A method and system for embodied intelligence low latency remote interaction and control

Abstract

The application relates to the technical field of interaction and control, and discloses a low-latency remote interaction and control method and system for embodied intelligence, which comprises the following steps: arranging a cloud collaborative platform, establishing a remote interaction channel between an edge control unit arranged on an embodied intelligent device side and a remote interaction client through a low-latency communication network; receiving and analyzing mixed data streams in the remote interaction channel through the edge control unit, separating corresponding instruction data and interaction data from the mixed data streams, and establishing a logically independent control data channel for the instruction data; and assigning a first priority to the instruction data in the control data channel and a second priority to the interaction data through the edge control unit. The application can solve the problem that mixed data stream transmission may cause congestion, packet loss and delay aggravation, which directly affects the motion control accuracy of the device and the safety of remote operation.

Description

Technical Field

[0001] This invention relates to the field of interaction and control technology, and in particular to a low-latency remote interaction and control method and system for embodied intelligence. Background Technology

[0002] Current mainstream solutions mostly use a single channel to carry a mixed data stream of control commands, video images, and sensor data. In this approach, control commands and non-control interaction data share the transmission link, bus resources, and interrupt permissions. Command transmission is easily crowded by high-bandwidth interaction data, resulting in large control command jitter and uncontrollable response delays. This makes it difficult to meet the hard real-time control requirements of the actuators of embodied intelligent devices. At the same time, when network signal attenuation and latency fluctuations exceed limits, the transmission of mixed data streams will experience congestion, packet loss, and increased latency, directly affecting the motion control accuracy of the device and the safety of remote operation. Summary of the Invention

[0003] This invention provides a low-latency remote interaction and control method and system for embodied intelligence, the main purpose of which is to address the problems raised in the background art above.

[0004] To achieve the above objectives, the present invention provides a low-latency remote interaction and control method for embodied intelligence, comprising: S1. Install a cloud-based collaborative platform to establish a remote interaction channel between the edge control unit deployed on the smart device side and the remote interaction client through a low-latency communication network; S2. The edge control unit receives and parses the mixed data stream in the remote interaction channel, separates the corresponding instruction data and interaction data from the mixed data stream, and establishes a logically independent control data channel for the instruction data. S3. The edge control unit assigns a first priority to the instruction data in the control data channel and a second priority to the interactive data. S4. Apply the allocated instruction data to the actuator of the embodied intelligent device, and collect the motion state data of the actuator in real time as a feedback signal. Asynchronously synchronize the log data and running status in the feedback signal as non-control data to the cloud collaborative platform. S5. Under the premise of ensuring that the first priority of the instruction data is not affected, transmit the interaction data to the remote interaction client through the remote interaction channel; S6. When the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, it automatically reduces the amount of interactive data and the transmission frequency, and maintains the first priority of the control data channel.

[0005] Preferably, establishing a remote interaction channel between the edge control unit deployed on the embodied intelligent device side and the remote interaction client via a low-latency communication network includes: The low-latency communication network transmits link probe pulses and receives return confirmation pulses of the transmission results. The one-way transmission delay fluctuation range and round-trip delay difference of the low-latency communication network are determined step by step by the transmit link detection pulse and the return confirmation pulse. By combining the one-way transmission delay fluctuation range and the round-trip delay difference, the uncertainty margin of the low-latency communication network is extracted to obtain the maximum continuous carrying period and the minimum idle protection interval corresponding to the low-latency communication network. The physical resource blocks of the low-latency communication network are divided into hard real-time time slot intervals according to the continuous bearing period and the idle protection interval, so as to obtain the boundary parameter frames of the hard real-time time slot intervals. Based on the boundary parameter frame, a two-way handshake synchronization and asynchronous transmission parameter negotiation are performed on the time slot boundary between the edge control unit and the remote interactive client to obtain a remote interactive channel.

[0006] Preferably, receiving and parsing the mixed data stream in the remote interaction channel through the edge control unit includes: The physical transmission medium of the remote interaction channel is collected to obtain the original electrical signal sequence captured by the physical layer interface corresponding to the edge control unit; A serial bit stream is obtained by synchronizing the original electrical signal sequence with clock edges and performing bit decisions. The serial bit stream is parsed and its payload is extracted to obtain the frame payload data block sequence of the serial bit stream; The consecutive data blocks in the frame payload data block sequence are concatenated end to end, and the frame delimiter and check field in the concatenation result are removed to obtain the mixed data stream of the remote interaction channel.

[0007] Preferably, the step of separating the corresponding instruction data and interaction data from the mixed data stream and establishing a logically independent control data channel for the instruction data includes: A sliding window energy integration is performed on consecutive bits in the mixed data stream to obtain the bit energy distribution sequence of the mixed data stream on the time axis; Determine the alternating boundaries between local energy valleys and local energy peaks in the bit energy distribution sequence; The mixed data stream is segmented into bit segments according to the alternating boundaries to obtain constant energy bit segments; Determine the duration of the constant energy bit segment, and calculate the length variance of the length sequence corresponding to the constant energy bit segment based on the duration; The constant energy bit segments with length variance smaller than the overall length variance of the mixed data stream are selected to obtain a sequence of instruction data frame segments that arrive stably in a periodic manner. The instruction data frame sequence is physically mapped from the mixed data stream to the first-in-first-out buffer queue of the edge control unit; Assign an independent bus arbitration identifier and interrupt request line to the first-in-first-out buffer queue to obtain a control data channel that shares a physical link with the interactive data but is isolated from the arbitration path.

[0008] Preferably, the step of assigning a first priority to the instruction data in the control data channel and assigning a second priority to the interaction data through the edge control unit includes: Analyze the difference between the arrival time of the instruction data and its adjacent data in the control data channel, and take the absolute value of the difference to obtain the instruction jitter amplitude sequence of the instruction data; Analyze the difference between the arrival times of the interactive data and the adjacent interactive data, and take the absolute value of all differences to obtain the interaction jitter amplitude sequence of the interactive data; When the maximum value in the instruction jitter amplitude sequence is less than the minimum value in the interaction jitter amplitude sequence, the instruction data is marked as deterministic periodic data and assigned the first priority; otherwise, the instruction data is marked as an abnormal stream and remains in an unassigned priority state. Based on the first priority already obtained from the instruction data, a second priority is assigned to the interactive data; Set the interrupt request response latency limit bound to the first priority to one-third of the interrupt request response latency limit bound to the second priority.

[0009] Preferably, the step of applying the allocated instruction data to the actuator of the embodied intelligent device and collecting the motion state data of the actuator in real time as a feedback signal includes: Obtain the instruction codeword from the allocated instruction data; The instruction codewords are sequentially written into the hold latch of the digital-to-analog converter register in the edge control unit to obtain the analog voltage control waveform of the actuator; Motion state data is obtained by applying the analog voltage control waveform to the winding drive input terminal of the actuator; The motion state data is converted into a feedback signal for the actuator.

[0010] Preferably, the step of asynchronously synchronizing the log data and operating status in the feedback signal as non-control data to the cloud collaboration platform includes: The log data segment and the running status data segment are parsed from the feedback signal; The log data segment and the running status data segment are written to the edge control unit in the order of their generation time to obtain a non-control data cache queue; A globally incrementing sequence number is appended to each data block in the non-control data buffer queue, and the data is encapsulated into an independent data frame to obtain an asynchronous synchronous data frame sequence. The asynchronous synchronous data frame sequence is sent to the network port of the cloud collaboration platform.

[0011] Preferably, transmitting the interaction data to the remote interaction client through the remote interaction channel, while ensuring that the first priority of the instruction data is not affected, includes: The depth of the transmission queue of the remote interaction channel is detected to obtain the fixed transmission duration; The minimum idle time interval between two adjacent instruction data is analyzed based on the fixed transmission duration. The window length is determined based on the minimum idle time interval. The interactive data is divided into data blocks that do not exceed the window length, and the second priority is attached to each data block to obtain an interactive data fragment sequence; The interactive data fragment sequence is transmitted to the remote interactive client.

[0012] Preferably, when the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, automatically reducing the amount of interactive data and the transmission frequency includes: The instantaneous value of the signal strength of the currently received signal in the low-latency communication network is collected, and the instantaneous value of the signal strength is compared with a preset threshold to obtain a judgment flag; When the determination flag is valid, the maximum capacity of the sending buffer for the interactive data is reduced to half of the original capacity, and the sending interval is extended to twice the original interval.

[0013] A low-latency remote interaction and control system for embodied intelligence, used to implement a low-latency remote interaction and control method for embodied intelligence, the system comprising: The building module is used to house the cloud collaboration platform and establishes a remote interaction channel between the edge control unit deployed on the smart device side and the remote interaction client through a low-latency communication network; The channel construction module is used to receive and parse the mixed data stream in the remote interaction channel through the edge control unit, separate the corresponding instruction data and interaction data from the mixed data stream, and establish a logically independent control data channel for the instruction data; The priority allocation module is used to allocate a first priority to the instruction data in the control data channel and allocate a second priority to the interactive data through the edge control unit. The synchronization module is used to apply the allocated instruction data to the actuator of the embodied intelligent device, and to collect the motion state data of the actuator in real time as a feedback signal. The log data and running status in the feedback signal are asynchronously synchronized to the cloud collaborative platform as non-control data. The interaction module is used to transmit the interaction data to the remote interaction client through the remote interaction channel, provided that the first priority of the instruction data is not affected. The optimization module is used to automatically reduce the amount of interactive data and the transmission frequency when the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, and to maintain the first priority of the control data channel.

[0014] Compared with the prior art, the present invention has the following beneficial effects: This invention establishes a logically independent control data channel for instruction data and assigns it a first priority higher than that of interactive data. This achieves isolated scheduling of control flow and interactive flow at the levels of transmission path, bus arbitration, and interrupt response, reducing instruction transmission jitter and response latency. This ensures that the actuators of the embedded intelligent device receive hard real-time, deterministic control guarantees, improving the accuracy and security of remote control. Simultaneously, based on network latency detection, it achieves hard real-time time slot interval division and bidirectional handshake synchronization, matching the maximum network carrying period and idle protection interval, reducing transmission conflicts and latency fluctuations, allowing the remote interaction channel to maintain stable low-latency characteristics even in complex network environments.

[0015] This invention employs an asynchronous synchronization mechanism for non-control data, decoupling logs and operational status data from the core control link. This avoids crowding out instruction transmission resources, optimizes overall system latency and link utilization, and automatically reduces interactive data capacity and transmission frequency when network link quality deteriorates. Without affecting the priority of control instructions and transmission reliability, it maintains the normal operation of core control functions and improves the system's anti-interference capability and environmental adaptability. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a low-latency remote interaction and control method for embodied intelligence, provided in an embodiment of the present invention. Figure 2A functional block diagram of a low-latency remote interaction and control system for embodied intelligence provided in an embodiment of the present invention; The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0018] This application provides a low-latency remote interaction and control method for embodied intelligence. The executing entity of this low-latency remote interaction and control method for embodied intelligence includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the low-latency remote interaction and control method for embodied intelligence can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks, and big data and artificial intelligence platforms.

[0019] Reference Figure 1 The diagram shown is a flowchart illustrating a low-latency remote interaction and control method for embodied intelligence according to an embodiment of the present invention. In this embodiment, the low-latency remote interaction and control method for embodied intelligence includes: S1. Install a cloud-based collaborative platform to establish a remote interaction channel between the edge control unit deployed on the smart device side and the remote interactive client through a low-latency communication network.

[0020] In this embodiment, establishing a remote interaction channel between the edge control unit deployed on the embodied smart device side and the remote interaction client via a low-latency communication network includes: The low-latency communication network transmits link probe pulses and receives return confirmation pulses of the transmission results. The one-way transmission delay fluctuation range and round-trip delay difference of the low-latency communication network are determined step by step by the transmit link detection pulse and the return confirmation pulse. By combining the one-way transmission delay fluctuation range and the round-trip delay difference, the uncertainty margin of the low-latency communication network is extracted to obtain the maximum continuous carrying period and the minimum idle protection interval corresponding to the low-latency communication network. The physical resource blocks of the low-latency communication network are divided into hard real-time time slot intervals according to the continuous bearing period and the idle protection interval, so as to obtain the boundary parameter frames of the hard real-time time slot intervals. Based on the boundary parameter frame, a two-way handshake synchronization and asynchronous transmission parameter negotiation are performed on the time slot boundary between the edge control unit and the remote interactive client to obtain a remote interactive channel.

[0021] Specifically, the cloud-based collaborative platform is a service platform deployed in the cloud that provides asynchronous synchronization and collaborative scheduling support for non-control data in remote interaction scenarios of embodied intelligent devices. It is the core carrier for data collaboration between edge control units and remote interactive clients.

[0022] Low-latency communication networks are communication networks that provide high-speed, low-jitter transmission capabilities for remote interaction of embody intelligent devices.

[0023] Embodied intelligent devices are intelligent physical devices with environmental perception and action execution capabilities, whose actuators are directly driven by remotely transmitted control commands.

[0024] The edge control unit is a hardware control unit deployed on the local side of the embodied smart device, and it is the core control node on the device side.

[0025] A remote interactive client is a remote control terminal for operators and is the initiator of remote control commands.

[0026] The remote interaction channel is a dedicated channel built on a low-latency communication network to enable bidirectional data transmission between the edge control unit and the remote interaction client. It is used to carry various data streams required for the remote control of intelligent devices.

[0027] Link probe pulses are special pulse signals transmitted through low-latency communication networks to detect the transmission characteristics of communication links.

[0028] The return confirmation pulse is a feedback pulse signal sent back by the receiving end after the link probe pulse is transmitted, used to confirm the result of the pulse reception.

[0029] The one-way transmission delay fluctuation range is the numerical range between the maximum and minimum values ​​of transmission delay during one-way data transmission in a low-latency communication network, calculated based on the link probe pulse and the return acknowledgment pulse. It can characterize the delay stability of one-way transmission of the link.

[0030] The round-trip delay difference is the total transmission time of the signal from the transmitter to the receiver and back to the transmitter, calculated based on the link probe pulse and the return acknowledgment pulse. It can characterize the overall delay level of the link's round-trip transmission.

[0031] Uncertainty margin is the fluctuation redundancy of transmission delay in low-latency communication networks, which is extracted by combining the one-way transmission delay fluctuation range and the round-trip delay difference.

[0032] The continuous carrying period is the maximum duration during which a low-latency communication network can continuously and stably carry data transmission, calculated based on the uncertainty margin.

[0033] The idle protection interval is the minimum length interval reserved between adjacent data transmission cycles, calculated based on the uncertainty margin.

[0034] Physical resource blocks are the basic resource units used to carry data transmission in the physical layer of low-latency communication networks.

[0035] The hard real-time time slot interval is a time interval with hard real-time transmission guarantee capability obtained by dividing physical resource blocks based on continuous bearing period and idle protection interval.

[0036] Boundary parameter frames are parameter frames generated after the hard real-time time slot interval is divided, and are used to calibrate the start and end boundaries of the hard real-time time slot interval.

[0037] The time slot boundary is the start and end time boundary of the hard real-time time slot interval, and it serves as the time reference for the edge control unit and the remote interactive client to achieve synchronous transmission.

[0038] Two-way handshake synchronization is a two-way time slot boundary time synchronization operation between the edge control unit and the remote interactive client, based on boundary parameter frames, which can ensure the alignment of transmission time slots at both ends.

[0039] Asynchronous transmission parameter negotiation is a bidirectional negotiation operation between the edge control unit and the remote interactive client to handle non-real-time data transmission parameters based on time slot synchronization.

[0040] Preferably, a link probe pulse is first transmitted to the other end through a low-latency communication network, while the communication link is continuously monitored and the return acknowledgment pulse returned by the other end in response to the probe pulse is received, thus completing the complete transmission and reception operation of the pulse signal. Based on the transmitted link probe pulses and the received return confirmation pulses, the one-way transmission delay fluctuation range and the round-trip delay difference of the low-latency communication network are determined. Then, combined with the calculated one-way transmission delay fluctuation range and the round-trip delay difference, an extraction operation is performed to obtain the uncertainty margin of the low-latency communication network. Based on this uncertainty margin, further calculations are performed to obtain the maximum continuous carrying period and the minimum idle protection interval that the low-latency communication network can support. Subsequently, the physical resource blocks of the low-latency communication network are divided into hard real-time time slot intervals. After the division is completed, boundary parameter frames that can be used to define the start and end boundaries of the hard real-time time slot intervals are generated synchronously. Finally, based on the generated boundary parameter frame, a two-way handshake synchronization is performed on the time slot boundary between the edge control unit and the remote interactive client to ensure that the transmission time slots at both ends are fully aligned, and the remote interactive channel is finally established.

[0041] S2. The edge control unit receives and parses the mixed data stream in the remote interaction channel, separates the corresponding instruction data and interaction data from the mixed data stream, and establishes a logically independent control data channel for the instruction data.

[0042] In this embodiment, receiving and parsing the mixed data stream in the remote interaction channel through the edge control unit includes: The physical transmission medium of the remote interaction channel is collected to obtain the original electrical signal sequence captured by the physical layer interface corresponding to the edge control unit; A serial bit stream is obtained by synchronizing the original electrical signal sequence with clock edges and performing bit decisions. The serial bit stream is parsed and its payload is extracted to obtain the frame payload data block sequence of the serial bit stream; The consecutive data blocks in the frame payload data block sequence are concatenated end to end, and the frame delimiter and check field in the concatenation result are removed to obtain the mixed data stream of the remote interaction channel.

[0043] Specifically, instruction data is the core control data used to drive the actuators of embodied intelligent devices to complete specified actions.

[0044] Interactive data is non-control data used for remote status display and environmental perception feedback.

[0045] The control data channel is a logically independent transmission channel specifically established for instruction data. It can share the physical transmission link with interactive data but achieves isolation through transmission arbitration.

[0046] The physical transmission medium is the physical carrier used to transmit electrical signals at the bottom layer of a remote interactive channel, including types such as wired transmission media or wireless transmission channels.

[0047] The raw electrical signal sequence is an ordered collection of raw analog electrical signals that are directly captured from the physical transmission medium by the edge control unit through the physical layer interface without any processing.

[0048] Clock edge synchronization is a clock alignment operation performed on the original electrical signal sequence to eliminate clock skew during signal transmission.

[0049] Bit decision is a numerical determination operation performed on the original electrical signal sequence after clock edge synchronization. It converts the continuously changing analog electrical signal into discrete binary bit values, realizing the conversion from analog signal to digital signal.

[0050] A serial bit stream is a stream of binary bits arranged in chronological order, obtained after clock edge synchronization and bit decision operations.

[0051] Frame parsing is a frame structure disassembly operation performed on a serial bit stream. According to a preset communication data frame format, it identifies the start and end positions and internal structure of complete data frames in the serial bit stream.

[0052] Payload extraction is based on frame parsing. It involves removing auxiliary fields of the frame structure from the decomposed data frame and extracting the effective data segment that actually carries the business content.

[0053] The frame payload data block sequence is an ordered set of multiple payload data blocks arranged in the data frame transmission order, obtained after frame parsing and payload extraction operations.

[0054] Beginning and end splicing is a splicing operation performed on consecutive data blocks in the frame payload data block sequence. According to the transmission order of the data blocks, the tail of the previous data block is seamlessly connected to the head of the next data block, integrating the scattered single frame data blocks into a continuous and complete data segment.

[0055] The frame delimiter is a specific identifier field embedded in the data frame structure, used to mark the start and end positions of a single data frame. It serves as the boundary identification identifier for the data frame.

[0056] The validation field is a specific field embedded in the data frame structure used to verify whether errors have occurred during data transmission.

[0057] In detail, firstly, the edge control unit captures the analog electrical signals transmitted on the transmission medium in real time through its corresponding physical layer interface to obtain the original electrical signal sequence. The edge control unit accurately aligns the clock beat of the signal transmission to eliminate the clock offset generated during transmission. The edge control unit performs segment-by-segment bit decision on the original electrical signal sequence, converting the continuously changing analog electrical signals into discrete binary bit values. Then, it identifies and disassembles each complete data frame in the serial bit stream according to the preset communication frame format, accurately locating the structural boundary of each data frame. By removing non-business auxiliary content from the data frames, the core effective data segments carried in each data frame are extracted, and finally a sequence of frame payload data blocks arranged in the data frame transmission order is obtained. Finally, the scattered single-frame payload data blocks are seamlessly integrated into continuous complete data segments according to the transmission sequence. All frame delimiters and check fields in the splicing result are completely removed, and the complete mixed data stream transmitted in the remote interactive channel is finally parsed.

[0058] In this embodiment, separating the corresponding instruction data and interaction data from the mixed data stream and establishing a logically independent control data channel for the instruction data includes: A sliding window energy integration is performed on consecutive bits in the mixed data stream to obtain the bit energy distribution sequence of the mixed data stream on the time axis; Determine the alternating boundaries between local energy valleys and local energy peaks in the bit energy distribution sequence; The mixed data stream is segmented into bit segments according to the alternating boundaries to obtain constant energy bit segments; Determine the duration of the constant energy bit segment, and calculate the length variance of the length sequence corresponding to the constant energy bit segment based on the duration; The constant energy bit segments with length variance smaller than the overall length variance of the mixed data stream are selected to obtain a sequence of instruction data frame segments that arrive stably in a periodic manner. The instruction data frame sequence is physically mapped from the mixed data stream to the first-in-first-out buffer queue of the edge control unit; Assign an independent bus arbitration identifier and interrupt request line to the first-in-first-out buffer queue to obtain a control data channel that shares a physical link with the interactive data but is isolated from the arbitration path.

[0059] Specifically, consecutive bits are a set of binary bits arranged sequentially in time within a mixed data stream.

[0060] Sliding window energy integration is a method for calculating the signal energy of consecutive bits. It involves sliding a fixed window along the time axis to traverse the bits and accumulating the signal energy within each window.

[0061] The bit energy distribution sequence is an ordered sequence obtained by sliding window energy integration, reflecting the signal energy distribution state of the mixed data stream at each position on the time axis.

[0062] A local energy valley is a point in the bit energy distribution sequence where the energy value is in a local minimum range.

[0063] A local energy peak is a point in the bit energy distribution sequence where the energy value is in a local range of maximum.

[0064] Alternating boundaries are the points where local energy valleys and local energy peaks connect in a bit energy distribution sequence.

[0065] Bit segmentation is a segmentation operation performed on a mixed data stream based on alternating boundaries, dividing the data stream into multiple independent bit segments.

[0066] A constant energy bit segment is a bit segment whose signal energy distribution remains stable after bit segmentation.

[0067] The instruction data frame sequence is a set of constant energy bit segments that have been selected and arrive periodically and stably.

[0068] The first-in-first-out (FIFO) buffer queue is a buffer queue within the edge control unit used to temporarily store instruction data frames, and it stores instruction data according to the first-in-first-out principle.

[0069] The bus arbitration identifier is a unique identifier assigned to the first-in-first-out buffer queue, used to distinguish the transmission rights of instruction data during bus transmission.

[0070] The interrupt request line is a dedicated signal line configured for the first-in-first-out buffer queue, used to send transmission requests for instruction data to the control module of the edge control unit.

[0071] In detail, for consecutive bits in the mixed data stream, a sliding window energy integration is performed by sliding along the time axis with a fixed window, accumulating the signal energy within each window, and generating a bit energy distribution sequence that reflects the energy distribution of the data stream's time axis; Next, the alternating boundaries of local energy valleys and local energy peaks in the sequence are located, and bit segmentation is performed on the mixed data stream based on these boundaries to obtain multiple constant energy bit segments; Then, the duration of each constant energy bit segment is determined, the length variance of its corresponding length sequence is calculated, and the variance is compared with the overall length variance of the mixed data stream. Constant energy bit segments with length variances smaller than the overall length variance are selected and integrated to obtain a sequence of instruction data segments that arrive periodically and stably. The instruction data frame sequence is then physically mapped from the mixed data stream to the first-in-first-out (FIFO) buffer queue of the edge control unit for temporary storage; finally, an independent bus arbitration flag and interrupt request line are assigned to the FIFO buffer queue.

[0072] S3. The edge control unit assigns a first priority to the instruction data in the control data channel and a second priority to the interaction data.

[0073] In this embodiment, the step of assigning a first priority to the instruction data in the control data channel and assigning a second priority to the interaction data through the edge control unit includes: Analyze the difference between the arrival time of the instruction data and its adjacent data in the control data channel, and take the absolute value of the difference to obtain the instruction jitter amplitude sequence of the instruction data; Analyze the difference between the arrival times of the interactive data and the adjacent interactive data, and take the absolute value of all differences to obtain the interaction jitter amplitude sequence of the interactive data; When the maximum value in the instruction jitter amplitude sequence is less than the minimum value in the interaction jitter amplitude sequence, the instruction data is marked as deterministic periodic data and assigned the first priority; otherwise, the instruction data is marked as an abnormal stream and remains in an unassigned priority state. Based on the first priority already obtained from the instruction data, a second priority is assigned to the interactive data; Set the interrupt request response latency limit bound to the first priority to one-third of the interrupt request response latency limit bound to the second priority.

[0074] In this embodiment, the difference in arrival times between the command data and its adjacent data in the control data channel is analyzed. in: The difference. For indexing, For the first The local hardware timestamp of the instruction data arriving at the edge control unit. The nominal transmission period for the instruction data. The frame length influence coefficient. For the first The total frame length of each instruction data This refers to the physical layer line bit rate.

[0075] Specifically, by subtracting the nominal sending period This highlights the deviation of the actual interval from the expected value, which better reflects the physical meaning of jitter. Secondly, it introduces an additional frame length difference term. This corrects the serialization latency differences caused by changes in data packet size, a factor that is ignored in the traditional definition of jitter but cannot be ignored in remote interaction scenarios.

[0076] Specifically, the first priority is the highest transmission priority specifically allocated to instruction data, used to ensure the hard real-time, deterministic transmission requirements of instruction data.

[0077] The second priority is the second highest transmission priority allocated to interactive data, which is lower than the priority of instruction data. It is used to ensure the stable transmission of interactive data but does not preempt instruction data resources.

[0078] The instruction jitter amplitude sequence is an ordered sequence that reflects the fluctuation range of the instruction data arrival time by calculating the difference between the arrival time of the instruction data and the arrival time of the adjacent instruction data and taking the absolute value.

[0079] Interactive jitter amplitude sequence is an ordered sequence that reflects the fluctuation range of the arrival time of interactive data by calculating the difference between the arrival time of interactive data and the arrival time of adjacent interactive data and taking the absolute value.

[0080] Deterministic periodic data is instruction data that satisfies the condition of minimal fluctuation at arrival time. Its transmission has predictable and stable periodic characteristics, making it the first priority data type to adapt.

[0081] Abnormal streams are instruction data that do not meet the conditions for deterministic periodic data, and their transmission time fluctuates greatly, so they are not assigned priority for the time being.

[0082] The interrupt request response delay limit is the maximum allowable delay time for the edge control unit to respond to command data or interactive data transmission requests after receiving an interrupt request.

[0083] In detail, the arrival time of each instruction data and its adjacent instruction data is obtained one by one, the time difference between the two is calculated and the absolute value of all differences is taken. At the same time, for each piece of interaction data, the arrival time of its adjacent interaction data is obtained one by one, the time difference is calculated and the absolute value is taken, and the acquisition and generation of the interaction jitter amplitude sequence is completed. Next, the maximum value in the instruction jitter amplitude sequence is compared with the minimum value in the interactive jitter amplitude sequence. If the maximum value in the instruction jitter amplitude sequence is less than the minimum value in the interactive jitter amplitude sequence, the instruction data is marked as deterministic periodic data and assigned the first priority. If this condition is not met, the instruction data is marked as an abnormal stream and remains in an unassigned priority state. After completing the priority allocation of instruction data, based on the first priority already obtained by the instruction data, a second priority is allocated to the interactive data. The upper limit of the interrupt request response delay bound to the first priority is set to one-third of the upper limit of the interrupt request response delay bound to the second priority. Finally, the priority differentiation allocation and delay threshold configuration of instruction data and interactive data are completed.

[0084] S4. Apply the allocated instruction data to the actuator of the embodied intelligent device, and collect the motion state data of the actuator in real time as a feedback signal. Asynchronously synchronize the log data and running status in the feedback signal to the cloud collaborative platform as non-control data.

[0085] In this embodiment, the step of applying the allocated instruction data to the actuator of the embodied intelligent device and collecting the motion state data of the actuator in real time as a feedback signal includes: Obtain the instruction codeword from the allocated instruction data; The instruction codewords are sequentially written into the hold latch of the digital-to-analog converter register in the edge control unit to obtain the analog voltage control waveform of the actuator; Motion state data is obtained by applying the analog voltage control waveform to the winding drive input terminal of the actuator; The motion state data is converted into a feedback signal for the actuator.

[0086] Specifically, motion state data refers to the motion state-related data that the actuator provides in real time after executing the command, including state information such as position, speed, and torque.

[0087] The instruction codeword is the core encoded field in the instruction data that carries specific control instruction information.

[0088] The digital-to-analog converter register is a register in the edge control unit used to convert digital instruction codewords into analog control signals.

[0089] A latch is a storage component integrated into the digital-to-analog converter register. It is used to latch instruction codewords to ensure stable signal output during digital-to-analog conversion and to prevent abnormal control signals caused by instruction data fluctuations.

[0090] The analog voltage control waveform is an analog voltage signal waveform generated by the instruction code word through digital-to-analog conversion and latching.

[0091] The winding drive input terminal is the hardware interface in the actuator used to input analog voltage control waveforms. It is the channel through which control signals are transmitted to the actuator, determining the actuator's operating state.

[0092] In detail, the instruction codewords carrying the core control instructions are first extracted from the instruction data with assigned priorities to clarify the action instruction type of the actuator; The extracted instruction codewords are then written sequentially into the hold latch of the digital-to-analog converter register in the edge control unit. The instruction codewords are stabilized by the latching function of the hold latch, and then the digital-to-analog converter register completes the conversion of digital instructions into analog signals. Next, the analog voltage control waveform is precisely applied to the winding drive input terminal of the actuator, and the control signal is transmitted to the actuator through the hardware interface to drive the actuator to perform the corresponding action; The collected motion state data is then converted into a new format to generate a feedback signal representing the action state of the actuator.

[0093] In this embodiment, asynchronously synchronizing the log data and operating status in the feedback signal as non-control data to the cloud collaboration platform includes: The log data segment and the running status data segment are parsed from the feedback signal; The log data segment and the running status data segment are written to the edge control unit in the order of their generation time to obtain a non-control data cache queue; A globally incrementing sequence number is appended to each data block in the non-control data buffer queue, and the data is encapsulated into an independent data frame to obtain an asynchronous synchronous data frame sequence. The asynchronous synchronous data frame sequence is sent to the network port of the cloud collaboration platform.

[0094] Specifically, non-control data consists of log data segments and running status data segments. It does not participate in the real-time action control of the actuators of the embodied intelligent device and does not preempt the transmission resources of core control commands.

[0095] Asynchronous synchronization is a synchronization method that is not bound to the transmission timing of core control commands, does not preempt control data channel resources, and uploads non-control data from the device edge to the cloud on demand.

[0096] The non-control data cache queue is a special first-in-first-out cache queue within the edge control unit that stores log data segments and running status data segments in the order they were generated. It is used to temporarily store non-control data that needs to be synchronized.

[0097] A data block is an independent data unit divided into data segments within a non-control data cache queue.

[0098] The globally incrementing sequence number is a globally unique and sequentially increasing numerical identifier assigned to each data block. It is used by the cloud-based collaborative platform to verify the integrity, timing accuracy, and continuity of received data.

[0099] A data frame is an independent transmission unit obtained by encapsulating a data block with an appended globally incrementing sequence number according to a preset communication transmission format.

[0100] An asynchronous synchronous data frame sequence is an ordered set composed of multiple encapsulated independent data frames arranged in the order in which the data was generated.

[0101] A network port is a dedicated network communication interface pre-configured on a cloud-based collaborative platform for receiving asynchronous and synchronous data frame sequences.

[0102] In detail, the field boundaries within the feedback signal are identified, and the log data segment and running status data segment carried in the feedback signal are split and extracted to complete the separation and extraction of the two types of non-control data. Subsequently, according to the generation time of the log data segment and the running status data segment, following the chronological order from early to late, the two types of data segments are sequentially written into the dedicated cache area opened in the edge control unit, generating a non-control data cache queue arranged in time sequence, and completing the orderly temporary storage of non-control data to be synchronized; For each independent data block, a globally unique and continuously increasing sequence number is assigned sequentially. Then, according to the preset network communication transmission format, each data block with the attached sequence number is encapsulated into an independent data frame that can be directly used for network transmission, and finally, an asynchronous synchronous data frame sequence arranged according to the data time sequence is generated. Finally, without preempting core control command transmission resources or affecting the priority of control data channels, the encapsulated asynchronous and synchronous data frame sequence is sent to the pre-configured dedicated network port of the cloud collaboration platform through a low-latency communication network.

[0103] S5. Under the premise of ensuring that the first priority of the instruction data is not affected, the interaction data is transmitted to the remote interaction client through the remote interaction channel.

[0104] In this embodiment, transmitting the interaction data to the remote interaction client through the remote interaction channel while ensuring that the first priority of the instruction data is not affected includes: The depth of the transmission queue of the remote interaction channel is detected to obtain the fixed transmission duration; The minimum idle time interval between two adjacent instruction data is analyzed based on the fixed transmission duration. The window length is determined based on the minimum idle time interval. The interactive data is divided into data blocks that do not exceed the window length, and the second priority is attached to each data block to obtain an interactive data fragment sequence; The interactive data fragment sequence is transmitted to the remote interactive client.

[0105] Specifically, the sending queue depth is the length of the data to be sent accumulated in the transmission queue within the remote interaction channel, used to characterize the current transmission load of the channel.

[0106] The fixed transmission duration is the time required for a single fixed-length data set to complete its full transmission, calculated based on the transmission queue depth of the remote interactive channel.

[0107] Adjacent instruction data refers to two independent instruction data that are immediately adjacent in time during the transmission sequence of the remote interactive channel.

[0108] The minimum idle time interval is the shortest time window between two adjacent instruction data transmissions that is not occupied by instruction data and can be used to transmit other data.

[0109] The window length is a threshold for the maximum data length that can be transmitted in a single block of interactive data, determined based on the minimum idle time interval.

[0110] An interactive data fragment sequence consists of multiple fragmented interactive data blocks with an additional second priority.

[0111] In detail, firstly, the edge control unit performs real-time detection of the sending queue depth of the remote interaction channel, and calculates the fixed sending duration of a single set of data based on the detected queue load. Subsequently, based on the calculated fixed transmission duration, the transmission timing of two adjacent instruction data in the remote interaction channel is analyzed to accurately locate the minimum idle time interval between two adjacent instruction data that is not occupied by instruction transmission, ensuring that this interval is completely within the instruction data transmission gap and will not interfere with the normal transmission of instructions. Next, the maximum window length that can be transmitted for a single block of interactive data is determined to ensure that the single block of data can be transmitted completely within the idle interval. Then, the edge control unit performs a segmentation operation on the complete interactive data, dividing it into multiple independent data blocks, none of which exceed the window length. At the same time, a second priority identifier is attached to each segmented data block to generate an interactive data fragment sequence arranged according to the original data time sequence.

[0112] S6. When the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, it automatically reduces the amount of interactive data and the transmission frequency, and maintains the first priority of the control data channel.

[0113] In this embodiment, when the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, automatically reducing the amount of interactive data and the transmission frequency includes: The instantaneous value of the signal strength of the currently received signal in the low-latency communication network is collected, and the instantaneous value of the signal strength is compared with a preset threshold to obtain a judgment flag; When the determination flag is valid, the maximum capacity of the sending buffer for the interactive data is reduced to half of the original capacity, and the sending interval is extended to twice the original interval.

[0114] Specifically, link quality is a comprehensive indicator that characterizes the transmission stability, signal strength, and transmission reliability of low-latency communication networks.

[0115] The preset threshold is a critical value pre-set within the edge control unit to determine whether the quality of the low-latency communication network link has deteriorated; it serves as the baseline for link quality judgment.

[0116] like Figure 2 The diagram shown is a functional block diagram of a low-latency remote interaction and control system for embodied intelligence provided in an embodiment of the present invention.

[0117] The low-latency remote interaction and control system 100 for embodied intelligence described in this invention can be installed in an electronic device. Depending on the functions implemented, the low-latency remote interaction and control system 100 for embodied intelligence may include a construction module 101, a channel construction module 102, a priority allocation module 103, a synchronization module 104, an interaction module 105, and an optimization module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0118] In this embodiment, the functions of each module / unit are as follows: Module 101 is used to house a cloud-based collaborative platform, which establishes a remote interaction channel between the edge control unit deployed on the embodied smart device and the remote interactive client through a low-latency communication network. Channel construction module 102 is used to receive and parse the mixed data stream in the remote interaction channel through the edge control unit, separate the corresponding instruction data and interaction data from the mixed data stream, and establish a logically independent control data channel for the instruction data; Priority allocation module 103 is used to allocate a first priority to the instruction data in the control data channel and allocate a second priority to the interactive data through the edge control unit; Synchronization module 104 is used to apply the allocated instruction data to the actuator of the embodied intelligent device, and to collect the motion state data of the actuator in real time as a feedback signal, and to asynchronously synchronize the log data and running status in the feedback signal as non-control data to the cloud collaborative platform. Interaction module 105 is used to transmit the interaction data to the remote interaction client through the remote interaction channel, provided that the first priority of the instruction data is not affected. The optimization module 106 is used to automatically reduce the amount of interactive data and the transmission frequency when the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, and to maintain the first priority of the control data channel.

[0119] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0120] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0121] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0122] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0123] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A low-latency remote interaction and control method for embodied intelligence, characterized in that, The method includes: S1. Install a cloud-based collaborative platform to establish a remote interaction channel between the edge control unit deployed on the smart device side and the remote interaction client through a low-latency communication network; S2. The edge control unit receives and parses the mixed data stream in the remote interaction channel, separates the corresponding instruction data and interaction data from the mixed data stream, and establishes a logically independent control data channel for the instruction data. S3. The edge control unit assigns a first priority to the instruction data in the control data channel and a second priority to the interactive data. S4. Apply the allocated instruction data to the actuator of the embodied intelligent device, and collect the motion state data of the actuator in real time as a feedback signal. Asynchronously synchronize the log data and running status in the feedback signal as non-control data to the cloud collaborative platform. S5. Under the premise of ensuring that the first priority of the instruction data is not affected, transmit the interaction data to the remote interaction client through the remote interaction channel; S6. When the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, it automatically reduces the amount of interactive data and the transmission frequency, and maintains the first priority of the control data channel.

2. The low-latency remote interaction and control method for embodied intelligence as described in claim 1, characterized in that, The establishment of a remote interaction channel between the edge control unit deployed on the embodied smart device and the remote interaction client via a low-latency communication network includes: The low-latency communication network transmits link probe pulses and receives return confirmation pulses of the transmission results. The one-way transmission delay fluctuation range and round-trip delay difference of the low-latency communication network are determined step by step by the transmit link detection pulse and the return confirmation pulse. By combining the one-way transmission delay fluctuation range and the round-trip delay difference, the uncertainty margin of the low-latency communication network is extracted to obtain the maximum continuous carrying period and the minimum idle protection interval corresponding to the low-latency communication network. The physical resource blocks of the low-latency communication network are divided into hard real-time time slot intervals according to the continuous bearing period and the idle protection interval, so as to obtain the boundary parameter frames of the hard real-time time slot intervals. Based on the boundary parameter frame, a two-way handshake synchronization and asynchronous transmission parameter negotiation are performed on the time slot boundary between the edge control unit and the remote interactive client to obtain a remote interactive channel.

3. The low-latency remote interaction and control method for embodied intelligence as described in claim 1, characterized in that, The process of receiving and parsing the mixed data stream in the remote interaction channel through the edge control unit includes: The physical transmission medium of the remote interaction channel is collected to obtain the original electrical signal sequence captured by the physical layer interface corresponding to the edge control unit; A serial bit stream is obtained by synchronizing the original electrical signal sequence with clock edges and performing bit decisions. The serial bit stream is parsed and its payload is extracted to obtain the frame payload data block sequence of the serial bit stream; The consecutive data blocks in the frame payload data block sequence are concatenated end to end, and the frame delimiter and check field in the concatenation result are removed to obtain the mixed data stream of the remote interaction channel.

4. The low-latency remote interaction and control method for embodied intelligence as described in claim 3, characterized in that, The step of separating the corresponding instruction data and interaction data from the mixed data stream and establishing a logically independent control data channel for the instruction data includes: A sliding window energy integration is performed on consecutive bits in the mixed data stream to obtain the bit energy distribution sequence of the mixed data stream on the time axis; Determine the alternating boundaries between local energy valleys and local energy peaks in the bit energy distribution sequence; The mixed data stream is segmented into bit segments according to the alternating boundaries to obtain constant energy bit segments; Determine the duration of the constant energy bit segment, and calculate the length variance of the length sequence corresponding to the constant energy bit segment based on the duration; The constant energy bit segments with length variance smaller than the overall length variance of the mixed data stream are selected to obtain a sequence of instruction data frame segments that arrive stably in a periodic manner. The instruction data frame sequence is physically mapped from the mixed data stream to the first-in-first-out buffer queue of the edge control unit; Assign an independent bus arbitration identifier and interrupt request line to the first-in-first-out buffer queue to obtain a control data channel that shares a physical link with the interactive data but is isolated from the arbitration path.

5. A low-latency remote interaction and control method for embodied intelligence as described in claim 1, characterized in that, The step of assigning a first priority to the instruction data in the control data channel and assigning a second priority to the interaction data through the edge control unit includes: Analyze the difference between the arrival time of the instruction data and its adjacent data in the control data channel, and take the absolute value of the difference to obtain the instruction jitter amplitude sequence of the instruction data; Analyze the difference between the arrival times of the interactive data and the adjacent interactive data, and take the absolute value of all differences to obtain the interaction jitter amplitude sequence of the interactive data; When the maximum value in the instruction jitter amplitude sequence is less than the minimum value in the interaction jitter amplitude sequence, the instruction data is marked as deterministic periodic data and assigned the first priority; otherwise, the instruction data is marked as an abnormal stream and remains in an unassigned priority state. Based on the first priority already obtained from the instruction data, a second priority is assigned to the interactive data; Set the interrupt request response latency limit bound to the first priority to one-third of the interrupt request response latency limit bound to the second priority.

6. The low-latency remote interaction and control method for embodied intelligence as described in claim 5, characterized in that, The step of applying the allocated instruction data to the actuator of the embodied intelligent device and collecting the motion state data of the actuator in real time as a feedback signal includes: Obtain the instruction codeword from the allocated instruction data; The instruction codewords are sequentially written into the hold latch of the digital-to-analog converter register in the edge control unit to obtain the analog voltage control waveform of the actuator; Motion state data is obtained by applying the analog voltage control waveform to the winding drive input terminal of the actuator; The motion state data is converted into a feedback signal for the actuator.

7. A low-latency remote interaction and control method for embodied intelligence as described in claim 1, characterized in that, The step of asynchronously synchronizing the log data and operating status in the feedback signal as non-control data to the cloud collaboration platform includes: The log data segment and the running status data segment are parsed from the feedback signal; The log data segment and the running status data segment are written to the edge control unit in the order of their generation time to obtain a non-control data cache queue; A globally incrementing sequence number is appended to each data block in the non-control data buffer queue, and the data is encapsulated into an independent data frame to obtain an asynchronous synchronous data frame sequence. The asynchronous synchronous data frame sequence is sent to the network port of the cloud collaboration platform.

8. A low-latency remote interaction and control method for embodied intelligence as described in claim 7, characterized in that, The step of transmitting the interaction data to the remote interaction client through the remote interaction channel, while ensuring that the first priority of the instruction data is not affected, includes: The depth of the transmission queue of the remote interaction channel is detected to obtain the fixed transmission duration; The minimum idle time interval between two adjacent instruction data is analyzed based on the fixed transmission duration. The window length is determined based on the minimum idle time interval. The interactive data is divided into data blocks that do not exceed the window length, and the second priority is attached to each data block to obtain an interactive data fragment sequence; The interactive data fragment sequence is transmitted to the remote interactive client.

9. A low-latency remote interaction and control method for embodied intelligence as described in claim 1, characterized in that, When the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, it automatically reduces the amount of interactive data and the transmission frequency, including: The instantaneous value of the signal strength of the currently received signal in the low-latency communication network is collected, and the instantaneous value of the signal strength is compared with a preset threshold to obtain a judgment flag; When the determination flag is valid, the maximum capacity of the sending buffer for the interactive data is reduced to half of the original capacity, and the sending interval is extended to twice the original interval.

10. A low-latency remote interaction and control system for embodied intelligence, used to implement the low-latency remote interaction and control method for embodied intelligence as described in any one of claims 1-9, characterized in that, The system includes: The building module is used to house the cloud collaboration platform and establishes a remote interaction channel between the edge control unit deployed on the smart device side and the remote interaction client through a low-latency communication network; The channel construction module is used to receive and parse the mixed data stream in the remote interaction channel through the edge control unit, separate the corresponding instruction data and interaction data from the mixed data stream, and establish a logically independent control data channel for the instruction data; The priority allocation module is used to allocate a first priority to the instruction data in the control data channel and allocate a second priority to the interactive data through the edge control unit. The synchronization module is used to apply the allocated instruction data to the actuator of the embodied intelligent device, and to collect the motion state data of the actuator in real time as a feedback signal. The log data and running status in the feedback signal are asynchronously synchronized to the cloud collaborative platform as non-control data. The interaction module is used to transmit the interaction data to the remote interaction client through the remote interaction channel, provided that the first priority of the instruction data is not affected. The optimization module is used to automatically reduce the amount of interactive data and the transmission frequency when the edge control unit detects that the link quality of the low-latency communication network is lower than a preset threshold, and to maintain the first priority of the control data channel.

Images