Body-fitting intelligent motion control system and controller using same

Abstract

The invention discloses an intelligent motion control system with a body, which comprises a system-on-chip, the system-on-chip is provided with a plurality of interfaces, the system-on-chip is integrated with a multi-core CPU and a graphics processor, the system-on-chip runs an NECRO real-time operating system, and the control system comprises a multi-mode acquisition unit for receiving environment information acquired by an external sensor; the large model processing unit is used for calculating a task instruction according to the environment information; the motion control unit is used for calculating the motion track and speed of each joint controller of the robot according to the task instruction, generating a motion instruction and issuing the motion instruction to the joint controllers of the robot, the CPU comprises a plurality of cores, each core is correspondingly connected with an EtherCAT network port, the EtherCAT network ports are used for being connected with the joint controllers of the robot, and the EtherCAT network ports are used for being connected with the joint controllers of the robot. And the CPU establishes a core of the CPU as a master station according to the external connection condition of the EtherCAT network port, and establishes a slave station at an interface of the joint controller. According to the control system, the plurality of master stations are adopted to control the joint controllers of the robot respectively, so that control is smoother.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent robots, and in particular to an embodied intelligent motion control system and a controller using the system. Background Art

[0002] With the continuous development of artificial intelligence technology, the demand for robot motion control is also growing. Traditional robot control systems use a heterogeneous architecture, in which AI chips and X86 industrial computers each undertake different computing tasks and transmit and collaborate data through a complex interconnected network. Although this architecture can give full play to the professional advantages of each module when dealing with complex intelligent decision-making and computing tasks, it also leads to low overall computing efficiency due to data transmission delays and the complexity of the interconnected network. In particular, when dealing with large-scale data sets or tasks with high real-time requirements, traditional heterogeneous architectures will appear to be unable to cope with the situation.

[0003] Although traditional motion control systems can realize basic motion command execution, their control accuracy and flexibility are often limited when facing complex environments and changing tasks. Because the existing control system uses a single master station controller to control the joint controllers of the robot through one EtherCAT, whether it is routing or trajectory planning, the complexity can only be achieved by a single master station. In this case, the EtherCAT control cycle will become very long, the model calculation will become extremely complex, and it will consume processor resources. Summary of the invention

[0004] In order to overcome the defects in the prior art, the present invention provides an embodied intelligent motion control system and a controller using the system. The embodied intelligent motion control system integrates multimodal acquisition, large model processing and motion control on a system-level chip, and is scheduled by the NECRO real-time operating system. The single system-level chip integrates high-performance CPUs, graphics processors and other computing units on a single chip, thereby significantly improving computing power. In addition, the system uses multiple master stations to control each joint controller of the robot respectively. Each master station can work independently, and multiple cores of the multi-core CPU can perform stable data exchange internally, realizing communication between master stations. Through reasonable task scheduling and resource allocation, it is ensured that the commands sent by each master station can work together on the joint controller.

[0005] To achieve the above object, the technical solution adopted by the present invention is: an embodied intelligent motion control system, including a system-level chip, the system-level chip is provided with multiple interfaces, the system-level chip integrates a multi-core CPU and a graphics processor, the system-level chip runs a NECRO real-time operating system, and the embodied intelligent motion control system includes:

[0006] A multimodal acquisition unit receives environmental information collected by an external sensor through the interface;

[0007] A large model processing unit, running in the graphics processor, the large model processing unit is used to calculate a task instruction according to the environmental information;

[0008] A motion control unit runs in the CPU, and the motion control unit calculates the motion trajectory and speed of each joint controller of the robot according to the task instruction, generates a motion instruction, and sends the motion instruction to the joint controller of the robot, wherein the CPU includes multiple cores, each core is connected to an EtherCAT network port, and the EtherCAT network port is used to connect to the joint controller of the robot. The CPU establishes the core of the CPU as a master station according to the external connection status of the EtherCAT network port, and establishes a slave station at the interface of the joint controller.

[0009] Furthermore, the NECRO real-time operating system interacts with the hardware device through the EtherCAT driver, and the EtherCAT driver is responsible for initializing the hardware device, sending control commands to the hardware, receiving data from the hardware, and processing hardware interrupts.

[0010] Furthermore, the CPU and the graphics processor implement information exchange through shared memory, and the NECRO real-time operating system provides a mutex lock or a semaphore to ensure synchronization and mutual exclusion between processes.

[0011] Further, the motion control unit includes a master station management module responsible for the configuration, management and scheduling of the master station, the master station management module is used to monitor the status and performance of each master station, the master station management module uses distributed clock technology to achieve clock synchronization between the master stations, and the master station management module allocates communication resources and tasks to each master station through a priority scheduling algorithm;

[0012] The motion control unit also includes a synchronization control module responsible for synchronization information between the master stations, the synchronization control module uses a trigger signal or a timestamp mechanism to achieve synchronization triggering between the master stations, the synchronization control module uses a synchronization protocol to specify the communication sequence and data format between the master stations, and the synchronization control module uses a real-time guarantee mechanism to ensure the accuracy and real-time performance of synchronization control;

[0013] The motion control unit also includes a data processing module, which uses data frame encapsulation technology to encapsulate the control parameters into EtherCAT data frames, and uses a data parsing algorithm to extract the control parameters from the received data frames. The data processing module can convert the data format as needed;

[0014] The motion control unit also includes a communication interface module responsible for data communication between the multi-core CPU, the master station and the slave station. The communication interface module uses an Ethernet communication protocol to achieve data transmission between the multi-core CPU and the master station. The communication interface module uses a communication interface circuit and a driver to ensure stable data transmission and efficient processing.

[0015] Furthermore, the large model processing unit includes a Transformer model, which binds environmental information to corresponding task instructions through key values, generates a task directory and stores it in the memory of the graphics processor. The graphics processor can generate trajectories based on the task directory.

[0016] Furthermore, the motion control unit also includes a feedback monitoring module, which receives feedback data from the slave station, monitors the state and performance of the joint controller, and provides feedback information for the intelligent control algorithm of the motion control processing module. The feedback monitoring module uses a feedback data parsing algorithm to extract joint operation information from the received data frame. The motion control processing module can adjust the parameters and strategies of the intelligent control algorithm based on the feedback information. The feedback monitoring module uses real-time display technology to display the state and performance of the joint to the operator in the form of graphics or numerical values.

[0017] Furthermore, after the graphics processor generates the trajectory using the Transformer model, it plans the interpolation information of each joint of the robot according to the trajectory, converts the interpolation information into the format required by the EtherCAT protocol, and transmits the interpolation information to each joint controller through the master station.

[0018] An embodied intelligent motion controller for running the above-mentioned embodied intelligent motion control system, characterized in that it includes a core board, a USB board connected to the core board, a camera board connected to the core board, and a shell for accommodating the core board, the USB board and the camera board, at least three surfaces of the shell are provided with fixing ports, the core board, the USB board and the camera board are stacked in the shell, and the various interfaces of the core board, the USB board and the camera board are fixed to the fixing ports.

[0019] Furthermore, three sides of the core board are connected with a 10 Gigabit Ethernet port, a Gigabit Ethernet port, an HDMI port and an AUX audio port, wherein the Gigabit Ethernet port includes at least five of the EtherCAT ports;

[0020] The three sides of the USB board are connected with a USB3.2 interface, a MIPI interface, an RS485 interface and a CANFD interface;

[0021] Two sides of the camera board are connected with GMSL camera interfaces.

[0022] Furthermore, the USB board is provided with a board-to-board connector, the core board and the camera board are provided with a high-speed board-to-board connector, the core board and the USB board are connected via the board-to-board connector and the high-speed board-to-board connector, and the USB board and the camera board are connected via the board-to-board connector and the high-speed board-to-board connector.

[0023] By means of the above technical solution, the beneficial effects of the present invention are as follows:

[0024] The embodied intelligent motion control system uses multiple master stations to control each joint controller of the robot. Each master station can work independently, and the multiple cores of the multi-core CPU can exchange data stably, realizing communication between master stations. Through reasonable task scheduling and resource allocation, it is ensured that the commands sent by each master station can work together on the joint controller.

[0025] In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0027] Figure 1 is a connection diagram of an embodied intelligent motion control system according to an embodiment of the present invention;

[0028] Figure 2 is a schematic diagram of the overall structure of an embodied intelligent motion controller in an embodiment of the present invention;

[0029] Figure 3 yes Figure 2 Front view of

[0030] Figure 4 yes Figure 2 Left view of

[0031] Figure 5 is a schematic diagram of the structure of the core board in an embodiment of the present invention;

[0032] Figure 6 is a schematic diagram of the structure of a USB board in an embodiment of the present invention;

[0033] Figure 7 It is a schematic diagram of the structure of a camera board in an embodiment of the present invention.

[0034] The figure marks of the above drawings are: 11, power terminal; 12, EtherCAT network port; 13, Gigabit Ethernet network port; 14, 10 Gigabit Ethernet network port; 15, HDMI interface; 16, AUX audio interface; 21, USB3.2 interface; 22, RS485 interface; 23, CANFD interface; 24, DI interface; 25, DO interface; 31, GMSL camera interface; 41, slave station one; 42, slave station two; 43, slave station three. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and to distinguish similar objects. There is no order of precedence between the two, and they cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, "plurality" means two or more.

[0037] Example: Combination Figure 1-7 As shown, embodiment 1 discloses an embodied intelligent motion control system, including a system-level chip, the system-level chip is provided with multiple interfaces, the system-level chip integrates a multi-core CPU and a graphics processor, and is characterized in that the system-level chip runs a NECRO real-time operating system, and the embodied intelligent motion control system includes:

[0038] A multimodal acquisition unit receives environmental information collected by an external sensor through the interface;

[0039] A large model processing unit, running in the graphics processor, the large model processing unit is used to calculate a task instruction according to the environmental information;

[0040] A motion control unit runs in the CPU, and the motion control unit calculates the motion trajectory and speed of each joint controller of the robot according to the task instruction, generates a motion instruction, and sends the motion instruction to the joint controller of the robot, wherein the CPU includes multiple cores, each core is connected to an EtherCAT network port, and the EtherCAT network port is used to connect to the joint controller of the robot. The CPU establishes the core of the CPU as a master station according to the external connection status of the EtherCAT network port, and establishes a slave station at the interface of the joint controller.

[0041] The microkernel structure of the NECRO real-time operating system is tailorable and can be customized according to specific application scenarios. Its core features include a hard real-time kernel and a preemptive scheduling algorithm, which supports up to 99 levels of priority to ensure strong performance. NECRO provides a 100% native API interface and is 100% compatible with Linux commands and Shell scripts. NECRO also provides a rich set of SDKs that support more than 400 industry libraries and ecological libraries, including NVIDIA CUDA and Huawei Ascend generative AI ecology. In addition, NECRO has also integrated a variety of self-developed floating-point operation acceleration algorithms and visual acceleration algorithms. NECRO supports a variety of industrial control dedicated buses, such as EtherCAT and CANOPEN. In terms of PLC operation modes, NECRO can meet the needs of multiple modes such as single, periodic, continuous operation, interruption, etc., and supports multi-task control.

[0042] The system-on-chip is an integrated circuit with a dedicated purpose, which contains the entire system and all the embedded software. The multi-core CPU is a multi-core processor, which means that two or more complete computing engines (cores) are integrated into one processor.

[0043] In the present application, the NECRO real-time operating system interacts with the hardware device through the EtherCAT driver, and the EtherCAT driver is responsible for initializing the hardware device, sending control commands to the hardware, receiving data from the hardware, and processing hardware interrupts.

[0044] The large model processing unit includes a Transformer model. Compared with traditional recurrent neural networks (RNNs) or convolutional neural networks (CNNs), the Transformer model can better handle long-distance dependencies and has higher parallel computing capabilities. The Transformer model mainly consists of two parts: an encoder and a decoder. Each encoder and decoder contains multiple identical layers, each of which contains a self-attention mechanism and a feedforward neural network. In the encoder, the self-attention mechanism allows the model to focus on different parts of the input sequence, thereby capturing the dependencies within the sequence. The decoder uses the self-attention mechanism and the encoder-decoder attention mechanism to generate the output sequence.

[0045] The graphics processor is used for the calculation of the Transformer model. The graphics processor first collects environmental information, historical trajectory data, and robot status information. These data are used as inputs to the Transformer model. In order to input these data into the Transformer model, the graphics processor needs to convert them into a series of Tokens (flags, features). Token is the basic data unit that the model can understand and process. The graphics processor uses the Transformer model to learn and generate trajectories through the self-attention mechanism and encoder-decoder structure. During the training phase, the model adjusts internal parameters through an optimization algorithm to minimize the difference between the predicted trajectory and the actual trajectory. After the model training is completed, the graphics processor can use the model for trajectory generation.

[0046] After the GPU generates the trajectory using the trained model, it is also necessary to perform trajectory planning to generate specific interpolation points. These interpolation points describe the position and posture of the robot at different time points. The planned trajectory data is converted into the format required by the EtherCAT protocol to ensure that the data can be correctly transmitted in the EtherCAT network. Through the EtherCAT multi-master synchronization mechanism, the GPU will release the interpolation point data to all joint controllers synchronously. Each joint controller will receive the corresponding instructions at the same time and move according to the planned trajectory. A perfect fusion of multi-modal generative AI and motion control is achieved.

[0047] The multi-core CPU is connected to multiple slave stations via a MAC network card. The slave stations are the control systems of the robot joints. A slave station group may include multiple slave stations, such as Figure 1As shown, slave station one 41 in one of the slave station groups can be a thigh controller of the robot, slave station two 42 is a knee controller of the robot, and slave station three is an ankle controller of the robot. It should be noted that a slave station group can also include more slave stations. For example, a toe joint controller can be added to the above slave station group as slave station four. Compared with the prior art in which the joint controllers of the entire robot are connected to the same control master station, the present application has a small number of joint controllers controlled by one master station, simple calculation, and high efficiency.

[0048] The CPU and the graphics processor exchange information via shared memory, and the NECRO real-time operating system provides a mutex lock or semaphore to ensure synchronization and mutual exclusion between processes.

[0049] Wherein, the motion control unit comprises:

[0050] The master station management module is responsible for the configuration, management and scheduling of the master station. The master station management module can monitor the status and performance of each master station. The master station management module uses distributed clock technology to achieve clock synchronization between the master stations and allocates communication resources and tasks to each master station through a priority scheduling algorithm.

[0051] The synchronization control module is used to realize the synchronization control between the master stations, ensuring that the commands sent by each master station can work together on the joint controller. The synchronization control module adopts a trigger signal or a timestamp mechanism to realize the synchronization trigger between the master stations; the synchronization control module adopts a synchronization protocol to specify the communication sequence and data format between the master stations; the synchronization control module uses a real-time guarantee mechanism to ensure the accuracy and real-time performance of the synchronization control.

[0052] The data processing module encapsulates, analyzes and converts the data processed by the multi-core CPU to meet the transmission requirements of the EtherCAT network. The data processing module uses data frame encapsulation technology to encapsulate the control parameters into EtherCAT data frames, and the data processing module can extract the control parameters from the received data frames.

[0053] The communication interface module realizes data communication between the multi-core CPU, the master station and the slave station. The data transmission between the multi-core CPU and the master station is realized by the Ethernet communication protocol. The real-time communication between the master station and the slave station follows the EtherCAT communication protocol. The communication interface module uses a communication interface circuit and a driver to ensure stable transmission and efficient processing of data.

[0054] The feedback monitoring module receives the feedback data from the slave station, monitors the state and performance of the joint controller, and provides feedback information for the intelligent control algorithm of the application layer. The feedback monitoring module can extract the position and speed feedback information of the joint from the received data frame, and pass the feedback information to the Transformer model, and the Transformer model can adjust the task catalog according to the feedback information. The feedback monitoring module can also display the state and performance of the joint to the operator in the form of graphics or numerical values.

[0055] Through the above technical solution, compared with the traditional single master station controller, the 40 to 60 joints of the robot's body are controlled by one EtherCAT. Whether it is routing or trajectory planning, the complexity can only be achieved by a single CPU. In this case, the EtherCAT control cycle becomes very long, and the model calculation becomes extremely complex, which consumes CPU resources. This application uses multiple master stations to control each joint controller of the robot separately. For example, one master station can control two arms, one master station can control two mechanical legs, and the remaining master stations can independently control the head and torso, as well as various sensors throughout the body. Each master station can work independently, and the multiple cores of the multi-core CPU can exchange data stably, realizing communication between master stations. Through reasonable task scheduling and resource allocation, it is ensured that the commands sent by each master station can work together on the joint controller.

[0056] When data needs to be loaded from the multi-core CPU to the GPU or when results need to be returned from the GPU to the multi-core CPU, data transfer occurs. This usually involves memory copy operations, which can be implemented through specific APIs (such as .to(device) or .cuda() in PyTorch).

[0057] In order to maximize the utilization of the GPU, asynchronous computing is usually adopted, that is, the multi-core CPU continues to execute other tasks (such as data loading and preprocessing) while the GPU calculates the model at the same time.

[0058] When the multi-core CPU needs to wait for the calculation results of the GPU (for example, to obtain the loss value or perform gradient updates), synchronization operations are required. This is usually achieved through the synchronization mechanism provided by the framework.

[0059] In order to reduce the data transmission overhead between the multi-core CPU and the graphics processor, data is usually processed in batches to reduce the number of data transmissions. In addition, through reasonable data loading and preprocessing strategies, it can be ensured that the graphics processor always has enough data to process during calculation, thus avoiding waiting time.

[0060] The external sensors include high-resolution cameras, 2D cameras, 3D cameras, binocular cameras, etc. They can capture visual information in the environment, including color, shape, texture, depth, etc., and provide the robot with detailed data about the surrounding environment;

[0061] Auditory sensors (microphones) are used to receive sound signals and convert them into processable electrical signals. Robots can use auditory sensors to recognize voice commands, environmental noise, and other sound clues to interact with humans or respond to sound events in the environment.

[0062] Tactile sensors can detect the tactile information on the surface of an object, such as pressure sensors, temperature sensors, etc. Tactile sensors can help miracle people achieve functions such as fine manipulation, object grasping and posture perception;

[0063] LiDAR, which emits a laser beam and measures the time it takes for it to reflect back, to obtain 3D point cloud data of the surrounding environment. Infrared sensors detect infrared radiation and are useful for tasks such as night vision, temperature measurement, and object detection.

[0064] Odor sensors are used to detect odor molecules in the environment and can be used for applications such as air quality monitoring, food safety testing, and robots identifying objects by smell;

[0065] It also includes sensors such as accelerometers, gyroscopes, and magnetometers to measure and track the robot's motion state, including posture, speed, acceleration, etc. This information is crucial for the robot's navigation, balance, and motion control.

[0066] Embodiment 2 discloses an embodied intelligent motion controller for running the above-mentioned embodied intelligent motion control system, the embodied intelligent motion controller includes a core board, a USB board and a camera board, and a housing for accommodating the core board, the USB board and the camera board, at least three faces of the housing are provided with fixing ports. The USB board includes a USB3.2 interface, an RS485 interface and a CANFD interface for external devices; the camera board includes a GMSL camera interface for external devices. Among them, the USB board is connected to a ring six-microphone and an auditory sensor, and the camera board is connected to a depth camera, an RGB camera and a lip reading recognition camera, etc.

[0067] Optionally, the USB board can expand one USB channel for a 5G module.

[0068] Among them, Figure 5As shown, the first side of the core board is arranged with a 1×3 power terminal 11 with a rated voltage of 9 to 36V, an EtherCAT network port 12, a Gigabit Ethernet network port 13, a 10 Gigabit Ethernet network port 14, an HDMI interface 15 and an AUX audio interface 16;

[0069] The second side and the third side of the core board are respectively provided with one Gigabit Ethernet port 13 and two EtherCAT network ports 12.

[0070] The 10 Gigabit Ethernet port 14 is used to connect to devices such as radar or camera.

[0071] like Figure 6 As shown, the first side of the USB board is provided with six USB3.2 interfaces 21;

[0072] The second side and the third side of the USB board are respectively arranged with one USB3.2 interface 21, one RS485 interface 22, one CANFD interface 23, two DI interfaces 24 and two DO interfaces 25.

[0073] Among them, the USB3.2 interface 21 is used for debugging the serial port or conversion port, or connecting a ring six-microphone, an auditory sensor, etc.; the RS485 interface 22 is used for connecting a dexterous hand, etc.; the CANFD interface 23 is used for connecting a robot joint controller that supports CAN.

[0074] like Figure 7 As shown, four GMSL camera interfaces 31 are arranged on the second side and the third side of the camera board respectively.

[0075] The GMSL camera interface 31 is used to connect a depth camera, an RGB camera, a lip reading recognition camera, etc.

[0076] The interfaces of the core board, USB board and camera board are all fixed at the fixing ports of the shell. It should be noted that the fixing ports are configured to match the shapes of the interfaces.

[0077] The core board, the USB board and the camera board are stacked in the housing. The two sides of the USB board are provided with board-to-board connectors, the core board and the camera board are provided with high-speed board-to-board connectors, the core board and the USB board are connected via the board-to-board connector and the high-speed board-to-board connector, and the USB board and the camera board are connected via the board-to-board connector and the high-speed board-to-board connector.

[0078] The core board, USB board and camera board are stacked in the housing to form an embodied intelligent motion controller. Figure 2The interfaces of the core board, the USB board and the camera board are directly connected through the fixed ports on the side of the housing.

[0079] The present invention uses specific embodiments to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. An embodied intelligent motion control system, comprising a system-level chip, the system-level chip having a plurality of interfaces, the system-level chip integrating a multi-core CPU and a graphics processor, characterized in that: The system-level chip runs the NECRO real-time operating system, and the embodied intelligent motion control system includes: A multimodal acquisition unit receives environmental information collected by an external sensor through the interface; A large model processing unit, running in the graphics processor, the large model processing unit is used to calculate a task instruction according to the environmental information; A motion control unit runs in the CPU, and the motion control unit calculates the motion trajectory and speed of each joint controller of the robot according to the task instruction, generates a motion instruction, and sends the motion instruction to the joint controller of the robot, wherein the CPU includes multiple cores, each core is connected to an EtherCAT network port, and the EtherCAT network port is used to connect to the joint controller of the robot. The CPU establishes the core of the CPU as a master station according to the external connection status of the EtherCAT network port, and establishes a slave station at the interface of the joint controller.

2. The embodied intelligent motion control system according to claim 1, characterized in that: The NECRO real-time operating system interacts with the hardware device through the EtherCAT driver, and the EtherCAT driver is responsible for initializing the hardware device, sending control commands to the hardware, receiving data from the hardware, and processing hardware interrupts.

3. The embodied intelligent motion control system according to claim 1, characterized in that: The CPU and the graphics processor exchange information via shared memory, and the NECRO real-time operating system provides a mutex lock or semaphore to ensure synchronization and mutual exclusion between processes.

4. The embodied intelligent motion control system according to claim 1, characterized in that: The motion control unit includes a master station management module responsible for the configuration, management and scheduling of the master station. The master station management module uses distributed clock technology to achieve clock synchronization between the master stations. The master station management module allocates communication resources and tasks to each master station through a priority scheduling algorithm. The motion control unit also includes a synchronization control module responsible for synchronization information between the master stations. The synchronization control module uses a trigger signal or a timestamp mechanism to achieve synchronization triggering between the master stations. The synchronization control module uses a synchronization protocol to specify the communication sequence and data format between the master stations. The synchronization control module uses a real-time guarantee mechanism to ensure the accuracy and real-time performance of synchronization control. The motion control unit also includes a data processing module, which uses a data frame encapsulation technology to encapsulate the control parameters into an EtherCAT data frame, and uses a data parsing algorithm to extract the control parameters from the received data frame.

5. The embodied intelligent motion control system according to claim 1, characterized in that: The large model processing unit includes a Transformer model, which binds environmental information with corresponding task instructions through key values, generates a task directory and stores it in the memory of the graphics processor. The graphics processor can generate trajectories according to the task directory.

6. The embodied intelligent motion control system according to claim 5, characterized in that: The motion control unit also includes a feedback monitoring module, which receives feedback data from the slave station, monitors the state and performance of the joint controller, and provides feedback information for the Transformer model.

7. The embodied intelligent motion control system according to claim 6, characterized in that: After the graphics processor generates a trajectory using the Transformer model, it plans the interpolation information of each joint of the robot according to the trajectory, converts the interpolation information into the format required by the EtherCAT protocol, and transmits the interpolation information to each joint controller through the master station.

8. An embodied intelligent motion controller for operating the embodied intelligent motion control system according to any one of claims 1 to 7, characterized in that: It includes a core board, a USB board connected to the core board, a camera board connected to the core board, and a shell for accommodating the core board, the USB board and the camera board, at least three surfaces of the shell are provided with fixing ports, the core board, the USB board and the camera board are stacked in the shell, and the various interfaces of the core board, the USB board and the camera board are fixed to the fixing ports.

9. The embodied intelligent motion controller according to claim 8, characterized in that: The three sides of the core board are connected with a 10 Gigabit Ethernet port, a Gigabit Ethernet port, an HDMI port and an AUX audio port, wherein the Gigabit Ethernet port includes at least five EtherCAT ports; The three sides of the USB board are connected with a USB3.2 interface, a MIPI interface, an RS485 interface and a CANFD interface; Two sides of the camera board are connected with GMSL camera interfaces.

10. The embodied intelligent motion controller according to claim 9, characterized in that: The USB board is provided with a board-to-board connector, the core board and the camera board are provided with a high-speed board-to-board connector, the core board and the USB board are connected via the board-to-board connector and the high-speed board-to-board connector, and the USB board and the camera board are connected via the board-to-board connector and the high-speed board-to-board connector.

Images