A multi-robot distributed collaborative carrying method based on model predictive control

Abstract

The application belongs to the technical field of automation, and discloses an improved multi-robot collaborative carrying control method based on model predictive control, which comprises the following steps: inputting and torque inverse solution; MPC modeling and DOF setting; optimization target and constraint condition; ensuring stable contact between the robot and the object during carrying, and defining special carrying constraints; and formulating a comprehensive cost function to realize optimization of the carrying process. Through the improved MPC method and torque distribution strategy, the application proposes an efficient and accurate collaborative carrying control method for humanoid robots, which can not only realize coordination and cooperation among the robots, but also ensure accurate control of the torque and stability of the robot posture during the carrying process, thereby improving the overall performance of the multi-robot collaborative carrying.

Description

Technical Field

[0001] This invention belongs to, but is not limited to, the field of automation technology, and particularly relates to an improved model predictive control-based distributed load sharing and collaborative handling control method and system for multi-robot robots. Background Technology

[0002] With the rapid development of industrial automation and robotics, robots are increasingly used in tasks such as handling and assembly, especially in tasks requiring high precision and efficiency, where their advantages are becoming more and more apparent. Humanoid robots, as a type of robot with high flexibility and adaptability, have shown great potential in multiple fields. However, when faced with complex handling tasks, a single robot is often limited by factors such as load, task environment, and movement space, resulting in limited efficiency and stability during task execution.

[0003] To overcome the limitations of single robots, multi-robot systems (MRS) have gradually become an important research direction in robotics in recent years. Multi-robot collaboration can increase the overall flexibility, stability, and efficiency of a system by sharing tasks, especially when handling large or heavy objects, where it can effectively improve the accuracy and speed of task completion. Through reasonable task allocation and coordination, multiple robots can achieve more complex and efficient handling operations than a single robot in collaborative operations. Among them, humanoid robots, due to their humanoid characteristics, environmental adaptability, and excellent terrain mobility, have broad research space and application prospects in collaborative handling tasks.

[0004] In multi-robot cooperative handling, task allocation and motion coordination are two important research topics. Although multi-robot systems have broad application potential in the field of handling, research on the following aspects is still insufficient when it comes to cooperative handling of humanoid robots:

[0005] Unreasonable allocation of upper-level handling tasks: In collaborative handling tasks, how to rationally allocate tasks and ensure load balance among robots in different tasks remains an unresolved problem. Uneven task allocation can lead to excessive burden on individual robots, resulting in dynamic imbalance or decreased work efficiency. Furthermore, since a multi-robot collaborative handling system is a multi-input system, without a reasonable algorithm to coordinate the load-bearing capacity among the robots, excessive internal forces can be exerted on the object being handled, leading to resource waste or even damage to the object.

[0006] The lower-level humanoid robots are uncoordinated: Unlike traditional planar motion robots, the handling system is composed of multiple humanoid robots, which move by the periodic back-and-forth swinging of their leg joints. Therefore, if there is no reasonable multi-humanoid robot gait synchronization algorithm to make the leg movements of the humanoid robots tend to be synchronized, it will have a great impact on the stability of the handling system, causing it to vibrate up and down, and in severe cases, damage the object being handled, and then damage the humanoid robots.

[0007] The shortcomings of existing technologies are: the lack of an effective upper-level collaborative transport algorithm for load task allocation; the lack of a predictive control algorithm for humanoid robot models oriented towards dynamic trajectory tracking collaborative transport problems; and the lack of a multi-humanoid robot gait synchronization algorithm to enhance coordination capabilities. Summary of the Invention

[0008] To address the problems existing in the prior art, this invention provides an improved model predictive control-based distributed load sharing and collaborative handling control method for multi-robot robots.

[0009] This invention is implemented as follows: an improved model predictive control-based distributed load-sharing collaborative transport control method for multi-robot robots, the method comprising:

[0010] S1: A collaborative material handling algorithm for trajectory tracking and load sharing;

[0011] S2: Lower-level control algorithm for model predictive control (MPC) and whole-body control (WBC) in cooperative transport;

[0012] S3: An algorithm for gait synchronization in multi-human robots.

[0013] Furthermore, S1 specifically includes:

[0014] First, a dynamic model of the object being transported is established. To increase the universality of the task, the parameters of the dynamic model are unknown.

[0015] Based on the desired motion, a control scheme for the transported object is established. The controller is designed as a basic model reference adaptive controller to handle trajectory tracking and adaptive estimation of unknown parameters.

[0016] In the adaptive estimation of unknown parameters, a distributed communication network is designed, and a distributed consistency term is designed in the estimator to achieve proactive load sharing.

[0017] By following the steps above, we can obtain the control signals that each robot needs to apply to the object being moved.

[0018] The above steps constitute the upper-level algorithm for collaborative handling by multi-human robots.

[0019] S2 specifically includes:

[0020] A dynamic model of a humanoid robot in a transport task is established. This model is based on the traditional dynamic model of a humanoid robot, with the addition of a reaction force term to characterize the coupling between the humanoid robot and the object being transported.

[0021] An MPC model is established, including a loss function and constraints. The objective function aims to minimize motion tracking error and control costs. The impact of these two factors is measured using a weight matrix. Constraints include dynamic constraints, input constraints, contact constraints (hands, feet), and friction constraints.

[0022] MPC solving and WBC implementation. After performing MPC solving based on an open-source library, the MPC output results of the foot reaction force and joint motion state of each robot corresponding to the control input to achieve S1 are input into the WBC module, thereby obtaining the control input of each motor.

[0023] Through the above steps, motor control commands for the humanoid robot to achieve collaborative handling can be obtained.

[0024] S3 specifically includes:

[0025] A gait synchronization controller is established based on the distributed MPC method. The optimization variables are phase and step frequency. The loss function is the step frequency difference and phase difference between adjacent robots in the communication graph. The constraints include initial and terminal constraints as well as amplitude limiting constraints.

[0026] After solving the MPC problem using an open-source library, the optimized step frequency result is obtained and used as input into the leg trajectory planner to output the complete leg trajectory.

[0027] By following the steps above, the synchronized gait trajectory of a multi-humanoid robot can be obtained.

[0028] Another objective of this invention is to provide an improved model predictive control-based distributed load sharing and cooperative transport control system for multi-person robots, based on the aforementioned model predictive control-based distributed load sharing and cooperative transport control method. This system specifically includes:

[0029] The upper-layer distributed load sharing collaborative handling module is used to calculate the input force when handling objects with unknown dynamic models without generating internal forces;

[0030] Lower-level model predictive control and whole-body control module: Connects to the upper-level distributed load sharing and collaborative transport module, used to calculate the control input and state of the humanoid robot that implements the upper-level module, and finally calculates the joint motor input signals;

[0031] Distributed multi-robot gait synchronization module: Used to synchronize the gait of multi-robots, making the multi-robot handling process more stable and reliable.

[0032] Another object of the present invention is to provide a computer device including a memory and a processor, the memory storing a computer program, which, when executed by the processor, causes the processor to perform the steps of the improved model predictive control-based distributed load sharing collaborative handling control method for multi-person robots.

[0033] Based on the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by this invention are as follows:

[0034] First, this invention addresses the shortcomings of existing multi-robot cooperative handling technologies by proposing a multi-robot distributed load-sharing cooperative handling control method based on model predictive control, which has the following significant advantages:

[0035] 1. Implement distributed load sharing to avoid excessive internal load caused by multiple input systems, thus avoiding resource waste.

[0036] In existing multi-robot collaborative handling, most work considers load distribution based on geometric angles when allocating load tasks, using weighted pseudo-inverse matrices to achieve force-free handling of the object being transported. However, this requires pre-setting the geometric parameters of all robots relative to the object being transported, or using a fixed weighted pseudo-inverse matrix, and cannot proactively distribute the load according to the differences in load capacity among the robots.

[0037] Advantages of this technology: This invention comprehensively studies the generation mechanism of internal forces in multi-robot collaborative handling, solves the problem from the perspective of control theory, and considers the individual load-bearing differences among robots. It proposes a parameter estimation algorithm for distributed load sharing, which ensures the internal force-free characteristics of multi-robot collaborative handling in the steady state stage, avoids resource waste, and protects the system being handled.

[0038] 2. To enable tracking of the dynamic trajectory of transported objects.

[0039] In existing legged multi-robot cooperative transport systems, most work focuses on achieving cooperative transport at the kinematic level, rather than considering trajectory tracking at the dynamic level, which is necessary but complex. Furthermore, some data-driven kinematic solutions cannot achieve real-time control due to limitations in the algorithms themselves.

[0040] Advantages of this technology: This invention divides the collaborative transport task into an upper-level collaborative transport and a lower-level multi-human robot, which can effectively track the dynamic trajectory at the dynamic level, and uses an unknown dynamic modeling and control scheme to universalize the transported objects, thereby improving the adaptability of the task.

[0041] 2. Achieve gait synchronization for multi-human robots.

[0042] Existing gait synchronization in multi-human robot / multi-quadruped robot systems is considered within cooperative control algorithms, but there is currently no gait synchronization algorithm specifically designed for cooperative transport. Especially when considering terrain variations, a suitable gait synchronization algorithm is crucial for ensuring the stability and reliability of cooperative transport in multi-human robots.

[0043] Advantages of this technology: This invention proposes a gait synchronization algorithm, which rationally plans the gait of a humanoid robot system based on distributed model predictive control. This part focuses on the phase and amplitude of the gait, and by establishing reasonable objective functions and constraints, achieves gait synchronization for multiple humanoid robots, thereby ensuring the stability and reliability of collaborative transport by multiple humanoid robots at the application level.

[0044] Secondly, the expected benefits and commercial value of the technical solution of this invention after its transformation are as follows: The multi-robot collaborative handling technical solution has potentially wide-ranging application scenarios and commercial value, especially in the fields of industry, logistics, and construction. After transforming this technical solution into an actual product, the following benefits and commercial value can be expected:

[0045] 1. Improve work efficiency and flexibility

[0046] Expected benefits: Compared to traditional single robots or specialized mechanical equipment, multiple robots working collaboratively can handle more complex and flexible material handling tasks. These robots can automatically adjust their posture, path, and handling strategies to adapt to different object sizes, shapes, weights, and environmental conditions. This flexibility makes them suitable for a variety of non-standard and unpredictable handling tasks, significantly improving work efficiency.

[0047] Commercial value: In scenarios requiring frequent handling, such as manufacturing, warehousing and logistics, and construction sites, collaborative handling robots can reduce manual labor, improve handling efficiency, reduce errors and damage, and lower operating costs.

[0048] 2. Replacement for humans in hazardous environments

[0049] Expected Benefits: In harsh or hazardous working environments (such as mines, chemical plants, and nuclear power plants), multi-robot robots can replace humans in performing dangerous or heavy handling tasks, thereby protecting worker safety. These robots can reduce the risks associated with human operation by working collaboratively to move heavy objects or hazardous materials. Commercial Value: The application of this technology in high-risk environments can significantly reduce the risk of workplace accidents, thereby lowering companies' safety costs and insurance expenditures. Furthermore, such systems can improve a company's ability to continue working in harsh environments, thus increasing productivity.

[0050] 3. Saves labor costs

[0051] Expected benefits: Humanoid robot collaborative handling can replace a significant amount of human labor, especially in labor-intensive industries (such as logistics warehousing and factory handling), reducing companies' reliance on manpower. Robots can operate continuously for extended periods without rest, thus significantly reducing labor costs.

[0052] Business Value: With rising global labor costs, the application of robotic collaborative handling systems can help companies reduce labor costs, especially in logistics centers, distribution stations, and manufacturing plants, resulting in long-term cost savings and improved profit margins.

[0053] 4. Improve the level of automation and intelligence

[0054] Expected benefits: Multi-robot collaborative systems possess intelligent sensing, planning, decision-making, and execution capabilities, enabling seamless integration with other intelligent systems (such as warehouse management systems and supply chain management systems) to form highly automated and intelligent material handling solutions. This system can dynamically adjust its working modes according to environmental and task requirements, optimize material handling paths, and coordinate the division of labor and cooperation among multiple robots.

[0055] Business Value: Increased intelligence not only improves material handling efficiency but also reduces equipment downtime and resource waste, helping businesses optimize operations. For some smart factories and automated warehouses, this technology is one of their core competitive advantages.

[0056] 5. Wide range of applications and huge market potential.

[0057] Expected benefits: Collaborative handling robots can be applied in many different fields. In addition to common manufacturing and logistics, they can be extended to scenarios requiring object handling, such as disaster relief, medical care, and the military. In disaster relief scenarios, robots can collaboratively move heavy objects or rescue equipment; on the battlefield, they can transport supplies; and even in operating rooms, they can move medical equipment.

[0058] Commercial Value: This technology possesses broad market application potential, enabling its expansion into multiple industry sectors, such as warehousing and logistics (e.g., automation of Amazon warehouses), construction (handling heavy objects), services (moving goods in hotels, shopping malls, and hospitals), and rescue. With the wave of intelligent upgrading in the logistics and manufacturing industries, the market demand for this technology will continue to grow.

[0059] 6. Enhance the differentiated competitiveness of products and services.

[0060] Expected benefits: By applying multi-robot collaborative handling systems, businesses can offer more competitive products or services and respond quickly to customer needs. For example, in scenarios requiring flexible and efficient logistics, such as rapid delivery and customized production, these collaborative robots can significantly shorten delivery times and improve customer satisfaction.

[0061] Business Value: Companies can gain a competitive edge and enhance brand value by offering differentiated products or services. For example, logistics companies can increase market share through faster and more precise handling services, while manufacturing companies can meet customized needs and strengthen customer loyalty through more flexible production capabilities.

[0062] 7. Scalability and Modular Design

[0063] Expected benefits: Multi-robot collaborative handling systems can be designed modularly, allowing for flexible configuration of the number and capabilities of robots according to task requirements. Enterprises can start with small-scale deployments and gradually expand the robot system as demand increases, reducing initial investment risk.

[0064] Business Value: This modularity and scalability allow businesses to flexibly adjust their investment plans according to business needs, avoiding large-scale one-time capital expenditures. This scalability is particularly attractive to SMEs, helping to reduce early-stage investment pressure and gradually increase automation levels.

[0065] 8. Long-term development and sustainability

[0066] Expected benefits: Collaborative material handling robot technology can improve resource utilization and reduce energy consumption and material loss. For example, by intelligently optimizing paths and task allocation, energy waste during the handling process can be significantly reduced. Furthermore, robots can be continuously maintained and upgraded, extending system lifespan and reducing long-term operating costs.

[0067] Business Value: In the global trend of emphasizing sustainable development, environmentally friendly and efficient collaborative robot systems meet the needs of enterprises to reduce carbon emissions and save energy. They can help enterprises obtain green certification, enhance their brand's social responsibility image, and attract more customers and investors.

[0068] Multi-robot collaborative material handling technology has significant expected benefits and commercial value, particularly in improving handling efficiency, reducing costs, enhancing safety, and increasing intelligence. With the increasing demand for automation and intelligence in industries and logistics, this technology is expected to play a vital role in multiple sectors and become an indispensable force in the future market. Attached Figure Description

[0069] Figure 1 This is a flowchart of an improved model predictive control-based distributed load sharing and collaborative handling control method for multi-human robots provided in an embodiment of the present invention.

[0070] Figure 2 This is a diagram of the comprehensive cost function provided in an embodiment of the present invention;

[0071] Figure 3 This is a schematic diagram of the upper-layer collaborative transport algorithm provided in an embodiment of the present invention:

[0072] Figure 4 This is a schematic diagram of internal forces provided in an embodiment of the present invention:

[0073] Figure 5 This is a trajectory tracking result diagram provided in an embodiment of the present invention:

[0074] Figure 6 The following are simulation results of parameter estimation and parameter estimation error provided in the embodiments of the present invention:

[0075] Figure 7 This is a diagram showing the internal force results provided in an embodiment of the present invention:

[0076] Figure 8 This is a diagram showing the experimental trajectory tracking results provided in an embodiment of the present invention:

[0077] Figure 9 This is a graph showing the experimental parameter estimation results provided in an embodiment of the present invention: Detailed Implementation

[0078] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0079] like Figure 1 As shown, this invention provides an improved model predictive control-based distributed load sharing and collaborative handling control method for multi-robot robots.

[0080] This invention is implemented as follows: an improved model predictive control-based distributed load-sharing cooperative transport control method for multi-human robots. First, the environmental characteristics of winter are analyzed to determine the specific requirements of the transport task. Next, considering the dynamic model of the object being transported and its load-sharing characteristics, a cooperative transport upper-level algorithm based on trajectory tracking and load sharing is used to obtain the desired force applied by the robot to the object. Then, an MPC model for the humanoid robot is established, including the establishment of the objective function and constraints, and the desired applied force is considered as part of the MPC model, which is then solved. Further, the results of the MPC solution are input into the WBC module, and the WBC problem is solved to obtain optimized input commands for each joint of the humanoid robot. Finally, a distributed gait synchronization algorithm is run among the humanoid robots to obtain gait commands, thereby achieving gait synchronization of the humanoid robot group and improving the overall performance of the transport process.

[0081] The present invention can also employ the following method, which includes:

[0082] S1: A collaborative material handling algorithm for trajectory tracking and load sharing;

[0083] S2: A lower-level control algorithm suitable for Model Predictive Control (MPC) and Whole Body Control (WBC) in cooperative transport;

[0084] S3: An algorithm for gait synchronization in multi-human robots.

[0085] like Figure 3 The core of the upper-level collaborative handling algorithm lies in distributed load sharing and trajectory tracking. Through force and displacement sensors between the robot and the object being handled, the system acquires real-time interaction data (such as contact forces and object displacement). Based on the object's dynamic model and task requirements, the upper-level algorithm calculates the force distribution and target trajectory to be applied by each robot, ensuring uniform load distribution during handling and avoiding internal force problems caused by force imbalances. This module further plans the handling path, ensuring stability and efficiency while the object is being manipulated.

[0086] The lower-level control algorithms include Model Predictive Control (MPC) and Whole-Body Control (WBC). MPC uses a physical model and future state predictions to calculate optimized control inputs for the robot joints, ensuring high precision and responsiveness in robot movements. Simultaneously, the Whole-Body Control module coordinates the movements of multiple robot joints to achieve overall balance and execute the target task. By integrating force feedback signals and trajectory planning results, WBC dynamically adjusts the robot's posture, optimizing the angle, velocity, and torque inputs of the robot joints, ultimately driving the joint motors to complete the transport task.

[0087] In multi-robot collaborative transport tasks, each robot needs to maintain gait synchronization to ensure the stability of the transport process. Gait synchronization algorithms coordinate the gait rhythm of all robots through real-time communication, based on each robot's position, stride length, gait phase, and other state data. During transport, if a robot experiences gait deviation or positional shift, the gait synchronization module adjusts the steps of other robots based on feedback signals to ensure overall coordination. The gait synchronization algorithm also utilizes predictive adjustment strategies to proactively correct potential inconsistencies, improving transport stability.

[0088] The entire system uses sensors to collect interaction data between the robot and the object, as well as the robot's own state information, forming a closed-loop control. A real-time feedback mechanism monitors the actions of each robot, and when deviations or disturbances occur, the system dynamically adjusts trajectory planning and control parameters. During control, MPC (Multi-Process Control) combines feedback data to optimize control inputs in real time, ensuring the robot maintains stability and accuracy in handling tasks even in complex environments. Furthermore, the system records task data for optimizing model parameters and improving performance in the next task.

[0089] like Figure 5 S1 specifically includes:

[0090] First, a dynamic model of the object being transported is established. To increase the universality of the task, the parameters of the dynamic model are unknown.

[0091] like Figure 6 Based on the desired motion, a control scheme for the transported object is established. The controller is designed as a basic model reference adaptive controller to handle trajectory tracking and adaptive estimation of unknown parameters.

[0092] In the adaptive estimation of unknown parameters, a distributed communication network is designed, and a distributed consistency term is designed in the estimator to achieve proactive load sharing.

[0093] The above steps yield the control signals that each robot needs to apply to the object being transported. These steps constitute the upper-level algorithm for collaborative transport by multi-robot systems.

[0094] S2 specifically includes:

[0095] A dynamic model of a humanoid robot in a transport task is established. This model is based on the traditional dynamic model of a humanoid robot, with the addition of a reaction force term to characterize the coupling between the humanoid robot and the object being transported.

[0096] An MPC model is established, including a loss function and constraints. The objective function aims to minimize motion tracking error and control costs. The impact of these two factors is measured using a weight matrix. Constraints include dynamic constraints, input constraints, contact constraints (hands, feet), and friction constraints.

[0097] MPC solving and WBC implementation. After performing MPC solving based on an open-source library, the MPC output results of the foot reaction force and joint motion state of each robot corresponding to the control input to achieve S1 are input into the WBC module, thereby obtaining the control input of each motor.

[0098] Through the above steps, motor control commands for the humanoid robot to achieve collaborative handling can be obtained.

[0099] S3 specifically includes:

[0100] A gait synchronization controller is established based on the distributed MPC method. The optimization variables are phase and step frequency. The loss function is the step frequency difference and phase difference between adjacent robots in the communication graph. The constraints include initial and terminal constraints as well as amplitude limiting constraints.

[0101] After solving the MPC problem using an open-source library, the optimized step frequency result is obtained and used as input into the leg trajectory planner to output the complete leg trajectory.

[0102] By following the steps above, the synchronized gait trajectory of a multi-humanoid robot can be obtained.

[0103] like Figure 4 This invention provides an improved model predictive control-based distributed load sharing and cooperative transport control system for multi-person robots, based on the improved model predictive control-based distributed load sharing and cooperative transport control method. The system specifically includes:

[0104] like Figure 7 The upper-layer distributed load sharing collaborative handling module is used to calculate the input force when handling objects with unknown dynamic models without generating internal forces;

[0105] Lower-level model predictive control and whole-body control module: Connects to the upper-level distributed load sharing and collaborative transport module, used to calculate the control input and state of the humanoid robot that implements the upper-level module, and finally calculates the joint motor input signals;

[0106] Distributed multi-robot gait synchronization module: Used to synchronize the gait of multi-robots, making the multi-robot handling process more stable and reliable.

[0107] This invention provides a computer device, which includes a memory and a processor. The memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the improved model predictive control-based distributed load sharing and collaborative handling control method for multi-person robots.

[0108] like Figure 8 The upper-layer distributed load-sharing collaborative transport module considers the task-oriented expected trajectory and the humanoid robot's load factor based on load capacity. It employs a collaborative transport algorithm based on trajectory tracking and load sharing to obtain the robot's expected force on the object. This part considers the model uncertainty of the transported object to improve the algorithm's versatility. The expected force obtained from running the algorithm serves as the input basis for MPC.

[0109] like Figure 9 The upper-level module calculates the force and torque required for each robot based on the desired trajectory signal to achieve a uniform load distribution and avoid internal force problems caused by unbalanced force application among the robots. During this process, estimated parameters from the object's unknown dynamic model (such as mass and moment of inertia) are used, combined with a load-sharing algorithm, to generate uniform force and torque commands for the robots and output control instructions to the lower-level module.

[0110] The lower-level model predictive control module calculates the required body motion commands and foot reaction force commands for each robot based on the forces and torques output by the upper-level module. These commands are then output by the MPC module and transmitted to the whole-body control (WBC) module. The WBC module calculates the control commands required for each joint based on the input signals from the MPC module and converts them into voltage and current signals required by the motor drivers. Each robot drives its respective joint motors according to these signals, ensuring coordinated movements. To ensure signal stability and interference resistance, filtering and noise suppression are also performed during signal generation.

[0111] The distributed multi-robot gait synchronization module is based on an additional model predictive control module. This module includes a distributed MPC objective function and constraints such as initial and terminal constraints and amplitude limiting constraints, and solves the MPC using an open-source library. The gait control algorithm adjusts the phase and stride of each robot's gait to ensure synchronization between robots, thereby ensuring the stability and accuracy of the transport task.

[0112] Throughout the control process, the system records input signals, output signals, and feedback data in real time and stores them in memory. The processor periodically analyzes the data to optimize the control strategy and improve control efficiency for future tasks. Furthermore, the stored historical data can be used for dynamic model calibration and adjustment of initial parameters for model predictive control in the next task, improving the system's adaptability and reliability.

[0113] like Figure 2This invention discloses an improved model predictive control (MPC)-based distributed load-sharing cooperative transport control method for multi-robot humanoid robots. The method includes: input and inverse torque solution; MPC modeling and degree-of-freedom setting; optimization objectives and constraints; ensuring stable contact between the robot and the object during transport by defining specific transport constraints; and formulating a comprehensive cost function to optimize the transport process. This invention proposes an efficient and accurate cooperative transport control method for humanoid robots through an improved MPC method and torque allocation strategy. It not only enables coordinated cooperation among robots but also ensures precise torque control and stable robot posture during transport, thereby improving the overall performance of multi-robot cooperative transport.

[0114] It should be noted that embodiments of the present invention can be implemented in hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by a suitable instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the above-described devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuitry such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of the above-described hardware circuitry and software, such as firmware.

[0115] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any modifications, equivalent substitutions, and improvements made by those skilled in the art within the scope of the technology disclosed in the present invention, and within the spirit and principles of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An improved model predictive control-based distributed load-sharing cooperative transport control method for multi-robot robots, characterized in that, The method includes: S1: A collaborative material handling algorithm for trajectory tracking and load sharing; S2: A lower-level control algorithm suitable for Model Predictive Control (MPC) and Whole Body Control (WBC) in cooperative transport; S3: An algorithm for gait synchronization in multi-human robots; S1 specifically includes: First, a dynamic model of the object being transported is established. To increase the universality of the task, the parameters of the dynamic model are unknown. Based on the desired motion, a control scheme for the transported object is established. The controller is designed as a model reference adaptive controller to handle trajectory tracking and adaptive estimation of unknown parameters. In the adaptive estimation part of unknown parameters, a distributed communication network is designed, and a distributed consensus term is designed in the estimator to achieve active load sharing. Through the above steps, the control signals that each robot needs to apply to the object being transported can be obtained; S2 specifically includes: establishing a humanoid robot dynamics model for the handling task, which is based on the traditional humanoid robot dynamics model, with the addition of a reaction force term to characterize the coupling effect between the humanoid robot and the object being handled; An MPC model is established, including a loss function and constraints; the objective function aims to minimize motion tracking error and control cost; the influence of both is measured by a weight matrix; constraints include dynamic constraints, input constraints, contact constraints, and friction constraints. S3 specifically includes: establishing a gait synchronization controller based on a distributed MPC method, with phase and step frequency as the optimization variables, the loss function being the step frequency difference and phase difference between adjacent robots in the communication graph, and constraints including initial and terminal constraints as well as amplitude limiting constraints; After solving the MPC problem using an open-source library, the optimized step frequency result is obtained and used as input into the leg trajectory planner to output the complete leg trajectory. Through the above steps, the synchronized gait trajectory of a multi-human robot can be obtained; It also includes the following steps: Acquire interaction data between multiple robots and the objects being transported, including contact forces, object displacement, and robot state information; In the upper-level collaborative handling module, the distributed load-sharing force between the target trajectory and the robot is calculated based on task requirements and object dynamics model to avoid generating internal forces during the handling process. In the lower-level control module, Model Predictive Control (MPC) is used to optimize and predict the robot's motion state, and Whole Body Control (WBC) is used to coordinate the movement of the robot's joints to achieve accurate tracking of the target in the upper-level module. The distributed gait synchronization module adjusts the rhythm, stride, and phase of multiple robot gaits in real time to ensure gait consistency and system stability during transport.

2. The method as described in claim 1, characterized in that, The distributed load sharing collaborative transport control system includes: The upper-layer distributed load sharing and collaborative handling module is used to calculate the force distribution and target trajectory applied by each robot under the task requirements; The lower-level Model Predictive Control (MPC) module and Whole Body Control (WBC) module are used to optimize robot joint control inputs and postures to achieve dynamic adjustment and precise manipulation; The distributed gait synchronization module is used to achieve gait coordination between robots through real-time communication and status feedback, so as to avoid system instability caused by gait deviation during the handling process; The data processing module is used to collect and analyze the interaction data between the robot and the object, and to provide real-time feedback during the task completion process to adjust the control strategy.

3. The method as described in claim 1, characterized in that, The model predictive control algorithm of the lower-level control module includes the following steps: Construct an optimization problem model that includes robot kinematics and dynamics constraints; By combining the robot's current state feedback and future goals, and by predicting the changes in the robot's state within the planning time window, the optimal solution for the joint inputs is calculated. The whole-body control module dynamically distributes joint torques to ensure robot posture balance and simultaneously achieve target trajectory tracking. During the transportation process, real-time interactive data is used to dynamically update the model's predictive control parameters to adapt to external disturbances and load changes.

4. An improved cooperative control system based on model predictive control (MPC) and torque distribution strategy for multi-robot distributed load sharing and cooperative handling control as described in any one of claims 1-3, characterized in that, The system specifically includes: The upper-layer distributed load sharing collaborative handling module is used to calculate the input force when handling objects with unknown dynamic models without generating internal forces; Lower-level model predictive control and whole-body control module: Connects to the upper-level distributed load sharing and collaborative transport module, used to calculate the control input and state of the humanoid robot that implements the upper-level module, and finally calculates the joint motor input signals; Distributed multi-robot gait synchronization module: Used to synchronize the gait of multi-robots, making the multi-robot handling process more stable and reliable.

5. A computer device, characterized in that, The computer device includes a memory and a processor. The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the improved model predictive control-based distributed load-sharing collaborative transport control method for multi-person robots as described in any one of claims 1-3.

Images