Robot control method and device, robot and storage medium

Abstract

The invention provides a robot control method and device, a robot and a storage medium, and the robot control method comprises the steps: receiving instruction information, and outputting a corresponding first action; receiving first input information and outputting corresponding first environment momentum, wherein the first input information comprises instruction information, a first action and a first evaluation result aiming at the first action; wherein the first evaluation result represents that the user is not satisfied when the robot executes the first action based on the instruction information; determining a second environment momentum under the condition that the set environment momentum having the similarity with the first environment momentum greater than or equal to a similarity threshold exists in a set database; adjusting the first action to obtain a second action; and executing the second action. According to the method and the device, the robot can adjust own behaviors according to user requirements and environment changes, so that the robot can provide more intelligent and adaptive services.

Description

Technical Field

[0001] The present disclosure relates to the field of robotics technology, and in particular to a robot control method, device, robot, and storage medium. Background Art

[0002] Robot control refers to the guidance and management of robots to achieve specific tasks or behaviors through pre-defined algorithms, programming, or artificial intelligence technologies. This includes managing the robot's motion, perception, and decision-making processes. In traditional robot control, humans directly control the robot's movements by writing code or using remote controls, such as performing specific industrial tasks or simple movements. Modern robot control, however, increasingly involves technologies for autonomous learning, perception, and environmental adaptation.

[0003] Robot control typically involves multiple levels of processing, including perception, planning, and execution. The perception stage involves the robot's perception of its surroundings and acquisition of data, which can be achieved through devices such as sensors and cameras. The planning stage involves decision-making, that is, determining the actions the robot should take, which are usually based on task requirements and environmental changes. The execution stage converts planned action instructions into actual robot movements or operations, involving the control of hardware such as motors and actuators.

[0004] In related technologies, robots are controlled by receiving audio commands from the user (e.g., "pick up this apple") and then performing corresponding actions based on the images captured by the robot. This method is very rigid and inflexible. The robot's behavior is pre-set and it cannot adapt to environmental feedback like humans do. Summary of the Invention

[0005] In view of this, in order to solve the above-mentioned technical problem that the robot control method is very rigid and inflexible, the present disclosure provides a robot control method, device, robot and storage medium.

[0006] According to a first aspect of an embodiment of the present disclosure, a robot control method is provided, the robot control method comprising:

[0007] Setting a command recognition model to receive command information and output a corresponding first action;

[0008] an environmental momentum extractor, receiving first input information and outputting a corresponding first environmental momentum, the first input information including the instruction information, the first action, and a first evaluation result for the first action; wherein the first evaluation result indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information; and the first environmental momentum indicates that the user's evaluation result when the robot executes the first action based on the instruction information is the first evaluation result;

[0009] If there is a set environmental momentum in the set database whose similarity to the first environmental momentum is greater than or equal to a similarity threshold, determining a second environmental momentum based on the set environmental momentum; wherein the set environmental momentum represents a user evaluation result when the robot performs a set action based on the set instruction information, as a set evaluation result;

[0010] adjusting the first action based on the second environmental momentum to obtain a second action;

[0011] Perform the second action.

[0012] In an optional embodiment, adjusting the first action based on the second environmental momentum to obtain the second action includes:

[0013] multiplying the set adjustment coefficient by the second environmental momentum to determine the action adjustment amount;

[0014] The first action is adjusted according to the action adjustment amount to obtain the second action.

[0015] In an optional embodiment, the robot control method includes:

[0016] If the setting database does not contain the setting environmental momentum whose similarity to the first environmental momentum is greater than or equal to the similarity threshold, the first action is performed.

[0017] In an optional embodiment, after performing the first action, the robot control method includes:

[0018] Obtaining a target evaluation result through evaluation information of the first action;

[0019] The environmental momentum extractor receives second input information and outputs a corresponding third environmental momentum, wherein the second input information includes the instruction information, the first action, and the target evaluation result; wherein the third environmental momentum represents that the user's evaluation result when the robot performs the first action based on the instruction information is the target evaluation result;

[0020] The third environmental momentum is stored in the setting database as the setting environmental momentum.

[0021] In an optional embodiment, the evaluation information includes an evaluation video, and obtaining a target evaluation result based on the evaluation information of the first action includes:

[0022] Outputting first prompt information; wherein the first prompt information is used to prompt the user to evaluate the first action;

[0023] In response to outputting the first prompt information, collecting an evaluation video of a set length; wherein the evaluation video is a video of the user evaluating the first action in response to the first prompt information;

[0024] The target evaluation result is obtained through the evaluation video.

[0025] In an optional embodiment, obtaining the target evaluation result through the evaluation video includes:

[0026] Extracting image information and audio information from the evaluation video;

[0027] a first prediction model, receiving the image information and outputting a corresponding image evaluation result;

[0028] a second prediction model, receiving the audio information and outputting a corresponding audio evaluation result;

[0029] When the image evaluation result and the audio evaluation result are the same evaluation result, the same evaluation result is determined as the target evaluation result.

[0030] In an optional embodiment, the robot control method includes:

[0031] When the image evaluation result and the audio evaluation result are different evaluation results, outputting second prompt information; wherein the second prompt information indicates that the image evaluation result and the audio evaluation result are different evaluation results;

[0032] When the output times of the second prompt information do not reach the set times, the first prompt information is output again.

[0033] According to a second aspect of an embodiment of the present disclosure, there is provided a robot control device, the robot control device comprising:

[0034] Setting a command recognition model for receiving command information and outputting a corresponding first action;

[0035] an environmental momentum extractor, configured to receive first input information and output a corresponding first environmental momentum, the first input information including the instruction information, the first action, and a first evaluation result for the first action; wherein the first evaluation result indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information; and the first environmental momentum indicates that the user's evaluation result when the robot executes the first action based on the instruction information is the first evaluation result;

[0036] an action determination module, configured to, if a set environmental momentum exists in a set database and has a similarity with the first environmental momentum greater than or equal to a similarity threshold, determine a second environmental momentum based on the set environmental momentum; wherein the set environmental momentum represents a user evaluation result when the robot performs a set action based on set instruction information, as a set evaluation result; and further configured to adjust the first action based on the second environmental momentum to obtain a second action;

[0037] An action execution module is used to execute the second action.

[0038] According to a third aspect of the embodiments of the present disclosure, a robot is provided, comprising:

[0039] processor;

[0040] a memory for storing instructions executable by the processor;

[0041] Wherein, the processor is configured to execute the robot control method as described in any one of the first aspects.

[0042] According to a fourth aspect of an embodiment of the present disclosure, a storage medium is provided, which stores one or at least one program, and the one or at least one program can be executed by one or at least one processor to implement the robot control method described in any one of the first aspects.

[0043] The technical solution provided by the embodiments of the present disclosure may include the following beneficial effects: In the present disclosure, after obtaining the first action corresponding to the instruction information based on the set instruction recognition model, the user's first evaluation result expressing dissatisfaction with the robot's first action specified based on the instruction information, as well as the corresponding instruction information and the first action, can be input into a pre-trained environmental momentum extractor, and the environmental momentum extractor then outputs the corresponding first environmental momentum. Then, based on the first environmental momentum, an environmental momentum is searched from a set database. If there is a set environmental momentum in the set database with a similarity greater than or equal to a similarity threshold to the first environmental momentum, the first action can be adjusted based on the set environmental momentum with the greatest similarity, thereby obtaining a second action. The robot can then execute the second action as a response to the instruction information. Due to the set environmental momentum representation, the user's evaluation result when the robot executes the set action based on the set instruction information is the set evaluation result. Therefore, adjusting the first action based on the appropriate set environmental momentum found can provide intelligent and flexible application possibilities for multiple scenarios, enabling the robot to adjust its own behavior according to user needs and environmental changes, so that the robot can provide more intelligent and adaptive services, thereby enhancing the application value of the robot in actual work and interaction.

[0044] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0046] Figure 1 is a flowchart of a robot control method according to an exemplary embodiment.

[0047] Figure 2 is a flowchart of a robot control method according to another exemplary embodiment.

[0048] Figure 3 is a block diagram of a robot control device according to an exemplary embodiment.

[0049] Figure 4 is a block diagram of a robot according to an exemplary embodiment. DETAILED DESCRIPTION

[0050] The following will describe the embodiments of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand the other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for the purpose of illustrating the present application and are not intended to limit the scope of protection of the present application.

[0051] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. Therefore, the illustrations only show components related to the present application rather than the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component can be changed at will, and the component layout type may also be more complicated.

[0052] The following will describe the embodiments of the present application with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand the other advantages and effects of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be understood that the preferred embodiments are only for the purpose of illustrating the present application and are not intended to limit the scope of protection of the present application.

[0053] The embodiments of the present disclosure provide a robot control method, device, robot, and storage medium. In the present disclosure, after obtaining a first action corresponding to instruction information based on a set instruction recognition model, the first evaluation result of the user expressing dissatisfaction with the first action specified by the robot based on the instruction information, as well as the corresponding instruction information and the first action, can be input as input information into a pre-trained environmental momentum extractor, and then the environmental momentum extractor outputs the corresponding first environmental momentum. Then, based on the first environmental momentum, an environmental momentum search is performed from a set database. If there is a set environmental momentum in the set database whose similarity with the first environmental momentum is greater than or equal to a similarity threshold, the first action can be adjusted based on the set environmental momentum with the greatest similarity, thereby obtaining a second action, and the robot can then execute the above-mentioned second action as a response to the instruction information. Due to the set environment momentum representation, the user's evaluation result when the robot performs the set action based on the set instruction information is the set evaluation result. Therefore, by adjusting the first action according to the queried appropriate set environment momentum, it can provide intelligent and flexible application possibilities for multiple scenarios, enabling the robot to adjust its own behavior according to user needs and environmental changes, so that the robot can provide more intelligent and adaptive services, thereby enhancing the application value of the robot in actual work and interaction.

[0054] In an exemplary embodiment, a robot control method is provided, which can be applied to a robot. Figure 1 As shown, the method may include:

[0055] S110, setting a command recognition model, receiving command information and outputting a corresponding first action;

[0056] S120, an environmental momentum extractor receives first input information and outputs a corresponding first environmental momentum, the first input information including instruction information, a first action, and a first evaluation result for the first action; wherein the first evaluation result indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information; and the first environmental momentum indicates that the user's evaluation result when the robot executes the first action based on the instruction information is the first evaluation result.

[0057] S130: If a set environmental momentum exists in the set database and its similarity to the first environmental momentum is greater than or equal to a similarity threshold, determining a second environmental momentum based on the set environmental momentum; wherein the set environmental momentum represents a user evaluation result when the robot performs a set action based on the set instruction information as a set evaluation result;

[0058] S140, adjusting the first action based on the second environmental momentum to obtain a second action;

[0059] S150: Execute the second action.

[0060] In step S110 , the instruction information may include text instruction information, image instruction information, or other instruction information, which is not limited.

[0061] For example, when a user needs to control a robot to perform certain actions, the user can issue control instructions to the robot, where the user can issue control instructions through audio and / or movement. After the robot collects the user's audio instructions, it can convert the user's audio instructions into text instructions through an audio-to-text conversion model deployed within the robot. The audio-to-text conversion model is a model that can convert audio information into text information, and its specific model type is not limited. After obtaining the text instruction information, the robot can use the currently captured image as the image instruction information, and then construct the above text instruction information and image instruction information into a complete instruction information.

[0062] It should be noted that, in addition to the above-mentioned methods, the robot may also obtain instruction information through other methods, which are not limited to this.

[0063] In this step, the instruction recognition model may include a multimodal large language model, or other models that can obtain the corresponding first action based on the instruction information, and there is no limitation on this.

[0064] A multimodal large language model is a computational model capable of processing and integrating multiple different data types or information sources. "Multimodal" refers to the model's ability to simultaneously process multiple inputs from different sensors, media, or data sources. The goal of this model is to comprehensively utilize diverse information to more comprehensively and accurately understand and interpret complex real-world scenarios.

[0065] For example, in image recognition tasks, a multimodal large language model can receive image data (visual information) and text data (linguistic information) and process these two information sources simultaneously to improve the understanding of the image content. Such a model helps to more fully grasp the context and meaning behind the image.

[0066] Multimodal large language models are used in a wide range of fields, including natural language processing, computer vision, and audio recognition. This approach enables models to better adapt to the complex and ever-changing real world, improves their ability to comprehensively analyze and understand multi-source information, and provides a powerful tool for solving problems involving multiple perceptions and understandings.

[0067] In some embodiments,

[0068] The setting instruction recognition model can be a large multimodal language model. When the user needs to control the robot to perform some actions, the user can issue control instructions to the robot, wherein the user can issue control instructions through audio and / or actions. After the robot collects the audio instructions issued by the user, it can convert the user's audio instruction information into text instruction information through the audio-to-text conversion model deployed inside the robot. The audio-to-text conversion model is a model that can convert audio information into text information, and its specific model type is not limited. After obtaining the text instruction information, the robot can use the currently captured image as image instruction information, and then input the above-mentioned text instruction information and image instruction information as complete instruction information into the large multimodal language model. After receiving the above-mentioned instruction information, the large multimodal language model can process the above-mentioned text instruction information and image instruction information to obtain the action corresponding to the above-mentioned instruction information. This action can be recorded as the first action.

[0069] It should be noted that, in addition to obtaining the first action through the above-mentioned method, the first action can also be obtained through other methods, which are not limited to this.

[0070] In step S120, after the robot obtains the first action corresponding to the instruction information, since the action recognized by the large model may deviate from the user's true intention, it can be further determined whether the above-mentioned first action needs to be adjusted.

[0071] The instruction information, the first action, and the first evaluation result for the first action may be used as input information, which may be recorded as first input information. The first input information is then input into an environmental momentum extractor. Upon receiving the first input information, the environmental momentum extractor may output a corresponding first environmental momentum. The first environmental momentum may represent a user evaluation result when the robot performs the first action based on the instruction information. The first evaluation result may be an unsatisfactory evaluation result, i.e., the first evaluation result may represent user dissatisfaction with the robot's first action based on the instruction information.

[0072] In this step, the input information is processed by the environmental momentum extractor to obtain the environmental momentum representing the worst user feedback (i.e., dissatisfaction with the first action identified by the set instruction recognition model). This environmental momentum can be recorded as the first environmental momentum.

[0073] The environmental momentum extractor can be a neural network model. During training, command information samples and first action samples can be collected and used as input samples. The command information samples, first action samples, and first evaluation result samples for the first action samples can be referenced to the method for determining command information, and the method for determining first action samples can be referenced to the method for determining first action, without limitation. The first evaluation result sample represents a user's dissatisfaction evaluation result of the robot's execution of the first action sample based on the command information sample.

[0074] Furthermore, after obtaining the command information sample, relevant personnel can determine the action corresponding to the command information sample based on experience or other means. This action is the action that better meets the user's control intention, and then use the action as the second action sample. The environmental momentum sample is then determined based on a set formula.

[0075] The setting formula can be: A' = A - α * E; where A' can be the second motion sample, A can be the first motion sample, E can be the environmental momentum sample, and α is the setting adjustment coefficient. For example, α can be 0.1, or it can be set to other values ​​based on actual needs. The first motion sample, the second motion sample, and the environmental momentum sample can all be represented by a high-dimensional mathematical vector, which is used to characterize the motion.

[0076] Based on the above formula, the environmental momentum sample corresponding to the first and second motion samples can be calculated. This environmental momentum sample can be recorded as the output sample. The above input sample and output sample can be constructed into a pair of training samples. In this way, a training sample set consisting of multiple training sample pairs can be obtained.

[0077] After obtaining the training sample set, the original neural network model can be trained based on the training sample set to obtain the required environmental momentum extractor. This environmental momentum extractor can obtain the corresponding first environmental momentum based on the input instruction information, the first action, and the first evaluation result.

[0078] It should be noted that, in addition to the above-mentioned method, the environmental momentum extractor can also be obtained by other methods, which are not limited to this.

[0079] In some embodiments,

[0080] After receiving the corresponding instruction information I, the first action A, and the first evaluation result R = "unsatisfied", the robot can input (I, A, R = "unsatisfied") as the first input information to the environmental momentum extractor. In response to the received first input information, the environmental momentum extractor outputs the corresponding first environmental momentum E, where the first environmental momentum E represents the environmental momentum corresponding to the worst user feedback (i.e., unsatisfied).

[0081] It should be noted that, in addition to obtaining the first environmental momentum through the above-mentioned method, the first environmental momentum may also be obtained through other methods, which are not limited thereto.

[0082] In step S130, a setting database may be preset in the robot, and the setting database may be used to store the setting environment momentum. The setting environment momentum in the setting database may be preset in advance, or may be newly added or modified as the robot is used, which is not limited.

[0083] After the robot obtains the first environmental momentum, it can search the setting database to determine whether there is a setting environmental momentum in the setting database whose similarity to the first environmental momentum is greater than or equal to a similarity threshold. The similarity threshold can be set based on actual conditions, and its specific value is not limited. For example, the similarity threshold can be 0.6, or other values.

[0084] If a set environmental momentum is obtained from the setting database and its similarity to the first environmental momentum is greater than or equal to the similarity threshold, it indicates that the user is likely to be dissatisfied when the robot performs the first action based on the command information. In this case, it is necessary to determine an appropriate second environmental momentum based on the set environmental momentum with a similarity greater than or equal to the similarity threshold, so as to appropriately adjust the first action.

[0085] For example, the set environmental momentums having similarities greater than or equal to a similarity threshold may be averaged to determine the second environmental momentum.

[0086] For another example, the first environmental momentum and the set environmental momentum with a similarity greater than or equal to a similarity threshold may be averaged to obtain the second environmental momentum.

[0087] For another example, the set environmental momentum with the greatest similarity may be selected from the set environmental momentums whose similarity is greater than or equal to the similarity threshold as the second environmental momentum.

[0088] It should be noted that, generally, the instruction information obtained by the robot each time will not be exactly the same, and its first action determined by the model will not be exactly the same. Therefore, after the first environmental momentum is determined based on the instruction information, the first action and the assumed first evaluation result, there will generally not be exactly the same set environmental momentum in the setting database. Therefore, using the set environmental momentum with the greatest similarity as the second environmental momentum to adjust the first action can better meet the user's control expectations.

[0089] It should be noted that, in addition to obtaining the second environmental momentum through the above-mentioned method, the second environmental momentum can also be obtained through other methods, which are not limited to this.

[0090] In step S140 , after obtaining the second environmental momentum, the first action may be adjusted based on the second environmental momentum, thereby obtaining an action that better meets the user's control expectations. This action may be recorded as the second action.

[0091] The product of the set adjustment coefficient and the second environmental momentum may be first determined as the action adjustment amount, and then the first action may be adjusted by the action adjustment amount to obtain the second action. For example, the second action may be obtained by subtracting the action adjustment amount from the first action.

[0092] The adjustment coefficient can be set according to actual conditions, and its specific value is not limited. Setting an appropriate adjustment coefficient can better ensure that the adjustment range is appropriate and avoid excessive interference with the robot's movements.

[0093] In some embodiments,

[0094] Actions can be adjusted based on a set formula. The set formula can be: A' = A - α * E; where A' can be the second action, A can be the first action, E can be the second environmental momentum, and α is the set adjustment coefficient, for example, α can be 0.1. The first action, second action, and second environmental momentum can all be represented by a high-dimensional mathematical vector, which is used to characterize the action. In this step, substituting the first action and second environmental momentum into the set formula above will yield the desired second action.

[0095] It should be noted that, in addition to obtaining the second action through the above-mentioned method, the second action can also be obtained through other methods, which are not limited to this.

[0096] In step S150 , after the second action is determined, the robot may execute the second action to complete the user's control over the robot.

[0097] In the present disclosure, after obtaining a first action corresponding to the instruction information based on a set instruction recognition model, the user's first evaluation result indicating dissatisfaction with the robot's first action specified based on the instruction information, as well as the corresponding instruction information and the first action, can be input into a pre-trained environmental momentum extractor. The environmental momentum extractor then outputs the corresponding first environmental momentum. Based on the first environmental momentum, an environmental momentum search is performed in a set database. If there is a set environmental momentum in the set database with a similarity greater than or equal to a similarity threshold to the first environmental momentum, the first action can be adjusted based on the set environmental momentum with the greatest similarity, thereby obtaining a second action. The robot can then execute the second action as a response to the instruction information. Due to the set environmental momentum representation, the user's evaluation result when the robot executes the set action based on the set instruction information is the set evaluation result. Therefore, adjusting the first action based on the appropriate set environmental momentum found can provide intelligent and flexible application possibilities for multiple scenarios, enabling the robot to adjust its behavior according to user needs and environmental changes, allowing the robot to provide more intelligent and adaptive services, thereby enhancing the application value of the robot in actual work and interaction.

[0098] In one exemplary embodiment, a robot control method is provided, applicable to a robot. In this method, if a setting database does not contain a setting environmental momentum whose similarity to a first environmental momentum is greater than or equal to a similarity threshold, this indicates that the robot cannot determine, based on the setting database, whether the user is satisfied with the first action when executing the first action based on the instruction information. In this case, the robot may execute the first action to complete the user's control of the robot.

[0099] After the robot performs the first action, refer to Figure 2As shown, the method may include:

[0100] S210, obtaining a target evaluation result through evaluation information of the first action;

[0101] S220, an environmental momentum extractor receives second input information and outputs a corresponding third environmental momentum, wherein the second input information includes the instruction information, the first action, and the target evaluation result; wherein the third environmental momentum represents that the user's evaluation result when the robot performs the first action based on the instruction information is the target evaluation result;

[0102] S230: Store the third environmental momentum as the set environmental momentum in the setting database.

[0103] In step S210 , after the robot performs the first action, an evaluation result may be determined in order to determine whether the first action meets the goal of the instruction information.

[0104] The robot may output a first prompt message. The first prompt message is used to prompt the user to evaluate the first action. For example, after the robot completes the first action, it may ask the user: "Please evaluate my action." The above-mentioned prompt message may be sent in the form of an audio message, an image, or text, without limitation.

[0105] After the user receives the first prompt from the robot, they can evaluate the first action through sound and movement. Simultaneously, in response to the first prompt, the robot can capture a video of the user's evaluation of the first action, lasting a set duration. This video, which includes both visual and audio information, captures the user's evaluation of the first action in response to the first prompt. The duration can be set based on actual needs, and there is no specific limit. For example, if the robot asks, "Please rate my action," it can record the user's video continuously for one minute and use this video as the evaluation video.

[0106] After the robot captures the evaluation video, it can determine a corresponding evaluation result based on the evaluation video. This evaluation result can be recorded as the target evaluation result. This evaluation result can generally be divided into a first evaluation result (e.g., "unsatisfactory") and a second evaluation result (e.g., "satisfactory"). It can also be set to include other evaluation results (e.g., "average") according to actual needs, without limitation.

[0107] It should be noted that "satisfied" indicates that the user is satisfied with the robot's execution of the first action based on the instruction information, and "unsatisfied" indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information.

[0108] When determining the target evaluation result based on the evaluation video, the image and audio information in the evaluation video can be extracted first. In other words, the image and audio information in the evaluation video are separated. Then, the corresponding prediction model is used to determine the evaluation result.

[0109] The robot can be pre-configured with two human satisfaction prediction models, designated as a first prediction model and a second prediction model. The first prediction model receives image information from an evaluation video and outputs a corresponding human satisfaction evaluation result, which can be recorded as an image evaluation result. The second prediction model receives audio information from an evaluation video and outputs a corresponding human satisfaction evaluation result, which can be recorded as an audio evaluation result.

[0110] After separating the image and audio information from the evaluation video, the image information can be input into a first prediction model. Upon receiving the image information, the first prediction model can output a corresponding image evaluation result. Alternatively, the audio information can be input into a second prediction model. Upon receiving the audio information, the second prediction model can output a corresponding audio evaluation result.

[0111] After obtaining the image evaluation result and the audio evaluation result, it can be determined whether the two evaluation results are the same. If the image evaluation result and the audio evaluation result are the same evaluation result (for example, both are the first evaluation result), the same evaluation result can be determined as the target evaluation result.

[0112] In the case where the image evaluation result and the audio evaluation result are different evaluation results (for example, the image evaluation result is the first evaluation result, and the audio evaluation result is the second evaluation result), it means that different evaluation results have been identified based on the image information and the audio information, and the robot can output the first prompt information again and re-determine the target evaluation result until the image evaluation result and the audio evaluation result are the same evaluation result, and the above-mentioned same evaluation result is used as the target evaluation result; or, until the target evaluation result determination process is completed a set number of times (for example, the second prompt information is output a set number of times). The set number of times can be set according to actual needs, and its specific value can be not limited. For example, if after the output of the second prompt information three times, the same image evaluation result and audio evaluation result cannot be obtained based on the collected evaluation video, then the determination of the target evaluation result is terminated, and it can be considered that the target evaluation result cannot be determined, and the process of this robot control method is terminated.

[0113] In some embodiments,

[0114] If the image and audio evaluation results are different, the robot may output a second prompt. The second prompt may indicate that the image and audio evaluation results are different. For example, the robot may output a voice message: "I'm sorry, I didn't understand what you meant."

[0115] In this embodiment, the number of times the second prompt information is output can be used as the number of times the target evaluation result is determined. That is, when the robot outputs the second prompt information, it can first determine whether the number of times the second prompt information is output reaches a set number (for example, three times).

[0116] If the number of outputs does not reach the set number, the robot can output the first prompt information again (for example, the voice message "Please evaluate my actions"), collect the evaluation video, and determine the image evaluation results and audio evaluation results, so as to re-judge whether the newly obtained image evaluation results and audio evaluation results are the same evaluation results (that is, judge whether the above-mentioned image evaluation results and audio evaluation results are the same). If they are the same, the same evaluation result will be used as the target evaluation result. If they are not the same, the second prompt information will be output. This cycle will continue until the number of outputs of the second prompt information reaches the set number, then it is considered that the target evaluation result cannot be obtained, and this process ends.

[0117] It should be noted that, in addition to determining the target evaluation results through the above-mentioned methods, the target evaluation results can also be determined through other methods, which are not limited to this.

[0118] In step S220, after the robot determines the target evaluation result, it uses the instruction information, the first action, and the target evaluation result as a set of input information, which can be recorded as the second input information. The second input information is then input into the environmental momentum extractor. After receiving the second input information, the environmental momentum extractor outputs the corresponding environmental momentum. This environmental momentum can be recorded as the third environmental momentum. The third environmental momentum can represent the user's evaluation result when the robot performs the first action based on the instruction information, which is the target evaluation result.

[0119] For example, the command information can be recorded as I, the action can be recorded as A, the evaluation result can be recorded as R, and the environmental momentum can be recorded as E. After the robot determines the command information, the first action, and the target evaluation result, it can input the corresponding (I, A, R) into the environmental momentum extractor, and the environmental momentum extractor can output the corresponding E.

[0120] Among them, when the types of evaluation results are divided into first evaluation results and second evaluation results, when training the environmental momentum extractor, it is also necessary to configure training sample pairs corresponding to the second evaluation result samples to obtain the required environmental momentum extractor.

[0121] When configuring a training sample pair corresponding to the second evaluation result sample, the environmental momentum sample in the training sample pair can be an environmental momentum representing an action adjustment of zero. In other words, the environmental momentum sample can be directly constructed based on the environmental momentum formation rules, such that the environmental momentum sample can represent the second evaluation result when the user evaluates the robot when it executes the first action sample based on the instruction information sample.

[0122] In step S230, after obtaining the third environmental momentum, the third environmental momentum is stored as the set environmental momentum in the setting database, thereby adding the new set environmental momentum to the setting database. If the robot subsequently receives similar instruction information, it will generally recognize a similar first action based on the instruction information. In this case, the robot can adjust the next action to be executed based on the third environmental momentum in the setting database.

[0123] It should be noted that if the target evaluation result is not determined after the robot performs the first action, the subsequent calculation and storage of the environmental momentum will no longer be performed.

[0124] This method incorporates real-time user feedback, establishing an iterative learning mechanism by asking users how satisfied they are with the robot's actions. It also proposes an environmental momentum extractor, which uses a neural network to extract environmental momentum from user instructions, robot actions, and evaluation results. By using this environmental momentum extractor and database retrieval, the robot can adjust current and future actions based on past user feedback, enabling personalized and adaptive behavior.

[0125] For example, robots can be used in smart collaborative office scenarios. The robot captures the user's voice commands through a microphone and uses cameras and sensors to perceive the office environment, such as the location of people and the movements of other robots. Using the methods of this application, the robot can perform tasks in a more intelligent and adaptive manner, such as delivering files or organizing filing cabinets.

[0126] For another example, in retail scenarios, robots can be used to provide personalized services. Using cameras and microphones, the robots can perceive the customer's appearance, voice commands, and feedback from the surrounding environment. Using the methods of this application, the robots can adjust their behavior based on the customer's personalized needs and environmental changes.

[0127] For another example, in medical scenarios, robots can be used to provide more intelligent medical care. The robot method of this application can perform tasks more flexibly, such as assisting patients with specific physical activities.

[0128] This environmental momentum-based robot control method offers intelligent and flexible application possibilities in a variety of scenarios. Its core concept is that the robot perceives feedback from its surroundings, such as sounds and movements, stores and processes this information into environmental momentum to influence its next action. Applications of this method include intelligent collaborative offices, personalized retail services, and healthcare. In these scenarios, the robot can adjust its behavior based on user needs and environmental changes, providing more intelligent and adaptive services, thereby enhancing its practical value in work and interaction.

[0129] In an exemplary embodiment, a robot control device is provided, which can be applied to a robot. The device can be used to implement the above-mentioned robot control method. For example, referring to Figure 3 As shown, the device may include:

[0130] Setting a command recognition model 10 for receiving command information and outputting a corresponding first action;

[0131] The environmental momentum extractor 20 is configured to receive first input information and output a corresponding first environmental momentum, the first input information including instruction information, a first action, and a first evaluation result for the first action; wherein the first evaluation result indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information; and the first environmental momentum indicates that the user's evaluation result when the robot executes the first action based on the instruction information is the first evaluation result.

[0132] The action determination module 30 is configured to determine a second environmental momentum based on the set environmental momentum, if a set environmental momentum exists in the set database and has a similarity greater than or equal to a similarity threshold with the first environmental momentum; wherein the set environmental momentum represents a user evaluation result when the robot performs the set action based on the set instruction information as the set evaluation result; and further configured to adjust the first action based on the second environmental momentum to obtain a second action;

[0133] The action execution module 40 is configured to execute a second action.

[0134] In one exemplary embodiment, a robot control device is provided, which can be applied to a robot. Figure 3 As shown, in the device, the action determination module 20 can be used to:

[0135] The product of the set adjustment coefficient and the second environmental momentum is determined as the action adjustment amount;

[0136] The first action is adjusted by the action adjustment amount to obtain the second action.

[0137] In one exemplary embodiment, a robot control device is provided, which can be applied to a robot. Figure 3 As shown, in the device, the action execution module 40 can be used to:

[0138] If there is no setting environmental momentum in the setting database whose similarity to the first environmental momentum is greater than or equal to the similarity threshold, a first action is performed.

[0139] In one exemplary embodiment, a robot control device is provided, which can be applied to a robot. Figure 3 As shown, the device includes an action evaluation module 50, wherein after the action execution module 40 executes the first action,

[0140] The action evaluation module 50 can be used to obtain a target evaluation result based on the evaluation information of the first action;

[0141] The environmental momentum extractor 20 is configured to receive second input information and output a corresponding third environmental momentum, wherein the second input information includes the instruction information, the first action, and the target evaluation result; wherein the third environmental momentum represents that the user's evaluation result when the robot performs the first action based on the instruction information is the target evaluation result;

[0142] The action evaluation module 50 may be configured to store the third environmental momentum as the set environmental momentum in the setting database.

[0143] In one exemplary embodiment, a robot control device is provided, which can be applied to a robot. Figure 3 As shown, in the device, the action evaluation module 50 can be used to:

[0144] Outputting first prompt information; wherein the first prompt information is used to prompt the user to evaluate the first action;

[0145] In response to outputting the first prompt information, collecting an evaluation video of a set length; wherein the evaluation video is a video of the user evaluating the first action in response to the first prompt information;

[0146] The target evaluation results are obtained by evaluating the video.

[0147] In one exemplary embodiment, a robot control device is provided, which can be applied to a robot. Figure 3 As shown, in the device, the action evaluation module 50 may include a first prediction model and a second prediction model.

[0148] An action evaluation module 50 can be used to extract image information and audio information from the evaluation video;

[0149] The first prediction model in the action evaluation module 50 can be used to receive the image information and output corresponding image evaluation results;

[0150] The second prediction model in the action evaluation module 50 can be used to receive the audio information and output corresponding audio evaluation results;

[0151] The motion evaluation module 50 may be configured to, when the image evaluation result and the audio evaluation result are identical, determine the identical evaluation result as a target evaluation result.

[0152] In one exemplary embodiment, a robot control device is provided, which can be applied to a robot. Figure 3 As shown, in the device, the action evaluation module 50 can be used to:

[0153] When the image evaluation result and the audio evaluation result are different evaluation results, outputting second prompt information; wherein the second prompt information indicates that the image evaluation result and the audio evaluation result are different evaluation results;

[0154] If the number of times the second prompt information is output does not reach the set number of times, the first prompt information is output again.

[0155] In an exemplary embodiment, a robot is provided. The robot may include a humanoid robot, an animal-shaped robot, or a robot of other shapes, without limitation.

[0156] refer to Figure 4 As shown, the robot 100 may include: at least one processor 101, a memory 102, at least one network interface 104, and another user interface 103. The various components of the robot 100 are coupled together via a bus system 105. It will be appreciated that the bus system 105 is used to enable communication between these components. In addition to a data bus, the bus system 105 also includes a power bus, a control bus, and a status signal bus. However, for clarity, all of these buses will be labeled as the bus system 105.

[0157] The user interface 103 may include a display, a keyboard, or a pointing robot (eg, a mouse, a trackball), a touch pad, or a touch screen.

[0158] It is understood that the memory 102 in the embodiment of the present application can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM bus random access memory (DRRAM). The memory 102 described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0159] In some embodiments, the memory 102 stores the following elements, executable units, or data structures, or a subset thereof, or an extended set thereof: an operating system 1021 and application programs 1022 .

[0160] Among them, the operating system 1021 includes various system programs, such as the framework layer, the core library layer, and the driver layer, which are used to implement various basic services and process hardware-based tasks. The application 1022 includes various application programs, such as a media player (Media Player), a browser (Browser), etc., which are used to implement various application services. The program implementing the method of the embodiment of the application can be included in the application 1022.

[0161] In the embodiment of the present application, the processor 101 is used to execute the methods provided in each method embodiment by calling the program or instructions stored in the memory 102, specifically, the program or instructions stored in the application 1022.

[0162] The methods disclosed in the above embodiments of the present application can be applied to or implemented by processor 101. Processor 101 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in processor 101 or by software instructions. The above processor 101 can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present application can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software units in the decoding processor. The software units can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory 102 , and the processor 101 reads the information in the memory 102 and implements the above method in combination with its hardware.

[0163] It is understood that the embodiments described herein may be implemented using hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit may be implemented in one or at least one application-specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing robot (DSPDevice, DSPD), programmable logic robot (PLD), field-programmable gate array (FPGA), general-purpose processor, controller, microcontroller, microprocessor, other electronic units for performing the functions described in the present application, or a combination thereof.

[0164] For software implementation, the technology described herein can be implemented by a unit that performs the functions described herein. The software code can be stored in a memory and executed by a processor. The memory can be implemented in the processor or outside the processor.

[0165] The present application also provides a storage medium (computer-readable storage medium). The storage medium stores one or at least one program. The storage medium may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive; and the memory may also include a combination of the aforementioned types of memory.

[0166] When one or at least one program in a storage medium is executable by one or at least one processor, the storage medium, when applied to a robot, can implement the aforementioned method executed on the robot. The processor is configured to execute the robot control program stored in the memory to implement the aforementioned method executed on the robot.

[0167] Professionals should also be further aware that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0168] It should be noted that references in this specification to "one embodiment," "an embodiment," "an exemplary embodiment," "some embodiments," and the like indicate that the described embodiments may include a particular feature, structure, or characteristic, but not necessarily every embodiment includes that particular feature, structure, or characteristic. Furthermore, such phrases do not necessarily refer to the same embodiment. Furthermore, when a particular feature, structure, or characteristic is described in conjunction with an embodiment, it is within the knowledge of those skilled in the art to implement such feature, structure, or characteristic in conjunction with other embodiments, whether explicitly described or not.

[0169] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or robot that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements that are inherent to such process, method, article or robot. In the absence of more limitations, an element defined by the sentence "comprises a..." does not exclude the presence of other identical elements in the process, method, article or robot that includes the element.

[0170] The above embodiments are only preferred embodiments for fully illustrating the present application, and the protection scope of the present application is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art based on the present application are within the protection scope of the present application.

Claims

1. A robot control method, characterized in that: The robot control method comprises: Setting a command recognition model to receive command information and output a corresponding first action; an environmental momentum extractor, receiving first input information and outputting a corresponding first environmental momentum, the first input information including the instruction information, the first action, and a first evaluation result for the first action; wherein the first evaluation result indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information; and the first environmental momentum indicates that the user's evaluation result when the robot executes the first action based on the instruction information is the first evaluation result; If there is a set environmental momentum in the set database whose similarity to the first environmental momentum is greater than or equal to a similarity threshold, determining a second environmental momentum based on the set environmental momentum; wherein the set environmental momentum represents a user evaluation result when the robot performs a set action based on the set instruction information, as a set evaluation result; adjusting the first action based on the second environmental momentum to obtain a second action; Perform the second action.

2. The robot control method according to claim 1, characterized in that: The adjusting the first action based on the second environmental momentum to obtain the second action includes: multiplying the set adjustment coefficient by the second environmental momentum to determine the action adjustment amount; The first action is adjusted according to the action adjustment amount to obtain the second action.

3. The robot control method according to claim 1 or 2, characterized in that: The robot control method comprises: If the setting database does not contain the setting environmental momentum whose similarity to the first environmental momentum is greater than or equal to the similarity threshold, the first action is performed.

4. The robot control method according to claim 3, characterized in that: After performing the first action, the robot control method includes: Obtaining a target evaluation result through evaluation information of the first action; The environmental momentum extractor receives second input information and outputs a corresponding third environmental momentum, wherein the second input information includes the instruction information, the first action, and the target evaluation result; wherein the third environmental momentum represents that the user's evaluation result when the robot performs the first action based on the instruction information is the target evaluation result; The third environmental momentum is stored in the setting database as the setting environmental momentum.

5. The robot control method according to claim 4, characterized in that: The evaluation information includes an evaluation video, and obtaining a target evaluation result based on the evaluation information of the first action includes: Outputting first prompt information; wherein the first prompt information is used to prompt the user to evaluate the first action; In response to outputting the first prompt information, collecting an evaluation video of a set length; wherein the evaluation video is a video of the user evaluating the first action in response to the first prompt information; The target evaluation result is obtained through the evaluation video.

6. The robot control method according to claim 5, characterized in that: Obtaining the target evaluation result through the evaluation video includes: Extracting image information and audio information from the evaluation video; a first prediction model, receiving the image information and outputting a corresponding image evaluation result; a second prediction model, receiving the audio information and outputting a corresponding audio evaluation result; When the image evaluation result and the audio evaluation result are the same evaluation result, the same evaluation result is determined as the target evaluation result.

7. The robot control method according to claim 6, characterized in that: The robot control method comprises: When the image evaluation result and the audio evaluation result are different evaluation results, outputting second prompt information; wherein the second prompt information indicates that the image evaluation result and the audio evaluation result are different evaluation results; When the output times of the second prompt information do not reach the set times, the first prompt information is output again.

8. A robot control device, characterized in that: The robot control device comprises: Setting a command recognition model for receiving command information and outputting a corresponding first action; an environmental momentum extractor, configured to receive first input information and output a corresponding first environmental momentum, the first input information including the instruction information, the first action, and a first evaluation result for the first action; wherein the first evaluation result indicates that the user is dissatisfied with the robot's execution of the first action based on the instruction information; and the first environmental momentum indicates that the user's evaluation result when the robot executes the first action based on the instruction information is the first evaluation result; an action determination module, configured to, if a set environmental momentum exists in a set database and has a similarity with the first environmental momentum greater than or equal to a similarity threshold, determine a second environmental momentum based on the set environmental momentum; wherein the set environmental momentum represents a user evaluation result when the robot performs a set action based on set instruction information, as a set evaluation result; and further configured to adjust the first action based on the second environmental momentum to obtain a second action; An action execution module is used to execute the second action.

9. A robot, characterized in that: The robot comprises: processor; a memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the robot control method according to any one of claims 1-7.

10. A storage medium, characterized in that: The storage medium stores one or at least one program, and the one or at least one program can be executed by one or at least one processor to implement the robot control method according to any one of claims 1 to 7.

Images