters in step 101, this solves the compensation control problem caused by the robot being affected by unknown disturbances. The control structure is as follows: Figure 6 shown.

[0053] 1. Unknown disturbance estimation and robot feedforward controller design: In this embodiment, a control strategy based on an extended state observer is proposed to solve the unknown disturbance problem in the multi-agent cooperative control process. Based on the cooperative control framework and combined with the tracking error model proposed in Equation (13), after linearizing the system control model near the reference trajectory point, the discrete error model of the agent can be described as follows: (28) in, , , is the Jacobian matrix of the error model, is the expected turning angular velocity of agent i, is the expected velocity of agent i, Is the agent i in The n-1th control input at time, represents the n-1th tracking error of agent i at time t.

[0054] In the distributed cooperative control process of multi-agent systems, unknown disturbances always exist and can adversely affect the system control performance, such as tire slip and terrain disturbances. Considering the impact of unknown disturbances on the motion of agents, these disturbances can be regarded as input additive disturbances.

[0055] Assumptions The kth external unknown disturbance at time The following increment equation is satisfied: (29) in, is the time increment of the external unknown disturbance, since is bounded at all times, so is also bounded, that is, it satisfies formula (30): (30) in, is the perturbation boundary.

[0056] Combined with Equation (29), the discrete error model Equation (28) of the intelligent agent system can be obtained as the following extended state equation: (31) in, . , , are the Jacobian matrices of the system. , is the coefficient matrix of the unknown disturbance, and I is the identity matrix.

[0057] The state information of the intelligent agent includes position coordinates and attitude angles. Error state model of the intelligent agent Considered as the output measurement of the extended state observer system, combined with Equation (31), the following observer form is designed in the proposed extended state observer system: (32) in, , and yes and The observed value of . is the gain matrix of the observer.

[0058] The extended state observer is an effective disturbance estimation method. When there are unknown disturbances in the control system, the estimated value of the observer can be used to compensate the control input in a feedforward control manner to reduce the impact of the disturbance on the system. Therefore, the composite compensation controller is designed as follows: (33) in, is the total sequence of the n-1th control input at time t of the extended state observer compensation, The optimal control sequence at time t obtained by the distributed model predictive controller is n-1, is the disturbance compensation gain.

[0059] like Figure 6 As shown, based on the characteristics of the DMPC controller, only The first control input in the control sequence is used to control the motion of the intelligent agent. As the time in the time domain increases, the controller continuously estimates and compensates to reduce the impact of unknown disturbances on the system motion.

[0060] By using the extended state observer control system designed by Equations (32) and (33), Equation (28) can be derived as the following closed-loop system: (34) Among them, let: (35) (36) in, ,because is bounded, so by choosing a suitable observer gain matrix Can be used The spectral radius is less than 1, making the observer error Bounded.

[0061] Through formula (35) and formula (36), we can get: (37) because Bounded, if , then the multi-agent system is asymptotically stable. Therefore, the interference can be attenuated by setting the interference compensation gain, so: (38) We can get: (39) By substituting Equation (39) into Equation (37), the following state equation of the system error model can be derived: (40) The control input of the agent calculated by the composite controller of the extended state observer combined with DMPC is designed in Equation (33), which includes the control quantity of DMPC and the compensation control quantity of the state observer.

[0062] The distributed model predictive control algorithm based on the state observer proposed in this embodiment is mainly used to solve the trajectory tracking problem of a six-wheeled robot under unknown disturbances, and to find the optimal control variable of the robot by iteratively looping the control input. The operation steps of the feedforward controller based on the extended state observer and the distributed control algorithm are as follows: Step 201: Initialize agent information: Given an agent i Initial state Reference trajectory parameter information , design constraint matching between agents ; Given the final optimization objective function coefficient matrix , and trajectory deviation threshold; Initialize the agent's control input , given the number of iterations and the maximum number of iterations ; For intelligent agents i Determine the appropriate observer gain matrix And design the appropriate disturbance compensation gain through formula (39) .

[0063] Step 202: Information synchronization: Agent i Exchange control inputs with other agents through the ROS system's shared communication network and trajectory parameter information .

[0064] Step 203: Iterative optimization: when When the agent i Solve the optimization problem through formula (27) and calculate the control sequence; If the motion deviations of all agents are less than the trajectory deviation threshold, the iteration ends. , go to step 205; Otherwise, , return to step 202.

[0065] Step 204: Observation of unknown disturbance: Through the designed extended state observer, the agent is estimated using formula (32) i disturbance .

[0066] Step 205: Combined with ESO intelligent body motion control: The agent is represented by formula (33) i Applying a composite compensation controller combined with DMPC control input With the gain after the compensation control input Generate compound control quantities to act on the intelligent agent; set up As an intelligent agent i The initial control input at the next moment; set up , , return to step 202.

[0067] Example 2 The purpose of this embodiment is to provide a robot distributed collaborative control system, wherein the robot is a six-wheel independently driven and independently steered robot, and the robot distributed collaborative control system includes: The model predictive control module is configured to treat each wheel of the robot as an independent agent, construct a distributed model predictive controller and corresponding constraints for each agent with the control objectives of trajectory tracking, collision avoidance, and motion matching between agents, and solve for the optimal control input for each agent within the prediction time domain; The control module is configured to: call an extended state observer constructed based on the discrete error model of the intelligent agent and external unknown disturbances, estimate the unknown disturbances in real time and generate compensating control quantities, and through iterative optimization, combine the current optimal control input with the generated compensating control quantities to generate a composite control quantity to control the robot.

[0068] In further embodiments, there is also provided: An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed by the processor. When the computer instructions are executed by the processor, the method described in Example 1 is performed. For the sake of brevity, no further details are given here.

[0069] It should be understood that in this embodiment, the processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), off-the-shelf field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0070] The memory may include a read-only memory and a random access memory, and provides instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0071] A computer-readable storage medium is used to store computer instructions, and when the computer instructions are executed by a processor, the method described in embodiment 1 is performed.

[0072] The method in Example 1 can be directly implemented as being executed by a hardware processor, or by a combination of hardware and software modules within the processor. The software module can be located in a storage medium well-established in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. The storage medium is located in the memory, and the processor reads the information in the memory and, in conjunction with its hardware, completes the steps of the above method. To avoid repetition, a detailed description is not given here.

[0073] A computer program product includes a computer program, and when the computer program is executed by a processor, the method described in embodiment 1 is implemented.

[0074] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as instructions contained in program modules, which are executed in a device on a real or virtual processor of a target to perform the process / method described above. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functionality of program modules can be combined or divided between program modules as needed. The machine-executable instructions for the program modules can be executed in local or distributed devices. In distributed devices, program modules can be located in local and remote storage media.

[0075] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the computer or other programmable data processing device, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on a computer, partially on a computer, as an independent software package, partially on a computer and partially on a remote computer, or entirely on a remote computer or server.

[0076] In the context of the present invention, computer program code or related data can be carried by any appropriate carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, and the like. Examples of signals include electrical, optical, radio, acoustic, or other forms of propagated signals, such as carrier waves, infrared signals, and the like.

[0077] Those skilled in the art will appreciate that the units and algorithm steps of the various examples described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0078] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. A robot distributed collaborative control method, characterized in that: The robot is a six-wheel independently driven and independently steered robot, and the robot distributed collaborative control method includes: Each wheel of the robot is considered an independent intelligent agent. With the control objectives of trajectory tracking, collision avoidance, and motion matching between agents, a distributed model predictive controller and corresponding constraints are constructed for each agent. The optimal control input of each agent in the prediction time domain is obtained. The extended state observer constructed based on the discrete error model of the intelligent agent and the external unknown disturbance is called to estimate the unknown disturbance in real time and generate the compensation control quantity. Through iterative optimization, the current optimal control input is combined with the generated compensation control quantity to generate a composite control quantity to control the robot.

2. A robot distributed collaborative control method according to claim 1, characterized in that: The optimization function corresponding to the distributed model predictive controller of the agent is constructed according to the robot system tracking deviation objective function, the multi-agent distributed collaborative control objective function and the robot optimization function.

3. A robot distributed collaborative control method according to claim 1 or 2, characterized in that: The constraints include: control quantity constraints on the driving linear velocity and steering angular velocity of the agent, hard constraints on the predicted states between agents used to describe the hard connection relationship between agents, and motion matching constraints between the urgency of the agent and the agent with the highest urgency.

4. A robot distributed collaborative control method according to claim 3, characterized in that: The hard constraints for predicting the state of each agent used to describe the hard connection relationship between agents are specifically: based on the agent i With the agent j The physical constraint distance between Medium Agent i Constrain the difference in state quantities between agents.

5. A robot distributed collaborative control method according to claim 1, characterized in that: The control strategy for motion matching between the agents is: According to the state of the obstacle and the state of the agent, an agent anti-collision constraint function is constructed to describe the possibility of collision between the agent and the obstacle in the time domain; Establish an urgency index based on the constructed agent anti-collision constraint function; Based on the mapping relationship between the agent and the reference trajectory and the calculation results of the urgency index, an obstacle avoidance relationship factor is designed to describe the motion matching between the agent and the agent with the highest urgency; When the system needs to perform an obstacle avoidance task, the most important optimization goal is to match the motion of other agents with the agent with the highest urgency. At this time, the obstacle avoidance relationship factor is given the highest weight, and the relaxation factor of the control convergence speed term is set to zero during the motion matching process to prevent the desired feasible solution from colliding with obstacles.

6. A robot distributed collaborative control method according to claim 1, characterized in that: The extended state observer constructed based on the discrete error model of the intelligent agent and the external unknown disturbance is called to estimate the unknown disturbance in real time and generate the compensation control quantity. The current optimal control input is combined with the generated compensation control quantity to generate a composite control quantity to control the robot. Specifically: Considering the impact of unknown disturbances on the agent's motion, the unknown disturbances are regarded as input additive disturbances and added to the agent's discrete error model; The error state model of the intelligent agent is regarded as the output measurement of the extended state observer system, and the extended state observer is designed; When there are unknown disturbances in the control system, the control input is compensated by feedforward control through the estimated value of the expanded state observer, and a compensation controller is designed. The composite control quantity for controlling the robot is obtained based on the compensation controller.

7. A robot distributed collaborative control method according to claim 1 or 6, characterized in that: The extended state observer is: in, and Yes and The observed value of , is the observer gain matrix, is the identity matrix, is the discrete error model of agent i, represents the time increment of the nth external unknown disturbance at time t, represents the n-1th control quantity of agent i at time t, represents the n-1th tracking error of agent i at time t , , , , is the expected turning angular velocity of agent i, is the expected velocity of agent i.

8. A robot distributed collaborative control system, characterized in that: The robot is a six-wheel independently driven and independently steered robot, and the robot distributed collaborative control system includes: The model predictive control module is configured to treat each wheel of the robot as an independent agent, construct a distributed model predictive controller and corresponding constraints for each agent with the control objectives of trajectory tracking, collision avoidance, and motion matching between agents, and solve for the optimal control input for each agent within the prediction time domain; The control module is configured to: call an extended state observer constructed based on the discrete error model of the intelligent agent and external unknown disturbances, estimate the unknown disturbances in real time and generate compensating control quantities, and through iterative optimization, combine the current optimal control input with the generated compensating control quantities to generate a composite control quantity to control the robot.

9. An electronic device, characterized in that: The method comprises a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 7 is completed.

10. A computer-readable storage medium, characterized in that Used to store computer instructions, which, when executed by a processor, complete the method according to any one of claims 1 to 7.

Images