Humanoid robot-oriented interconnection communication system and method based on machine learning

Abstract

The invention discloses a humanoid robot-oriented interconnection communication system and method based on machine learning, and relates to the field of robot communication. Through collaborative integration of visible light and infrared light communication facilities, the high bandwidth characteristic of visible light communication is reserved, the interference problem in a complex environment is solved by using infrared light, and the communication efficiency is improved. Complementarity communication guarantee is formed; by means of the dual-band design, the anti-electromagnetic interference capacity in an industrial scene is remarkably improved, and time delay can be controlled at the microsecond level; based on correlation analysis of filter parameters of historical task data and damping of a damping bracket, an external cause-success rate mapping model is constructed, so that a robot can dynamically adjust mechanical and communication parameters according to real-time working conditions; traditional fixed threshold value control is upgraded into self-adaptive adjustment, and the method is suitable for a high-precision control scene; through combined adjustment of filter parameters and damping of the damping support, limitation of single module optimization is broken through.

Description

Technical Field

[0001] The present invention belongs to the field of robot communication, and more specifically, relates to a humanoid robot interconnection communication system and method based on machine learning. Background Art

[0002] The single communication mode used by traditional humanoid robots has signal attenuation problems under complex working conditions such as electromagnetic interference and occlusion; and the data transmission rate of 5G and broadband WIFI used by traditional humanoid robots is slow; in addition, the traditional model with fixed parameters leads to a large number of trial and error, which in turn causes response delays. Summary of the Invention

[0003] In response to the problems in the related art, the present invention proposes a humanoid robot interconnection communication system and method based on machine learning to overcome the above-mentioned technical problems existing in the existing related art.

[0004] To solve the above technical problems, the present invention is achieved through the following technical solutions: The present invention is a humanoid robot interconnection communication method based on machine learning, comprising the following steps: S1. Integrating visible light and infrared light network communication facilities into the robot to be communicated with, thereby obtaining an integrated robot to be communicated with; S2. Collect filter parameters, silicone shock-absorbing bracket damping data, and task execution feature data of the integrated communication robot during multiple experiments in history; S3, by collecting the task execution success rate data during multiple experiments in S2, and constructing the external cause task success rate mapping model using the filter parameters, damping data and feature data; S4, collecting filter parameters, damping data of the silicone shock-absorbing bracket, and characteristic data of the current task when executing the current task, and inputting them into the external cause task success rate mapping model for mapping; S5. Adjust the current filter parameters according to the comparison result of the mapping result in S4 and the preset threshold, and repeat S4 and S5 until the first maximum number of repetitions is reached; if the comparison result still does not meet the condition, enter S6; otherwise, enter S7; S6. Coordinately adjust the parameters of the current filter and the shock-absorbing bracket, and repeat S4, S5, and S6 until the comparison result in S5 meets the conditions; S7. Pre-adjust the filter parameters and damping data when executing the current task according to the adjustment result in S5 or S6.

[0005] Preferably, the S1 comprises the following steps: S11. Setting several optical communication modes of humanoid robots to obtain a set of robot optical communication modes; the set of robot optical communication modes includes visible light communication and infrared light communication; S12. A humanoid robot to be communicated with using an optical network is set, denoted as the robot to be communicated; a photoelectric sensor is integrated into the robot to be communicated according to the robot optical communication method set, and LED light sources are evenly arranged around the activity space of the robot to be communicated; a plurality of active joints of the robot to be communicated are set to obtain a set of active joints to be communicated; an infrared transmitting module and an infrared receiving module are installed at the active joints of the robot to be communicated according to the set of active joints to be communicated, to obtain an integrated robot to be communicated; The design of separate visible light and infrared communication bands effectively avoids congestion in a single frequency band. While visible light is used for high-speed data transmission, the infrared channel can simultaneously execute obstacle avoidance commands, ensuring dual-channel parallel operation without conflict. The wide-area deployment of LED light sources combined with the localized sensing of the joint infrared modules achieves layered coverage of "global communication + local obstacle avoidance." The infrared transceiver modules at the joints detect obstacle distances using nanosecond pulses, forming a closed-loop control loop with the central processor.

[0006] Preferably, said S2 comprises the following steps: S21, setting several task types and corresponding feature types of the integrated robots to be communicated, to obtain a robot task type set and a robot task feature type set; S22. Based on the robot task type, setting an experiment in which the integrated robot to be communicated repeatedly performs different types of tasks; during each experiment, installing an acousto-optic filter at the photoelectric sensor and infrared receiving module of the integrated robot to be communicated, and installing a silicone shock-absorbing bracket at its joints; setting several adjustable parameter types of the acousto-optic filter to obtain a filter parameter type set; S23. Based on the robot task feature type set and the filter parameter type set, collect the feature data of each experimental task in S22, the damping data of the silicone shock-absorbing bracket, and the adjustable parameter data of the acousto-optic filter to be installed on the communication robot after integration, to obtain a historical task feature dataset, a historical shock-absorbing bracket damping dataset, and a historical filter parameter set; Dynamic adjustment of the RF power and stopband attenuation ratio of the acousto-optic filter can effectively suppress electromagnetic interference and optical noise in industrial environments; historical data sets provide customized filtering strategies for different task types by quantitatively analyzing the mapping relationship between task characteristics and filter parameters; in addition, by collecting historical task feature data sets and historical filter parameter sets, data support is provided for the subsequent construction of a mapping model between the adjustable parameters of the acousto-optic filter and the task execution success rate.

[0007] Preferably, the step S3 includes the following steps: S31, collecting the execution success rate data of each type of task performed multiple times in S22 to obtain a historical task execution success rate data set; S32, constructing a mapping model between the task execution success rate and the task feature data, the shock absorber bracket damping data, and the filter parameter according to the historical task execution success rate dataset, the historical task feature dataset, the historical shock absorber bracket damping data set, and the historical filter parameter set, to obtain an external cause task success rate mapping model; The mapping model can recommend the optimal parameter combination in real time by analyzing the coupling relationship between historical task characteristics and filter parameters; The model supports generalized parameter recommendations for differentiated scenarios such as high-risk environments and multi-machine collaboration, improving the stability of mission execution.

[0008] Preferably, the external task success rate mapping model in S32 adopts an integrated learning framework model; The integrated learning framework model has greatly improved the generalization performance across task types compared to a single model.

[0009] Preferably, the S4 comprises the following steps: S41. Collect, based on the robot task type set, the robot task feature type set, and the filter parameter type set, adjustable parameter data of the acousto-optic filter installed on the integrated communication robot when it performs the current task, damping data of the silicone shock-absorbing bracket, and feature data of the current task, to obtain a current filter parameter dataset, a current shock-absorbing bracket damping dataset, and a current task feature dataset; S42, inputting the current filter parameter data set, the current shock-absorbing bracket damping data set, and the current task feature data set into the external cause task success rate mapping model for mapping, to obtain current task success rate data; By dynamically collecting current task features and filter parameters and combining them with a pre-trained success rate mapping model, the system can adjust parameter combinations in real time, improving the stability of task execution. When the predicted success rate is lower than the threshold, the obstacle avoidance strategy switch or filter parameter adjustment can be triggered in advance to reduce the sudden failure rate.

[0010] Preferably, the S5 comprises the following steps: S51, setting a current task success rate threshold; if the current task success rate data is less than the current task success rate threshold, adjusting various filter parameter data in the current filter parameter data set to obtain a first current adjusted filter parameter data set; otherwise, no adjustment is required; S52, setting a first maximum number of repetitions and an adjustment time threshold; replacing the current filter parameter data set with the first current adjusted filter parameter data set and repeating S42 and S51; if the number of repetitions is greater than or equal to the first maximum number of repetitions, stopping the repetition; after stopping the repetition, if the current task success rate data is less than the current task success rate threshold, proceeding to S61; otherwise, the adjustment is completed, obtaining the first current final filter parameter data set, and proceeding to S71; By prioritizing the adjustment of the RF power and the stopband attenuation ratio, since the quality of the incident light signal in optical communication has a decisive influence on the accuracy of the humanoid robot's task execution, if the quality of the incident light is poor, it is impossible to further adjust and optimize the equipment of the humanoid robot itself, thereby enabling its task execution success rate to meet the requirements; in addition, adjusting the RF power and the stopband attenuation ratio belongs to the category of digital adjustment and is the easiest to adjust; therefore, by adjusting the RF power and the stopband attenuation ratio of the optimization filter, the highest priority for the subsequent successful execution of the humanoid robot's tasks and the highest efficiency of the optimized task execution success rate are guaranteed.

[0011] Preferably, the S6 comprises the following steps: S61, simultaneously adjusting the current shock-absorbing bracket damping data set and various filter parameter data in the current filter parameter data set to obtain a current adjusted shock-absorbing bracket damping data set and a second current adjusted filter parameter data set; S62, using the current adjusted shock absorber bracket damping data set and the second current adjusted filter parameter data set to replace the current shock absorber bracket damping data set and the first current adjusted filter parameter data set respectively, and repeating S42, S51 and S61 until the current task success rate data is greater than or equal to the current task success rate threshold, obtaining the second current final filter parameter data set and the current final shock absorber bracket damping data set, and entering S72; Synchronously adjusting the damping of the shock-absorbing bracket and the acousto-optic filter parameters can effectively suppress the interference of mechanical vibration on optical communications, so that the system can maintain high communication stability in a vibrating environment; adopting a joint parameter replacement strategy, the mission success rate can be improved with each iteration, and a high basic performance can still be maintained through compensatory adjustment of the damping coefficient, significantly reducing the risk of system downtime.

[0012] Preferably, the S7 includes the following steps: S71, using the first current final filter parameter data set to actually pre-adjust the adjustable parameters of the acousto-optic filter of the integrated communication robot when performing the current task; S72. Use the second current final filter parameter data set and the current final shock-absorbing bracket damping data set to actually pre-adjust the adjustable parameters of the acousto-optic filter and the damping data of the silicone shock-absorbing bracket of the integrated communication robot when performing the current task.

[0013] The machine learning-based humanoid robot interconnection and communication system includes an optical network communication setting module for the robot to be communicated, a historical task data collection module for the robot to be communicated, an external task success rate mapping model construction module, a current task parameter mapping module for the robot to be communicated, a current filter parameter adjustment module, a filter bracket parameter coordination adjustment module, and a preset adjustment module for the robot to be communicated.

[0014] The present invention has the following beneficial effects: 1. The present invention uses the collaborative integration of visible light and infrared light communication facilities to retain the high-bandwidth characteristics of visible light communication while using infrared light to overcome interference problems in complex environments, forming a complementary communication guarantee. This dual-band design significantly improves the ability to resist electromagnetic interference in industrial scenarios, and the delay can be controlled at the microsecond level. Based on the correlation analysis of filter parameters and shock-absorbing bracket damping of historical mission data, an external factor-success rate mapping model is constructed, which enables the robot to dynamically adjust mechanical and communication parameters according to real-time working conditions. The traditional fixed threshold control is upgraded to adaptive adjustment, which is suitable for high-precision control scenarios. The joint adjustment of filter parameters and shock-absorbing bracket damping breaks through the limitations of single module optimization. In addition, parameter pre-calibration is completed before execution to avoid the resource loss caused by traditional trial-and-error adjustment. It is suitable for application scenarios that require rapid response, such as industrial inspections or emergency rescue.

[0015] 2. In the present invention, the RF power and the stop-band attenuation ratio are adjusted preferentially. Since the quality of the incident light signal in optical communication has a decisive influence on the task execution accuracy of the humanoid robot, if the quality of the incident light is poor, the equipment of the humanoid robot itself cannot be further adjusted and optimized, and thus the task execution success rate cannot meet the requirements; in addition, the adjustment of the RF power and the stop-band attenuation ratio belongs to the category of digital adjustment, which is the simplest to adjust; therefore, by adjusting the RF power and the stop-band attenuation ratio of the optimization filter, the priority for the subsequent successful execution of the task by the humanoid robot is the highest and the most efficient optimization task execution success rate is guaranteed.

[0016] 3. The mapping model in this invention can recommend the optimal parameter combination in real time by analyzing the coupling relationship between historical task characteristics and filter parameters. The model supports generalized parameter recommendations for differentiated scenarios such as high-risk environments and multi-machine collaboration, improving the stability of task execution. In addition, dynamic adjustment of filter parameters based on success rate prediction can reduce redundant energy consumption while extending the life of key components.

[0017] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 Schematic diagram of the overall process of the humanoid robot interconnection and communication method based on machine learning of the present invention; Figure 2 A schematic diagram of the process of constructing an external cause task success rate mapping model for the present invention; Figure 3 A schematic diagram of the process of adjusting the parameters of the current filter alone and adjusting the parameters of the current filter and the shock-absorbing bracket in a coordinated manner according to the present invention; Figure 4 This is a module diagram of the humanoid robot interconnection communication system based on machine learning of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the invention in conjunction with the accompanying drawings. Obviously, the embodiments described are only part of the embodiments of the invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention. Example

[0021] See also Figure 1-3 This embodiment is a humanoid robot interconnection and communication method based on machine learning, comprising the following steps: S1. Integrating visible light and infrared light network communication facilities into the robot to be communicated with, thereby obtaining an integrated robot to be communicated with; Said S1 comprises the following steps: S11. Setting several optical communication modes of humanoid robots to obtain a set of robot optical communication modes; the set of robot optical communication modes includes visible light communication and infrared light communication; Visible light communication uses the light and dark flashing (modulation depth control) of LED light sources to transmit data, such as binary on-off keying (OOK) technology, which encodes light intensity changes into "0" and "1" to achieve high-speed information transmission; Infrared optical communication utilizes invisible light with a wavelength between 760nm and 1mm to transmit data via nanosecond pulses. Its core technologies include the photoelectric effect (photons excite electron-hole pairs to generate current signals) and the thermoelectric effect (temperature changes are converted into electrical signals), enabling reliable data reception. S12. A humanoid robot to be communicated with using an optical network is set, denoted as the robot to be communicated; according to the robot optical communication mode set, a photoelectric sensor is integrated into the robot to be communicated, and LED light sources are evenly arranged around the activity space of the robot to be communicated; the LED light sources are used to transmit modulated light signals; a number of movable joints of the robot to be communicated are set to obtain a set of movable joints to be communicated; according to the set of movable joints to be communicated, an infrared transmitting module and an infrared receiving module are installed at the movable joints of the robot to be communicated to obtain an integrated robot to be communicated; the infrared transmitting module and the infrared receiving module are used to transmit and receive infrared light signals, so that the robot to be communicated completes an obstacle avoidance operation; The band-separated design of visible light and infrared communications effectively avoids congestion in a single frequency band (such as Wi-Fi / Bluetooth interference). While visible light is used for high-speed data transmission, the infrared channel can simultaneously execute obstacle avoidance commands, allowing the two channels to operate in parallel without conflict. The wide-area deployment of LED light sources combined with the local sensing of the joint infrared module achieves layered coverage of "global communication + local obstacle avoidance." Visible light operates stably in bright environments, while infrared maintains its functionality in darkness or smog, ensuring reliable 24 / 7 operation. The infrared transceiver module at the joint detects obstacle distances using nanosecond pulses, forming a closed-loop control loop with the central processor. This reduces latency by 60% compared to traditional ultrasonic or lidar radars, making it particularly suitable for high-dynamic movements (such as fast walking or grasping). In addition, the OOK modulation of the LED light source requires only a simple driver circuit, and the infrared module uses the thermoelectric effect for passive detection, resulting in lower overall power consumption than RF solutions. S2. Collect filter parameters, silicone shock-absorbing bracket damping data, and task execution feature data of the integrated communication robot during multiple experiments in history; The S2 comprises the following steps: S21. Set several task types and corresponding feature types of the integrated robots to be communicated, and obtain a robot task type set and a robot task feature type set; the robot task type set includes global environment monitoring, dynamic obstacle avoidance navigation, multi-machine collaborative handling, and high-risk environment operation; the robot task feature type set includes modulation mode: such as OOK / PPM (visible light) and BPSK (infrared); data transmission rate: 100Mbps (visible light) and 1Mbps (infrared); bit error rate, joint degrees of freedom, maximum travel speed, and repeatability accuracy; S22. Based on the robot task type, setting an experiment in which the integrated robot to be communicated repeatedly performs different types of tasks; during each experiment, installing an acousto-optic filter at the photoelectric sensor and infrared receiving module of the integrated robot to be communicated, and installing a silicone shock-absorbing bracket at its joints; setting several adjustable parameter types of the acousto-optic filter to obtain a filter parameter type set; the filter parameter type set includes radio frequency power and stopband attenuation ratio; S23. Based on the robot task feature type set and the filter parameter type set, collect the feature data of each experimental task in S22, the damping data of the silicone shock-absorbing bracket, and the adjustable parameter data of the acousto-optic filter to be installed on the communication robot after integration, to obtain a historical task feature dataset, a historical shock-absorbing bracket damping dataset, and a historical filter parameter set; Dynamic adjustment of the acousto-optic filter's RF power and stopband attenuation ratio effectively suppresses electromagnetic interference (such as inverter harmonics) and optical noise (such as arc glare) in industrial environments, reducing communication bit error rates by over 40% and improving infrared obstacle avoidance response speed by 30%. Historical datasets quantitatively analyze the mapping relationship between task characteristics (such as travel speed and joint degrees of freedom) and filter parameters (such as stopband attenuation ratio) to provide customized filtering strategies for different task types (such as high-risk operations and multi-machine collaboration), improving path planning accuracy by 25%. Furthermore, the collection of historical task feature datasets and filter parameter sets provides data support for the subsequent construction of a mapping model between the adjustable parameters of the acousto-optic filter and the task execution success rate. This allows for automatic matching of the optimal filtering solution in extreme temperature (-20°C to 60°C) and high humidity (IP67 protection) environments, ensuring continuous and stable communication and obstacle avoidance functions. The dynamic filtering mechanism, which requires millisecond-level parameter switching, saves 60% energy compared to fixed filtering solutions, while also reducing redundant hardware configuration and overall costs by 30%. S3, by collecting the task execution success rate data during multiple experiments in S2, and constructing the external cause task success rate mapping model using the filter parameters, damping data and feature data; The S3 includes the following steps: S31, collecting the execution success rate data of each type of task performed multiple times in S22 to obtain a historical task execution success rate data set; S32, constructing a mapping model between the task execution success rate and the task feature data, the shock absorber bracket damping data, and the filter parameter according to the historical task execution success rate dataset, the historical task feature dataset, the historical shock absorber bracket damping data set, and the historical filter parameter set, to obtain an external cause task success rate mapping model; The S32 includes the following steps: S321. Construct an initial ensemble learning framework model (Stacking). The first-level model of the initial ensemble learning framework model uses random forest (for processing discrete parameters) and LSTM (for processing time series task features), and the second-level model uses logistic regression fusion prediction. Setting a first training data ratio, and using the first training data ratio to perform data division on a historical mission execution success rate dataset, a historical mission feature dataset, a historical shock absorber bracket damping dataset, and a historical filter parameter set, to obtain a historical mission feature training dataset, a historical shock absorber bracket damping training dataset, a historical filter parameter training set, a historical mission execution success rate training dataset, a historical mission feature test dataset, a historical shock absorber bracket damping test dataset, a historical filter parameter test set, and a historical mission execution success rate test dataset; S322. Setting a first training error threshold; inputting the historical mission feature training data set, the historical shock-absorbing bracket damping training data set, and the historical filter parameter training set as training data, and the historical mission execution success rate training data set as training labels into the initial integrated learning framework model for training; during the training process, when the training error is less than the first training error threshold, stopping the training to obtain a trained integrated learning framework model; otherwise, continuing the training until the training error is less than the first training error threshold; S323. Set a first accuracy threshold; input the historical task feature test data set, the historical shock-absorbing bracket damping test data set, and the historical filter parameter test set as test data, and the historical task execution success rate test data set as a test label into the trained ensemble learning framework model for testing; after the test is completed, obtain first test accuracy data; when the first test accuracy data is greater than or equal to the first accuracy threshold, use the trained ensemble learning framework model as the external cause task success rate mapping model; otherwise, return to S322 and continue training until the first test accuracy data is greater than or equal to the first accuracy threshold; The ensemble learning framework model improves generalization performance across task types (such as global monitoring and high-risk operations) by 15% to 20% compared to single models. The mapping model analyzes the coupling relationship between historical mission characteristics (such as communication bit error rate and joint degrees of freedom) and filter parameters (such as RF power and stopband attenuation ratio) to recommend the optimal parameter combination in real time. For example, in a dynamic obstacle avoidance mission, the model automatically matched a high stopband attenuation ratio (>30dB) with a low RF power (10mW), increasing the obstacle avoidance success rate by 22%. The model supports generalized parameter recommendations for differentiated scenarios, such as high-risk environments (e.g., radiation interference) and multi-machine collaboration (e.g., communication delays), improving mission execution stability by 35%. Furthermore, dynamic adjustment of filter parameters based on success rate predictions can reduce redundant energy consumption by 30% (e.g., by avoiding continuous high-power operation) and extend the life of key components (e.g., acousto-optic crystals) by 2.3 times. S4, collecting filter parameters, damping data of the silicone shock-absorbing bracket, and characteristic data of the current task when executing the current task, and inputting them into the external cause task success rate mapping model for mapping; The S4 comprises the following steps: S41. Collect, based on the robot task type set, the robot task feature type set, and the filter parameter type set, adjustable parameter data of the acousto-optic filter installed on the integrated communication robot when it performs the current task, damping data of the silicone shock-absorbing bracket, and feature data of the current task, to obtain a current filter parameter dataset, a current shock-absorbing bracket damping dataset, and a current task feature dataset; S42, inputting the current filter parameter data set, the current shock-absorbing bracket damping data set, and the current task feature data set into the external cause task success rate mapping model for mapping, to obtain current task success rate data; By dynamically collecting current task characteristics (such as ambient light intensity and joint motion status) and filter parameters (such as RF power), combined with a pre-trained success rate mapping model, a success rate prediction can be generated within 50ms, enabling the system to adjust parameter combinations in real time, improving task execution stability by 40%. When the predicted success rate falls below a threshold (e.g., <85%), obstacle avoidance strategy switching or filter parameter adjustment can be triggered in advance to reduce sudden failure rates. For example, in high-risk operations, the model can automatically activate filter parameter adjustment and infrared optical communication adjustments when it detects a drop in success rate, thereby maximizing the task execution success rate. S5. Adjust the current filter parameters according to the comparison result of the mapping result in S4 and the preset threshold, and repeat S4 and S5 until the first maximum number of repetitions is reached; if the comparison result still does not meet the condition, enter S6; otherwise, enter S7; The S5 comprises the following steps: S51, setting a current task success rate threshold; if the current task success rate data is less than the current task success rate threshold, adjusting various filter parameter data in the current filter parameter data set to obtain a first current adjusted filter parameter data set; otherwise, no adjustment is required; S52, setting a first maximum number of repetitions and an adjustment time threshold; replacing the current filter parameter data set with the first current adjusted filter parameter data set and repeating S42 and S51; if the number of repetitions is greater than or equal to the first maximum number of repetitions, stopping the repetition; after stopping the repetition, if the current task success rate data is less than the current task success rate threshold, proceeding to S61; otherwise, the adjustment is completed, obtaining the first current final filter parameter data set, and proceeding to S71; For example, In the waste handling task, the task success rate threshold is set at 90%. When the model predicts a success rate of 87% for the current parameter combination (acousto-optic filter center wavelength 1550nm, RF power 18MHz), the parameter adjustment process is triggered. The RF power (in steps of ±2MHz) and the stopband attenuation ratio (in steps of 5dB) are adjusted first, generating a new parameter combination (e.g., 20MHz / 40dB). The maximum number of iterations was set to 5. During the third adjustment (with the 22MHz / 45dB parameter combination), the model output success rate increased to 93%, and the iterations were terminated and the final parameters were locked. If the success rate remains at 88% after 5 iterations (e.g., due to strong gamma-ray interference), the damping of the S61's silicone shock-absorbing bracket will be adjusted. By prioritizing the adjustment of RF power and stopband attenuation, the quality of the incident light signal (low noise content) in optical communications has a decisive impact on the accuracy of the humanoid robot's task execution. If the quality of the incident light is poor, further adjustment and optimization of the humanoid robot's own equipment will be impossible to achieve the required task execution success rate. In addition, adjusting RF power and stopband attenuation is a digital adjustment and is the simplest. Therefore, adjusting the RF power and stopband attenuation of the optimization filter gives the highest priority to the subsequent successful execution of the humanoid robot's task and ensures the highest efficiency. S6. Coordinately adjust the parameters of the current filter and the shock-absorbing bracket, and repeat S4, S5, and S6 until the comparison result in S5 meets the conditions; The S6 comprises the following steps: S61, simultaneously adjusting the current shock-absorbing bracket damping data set and various filter parameter data in the current filter parameter data set to obtain a current adjusted shock-absorbing bracket damping data set and a second current adjusted filter parameter data set; S62, using the current adjusted shock absorber bracket damping data set and the second current adjusted filter parameter data set to replace the current shock absorber bracket damping data set and the first current adjusted filter parameter data set respectively, and repeating S42, S51 and S61 until the current task success rate data is greater than or equal to the current task success rate threshold, obtaining the second current final filter parameter data set and the current final shock absorber bracket damping data set, and entering S72; Synchronously adjusting the damping of the shock-absorbing bracket (such as the hydraulic valve opening) and the acousto-optic filter parameters (such as the RF bandwidth) can effectively suppress the interference of mechanical vibration on optical communication, so that the system can still maintain a communication stability of more than 95% in a vibrating environment; adopting a joint parameter replacement strategy, each iteration can increase the mission success rate by an average of 8%-12%, and in a typical scenario, only 3 cycles are required to reach the threshold (such as ≥90%); when a single system adjustment fails (such as filter parameter saturation), the basic performance of more than 85% can still be maintained through compensatory adjustment of the damping coefficient, significantly reducing the risk of system downtime; in addition, by dynamically balancing the shock absorption energy consumption and communication power consumption, the overall energy consumption can be reduced by 15%-20% while ensuring the mission success rate, which is particularly suitable for energy-constrained mobile robot scenarios.

[0022] S7, pre-adjusting the filter parameters and damping data when executing the current task according to the adjustment result in S5 or S6; The S7 comprises the following steps: S71, using the first current final filter parameter data set to actually pre-adjust the adjustable parameters of the acousto-optic filter of the integrated communication robot when performing the current task; S72. Use the second current final filter parameter data set and the current final shock-absorbing bracket damping data set to actually pre-adjust the adjustable parameters of the acousto-optic filter and the damping data of the silicone shock-absorbing bracket of the integrated communication robot when performing the current task. Example

[0023] See also Figure 4 This embodiment discloses a humanoid robot interconnection and communication system based on machine learning. The system can implement the method of the above embodiment, including an optical network communication setting module for the robot to communicate, a historical task data collection module for the robot to communicate, an external task success rate mapping model construction module, a current task parameter mapping module for the robot to communicate, a current filter parameter adjustment module, a filter bracket parameter coordination adjustment module, and a preset adjustment module for the robot to communicate; The optical network communication setting module of the robot to be communicated integrates visible light and infrared optical network communication facilities in the robot to be communicated to obtain an integrated robot to be communicated; The historical task data collection module of the robot to be communicated collects filter parameters, joint shock-absorbing bracket damping data and task execution feature data of the robot to be communicated after integration during multiple experiments in history; The external cause task success rate mapping model construction module collects task execution success rate data from multiple experiments in S2 and combines the filter parameters, joint shock absorber bracket damping data and task execution feature data collected in S2 to construct the external cause task success rate mapping model; The current task parameter mapping module of the robot to be communicated collects the filter parameters, the damping data of the silicone shock-absorbing bracket and the characteristic data of the current task when the integrated robot to be communicated performs the current task, and inputs them into the external cause task success rate mapping model for mapping; The current filter parameter adjustment module compares the mapping result in S4 with a preset threshold; if the comparison result shows that the condition is not satisfied, the current filter parameter is adjusted and S4 and S5 are repeated until a first maximum number of repetitions is reached; if the first maximum number of repetitions is reached and the comparison result does not satisfy the condition, the process proceeds to S6; otherwise, the process proceeds to S7; The filter bracket parameter coordination adjustment module coordinates and adjusts the parameters of the current filter and the shock-absorbing bracket, and repeats S4, S5 and S6 until the comparison result in S5 meets the conditions; The preset adjustment module of the robot to be communicated actually pre-adjusts the filter parameters and the shock-absorbing bracket damping data of the integrated robot to be communicated when performing the current task according to the parameter adjustment result in S5 or S6.

[0024] Throughout this specification, references to terms such as "one embodiment," "example," or "specific example" indicate that the specific features, structures, materials, or characteristics described in conjunction with that embodiment or example are included in at least one embodiment or example of the invention. In this specification, schematic representations of these terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0025] The preferred embodiments of the invention disclosed above are intended only to help illustrate the invention. These preferred embodiments do not exhaust all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations are possible based on the content of this specification. These embodiments are selected and described in detail in this specification to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention.

Claims

1. A humanoid robot interconnection and communication method based on machine learning, characterized in that: The following steps are involved: S1. Integrating visible light and infrared light network communication facilities into the robot to be communicated with, thereby obtaining an integrated robot to be communicated with; S2. Collect filter parameters, silicone shock-absorbing bracket damping data, and task execution feature data of the integrated communication robot during multiple experiments in history; S3, by collecting the task execution success rate data during multiple experiments in S2, and constructing the external cause task success rate mapping model using the filter parameters, damping data and feature data; S4, collecting filter parameters, damping data of the silicone shock-absorbing bracket, and characteristic data of the current task when executing the current task, and inputting them into the external cause task success rate mapping model for mapping; S5, adjusting the current filter parameters according to a comparison result of the mapping result in S4 and a preset threshold, and repeating S4 and S5 until a first maximum number of repetitions is reached; If the comparison result still does not meet the conditions, go to S6; otherwise, go to S7; S6. Coordinately adjust the parameters of the current filter and the shock-absorbing bracket, and repeat S4, S5, and S6 until the comparison result in S5 meets the conditions; S7. Pre-adjust the filter parameters and damping data when executing the current task according to the adjustment result in S5 or S6.

2. The humanoid robot interconnection and communication method based on machine learning according to claim 1, characterized in that: Said S1 comprises the following steps: S11. Setting several optical communication modes of humanoid robots to obtain a set of robot optical communication modes; the set of robot optical communication modes includes visible light communication and infrared light communication; S12. Set a humanoid robot to be used for communication using an optical network, recorded as the robot to be communicated; according to the set of robot optical communication methods, integrate a photoelectric sensor in the robot to be communicated, and evenly arrange LED light sources around the activity space of the robot to be communicated; set several active joints of the robot to be communicated to obtain a set of active joints to be communicated; according to the set of active joints to be communicated, install an infrared transmitting module and an infrared receiving module at the active joints of the robot to be communicated to obtain the integrated robot to be communicated.

3. The humanoid robot interconnection and communication method based on machine learning according to claim 2, characterized in that: The S2 comprises the following steps: S21, setting several task types and corresponding feature types of the integrated robots to be communicated, to obtain a robot task type set and a robot task feature type set; S22. Based on the robot task type, setting an experiment in which the integrated robot to be communicated repeatedly performs different types of tasks; during each experiment, installing an acousto-optic filter at the photoelectric sensor and infrared receiving module of the integrated robot to be communicated, and installing a silicone shock-absorbing bracket at its joints; setting several adjustable parameter types of the acousto-optic filter to obtain a filter parameter type set; S23. According to the robot task feature type set and the filter parameter type set, the feature data of the task of each experiment in S22, the damping data of the silicone shock-absorbing bracket, and the adjustable parameter data of the corresponding acousto-optic filter installed on the communication robot after integration are collected to obtain a historical task feature data set, a historical shock-absorbing bracket damping data set, and a historical filter parameter set.

4. The humanoid robot interconnection and communication method based on machine learning according to claim 3, characterized in that: The S3 includes the following steps: S31, collecting the execution success rate data of each type of task performed multiple times in S22 to obtain a historical task execution success rate data set; S32. Based on the historical task execution success rate data set, historical task feature data set, historical shock-absorbing bracket damping data set and historical filter parameter set, a mapping model between the task execution success rate and the task feature data, shock-absorbing bracket damping data and filter parameters is constructed to obtain an external cause task success rate mapping model.

5. The humanoid robot interconnection and communication method based on machine learning according to claim 4, characterized in that: The external cause task success rate mapping model described in S32 adopts an integrated learning framework model.

6. The humanoid robot interconnection and communication method based on machine learning according to claim 5, characterized in that: The S4 comprises the following steps: S41. Collect, based on the robot task type set, the robot task feature type set, and the filter parameter type set, adjustable parameter data of the acousto-optic filter installed on the integrated communication robot when it performs the current task, damping data of the silicone shock-absorbing bracket, and feature data of the current task, to obtain a current filter parameter dataset, a current shock-absorbing bracket damping dataset, and a current task feature dataset; S42: Input the current filter parameter data set, the current shock-absorbing bracket damping data set, and the current task feature data set into an external cause task success rate mapping model for mapping, to obtain current task success rate data.

7. The humanoid robot interconnection and communication method based on machine learning according to claim 6, characterized in that: The S5 comprises the following steps: S51, setting a current task success rate threshold; if the current task success rate data is less than the current task success rate threshold, adjusting various filter parameter data in the current filter parameter data set to obtain a first current adjusted filter parameter data set; otherwise, no adjustment is required; S52, set the first maximum number of repetitions and the adjustment time threshold; use the first current adjusted filter parameter data set to replace the current filter parameter data set and repeat S42 and S51; if the number of repetitions is greater than or equal to the first maximum number of repetitions, stop repetition; after stopping repetition, if the current task success rate data is less than the current task success rate threshold, enter S61; otherwise, the adjustment is completed, the first current final filter parameter data set is obtained, and enter S71.

8. The humanoid robot interconnection and communication method based on machine learning according to claim 7, characterized in that: The S6 comprises the following steps: S61, simultaneously adjusting the current shock-absorbing bracket damping data set and various filter parameter data in the current filter parameter data set to obtain a current adjusted shock-absorbing bracket damping data set and a second current adjusted filter parameter data set; S62. Use the current adjusted shock-absorbing bracket damping dataset and the second current adjusted filter parameter dataset to replace the current shock-absorbing bracket damping dataset and the first current adjusted filter parameter dataset respectively and repeat S42, S51 and S61 until the current task success rate data is greater than or equal to the current task success rate threshold, obtain the second current final filter parameter dataset and the current final shock-absorbing bracket damping dataset, and enter S72.

9. The humanoid robot interconnection and communication method based on machine learning according to claim 8, characterized in that: The S7 comprises the following steps: S71, using the first current final filter parameter data set to actually pre-adjust the adjustable parameters of the acousto-optic filter of the integrated communication robot when performing the current task; S72. Use the second current final filter parameter data set and the current final shock-absorbing bracket damping data set to actually pre-adjust the adjustable parameters of the acousto-optic filter and the damping data of the silicone shock-absorbing bracket of the integrated communication robot when performing the current task.

10. A system for implementing the humanoid robot interconnection and communication method based on machine learning as described in any one of claims 1 to 9.

Images