Growth type robot control method and device, storage medium and computer equipment

Abstract

The invention discloses a control method and device of a growth type robot, a storage medium and computer equipment, relates to the technical field of artificial intelligence and the technical field of medical treatment, and mainly aims at solving the problem that an existing robot lacks the reflecting ability and cannot process complex task instructions. Performing character conversion processing on the voice information to obtain a voice text; calling a matched target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a to-be-processed text; a large language model is adopted to generate reply content corresponding to the to-be-processed text, and if the reply content contains an execution step, an action instruction corresponding to the execution step is generated, so that an execution component of the robot executes a task based on the action instruction.

Description

Technical Field

[0001] The present invention relates to the fields of artificial intelligence technology and medical technology, and in particular to a control method and device, storage medium, and computer equipment for a growing robot. Background Art

[0002] With the development of artificial intelligence, the field of robotics has made significant progress, including industrial robots, service robots, medical robots, educational robots, etc. However, traditional robots have many problems in answering questions and performing tasks. For example, elderly care robots in the medical field usually have limited short-term memory and lack the ability to reflect, so they cannot handle complex task instructions, making the interaction between users and robots low. Summary of the invention

[0003] In view of this, the present invention provides a control method and device, storage medium, and computer equipment for a growing robot, the main purpose of which is to solve the problem that existing robots lack reflective ability and cannot process complex task instructions.

[0004] According to one aspect of the present invention, a control method for a growing robot is provided, comprising:

[0005] Receive the user's voice information and convert the voice information into text to obtain voice text;

[0006] Retrieving a matching target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a text to be processed;

[0007] A large language model is used to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, action instructions corresponding to the execution steps are generated so that the execution component of the robot performs the task based on the action instructions.

[0008] Furthermore, the memory stream database stores historical record data with timestamps and importance values;

[0009] The retrieving a matching target record from a memory stream database based on the voice text includes:

[0010] Performing semantic recognition processing on the speech text to obtain semantic information; and acquiring semantically relevant records from the historical record data based on the semantic information;

[0011] Obtaining the timestamp and the importance value of the semantically related record, and scoring the semantically related record based on the timestamp and the importance value to obtain a scoring value for each of the semantically related records;

[0012] The K semantically related records with the highest scoring values ​​are determined as the target records.

[0013] Furthermore, the method further comprises:

[0014] Collecting image data when the robot performs a task, and performing image recognition processing on the image data to obtain a recognition result corresponding to the image data;

[0015] The execution status of the robot is judged based on the recognition result to obtain an execution status judgment result.

[0016] Furthermore, after the execution status of the robot is judged based on the recognition result and the execution status judgment result is obtained, the method further includes:

[0017] After integrating the voice text, the reply content and the execution status judgment result, a record data is obtained;

[0018] Obtaining a preset importance analysis rule, and performing importance analysis processing on the execution situation judgment result based on the importance analysis rule to obtain an importance value corresponding to the record data;

[0019] A corresponding timestamp and the importance value are added to the record data, and the record data to which the timestamp and the importance value are added are saved in the memory stream database.

[0020] Furthermore, the method further comprises:

[0021] Determine the target history record data of performing the same task from the memory stream database based on the semantic information of the speech text;

[0022] Performing reflection processing on the target historical record data using the large language model according to a preset time period to obtain a reflection result corresponding to the execution task;

[0023] The target history record data is updated based on the reflection result to obtain an updated memory stream database.

[0024] Furthermore, after the target historical record data is updated based on the reflection result to obtain an updated memory stream database, the method further includes:

[0025] Generate a growth-type reply content and a growth-type action instruction corresponding to the user's voice information based on the updated memory stream database, so that the execution component of the robot performs the task based on the growth-type action instruction;

[0026] Collecting image data when the robot performs a task, and obtaining a corresponding execution status judgment result based on the image data;

[0027] Importance analysis is performed based on preset importance analysis rules and the execution status judgment result to obtain a growing importance value corresponding to the user voice information.

[0028] Furthermore, the method further comprises:

[0029] Obtaining a preset importance threshold, and comparing the growth importance value with the importance threshold;

[0030] If the growth-type importance value is greater than the importance threshold, the corresponding growth-type reply content is obtained; and information is extracted from the growth-type reply content to obtain a growth-type reply summary;

[0031] The relevant historical record data in the memory stream database is updated based on the growing response summary to obtain an updated memory stream database.

[0032] According to another aspect of the present invention, a control device for a growing robot is provided, comprising:

[0033] The receiving and converting module is used to receive the user's voice information and convert the voice information into text to obtain the voice text;

[0034] A retrieval and integration module, used for retrieving a matching target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a text to be processed;

[0035] A generation and execution module is used to use a large language model to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, action instructions corresponding to the execution steps are generated so that the robot's execution component performs the task based on the action instructions.

[0036] Furthermore, the memory stream database stores historical record data with timestamps and importance values; the retrieval and integration module is also used to:

[0037] Performing semantic recognition processing on the speech text to obtain semantic information; and acquiring semantically relevant records from the historical record data based on the semantic information;

[0038] Obtaining the timestamp and the importance value of the semantically related record, and scoring the semantically related record based on the timestamp and the importance value to obtain a scoring value for each of the semantically related records;

[0039] The K semantically related records with the highest scoring values ​​are determined as the target records.

[0040] Furthermore, the device further comprises an execution judgment module, which is used to:

[0041] Collecting image data when the robot performs a task, and performing image recognition processing on the image data to obtain a recognition result corresponding to the image data;

[0042] The execution status of the robot is judged based on the recognition result to obtain an execution status judgment result.

[0043] Furthermore, the device also includes a data storage module, which is used to:

[0044] After integrating the voice text, the reply content and the execution status judgment result, a record data is obtained;

[0045] Obtaining a preset importance analysis rule, and performing importance analysis processing on the execution situation judgment result based on the importance analysis rule to obtain an importance value corresponding to the recorded data;

[0046] A corresponding timestamp and the importance value are added to the record data, and the record data to which the timestamp and the importance value are added are saved in the memory stream database.

[0047] Furthermore, the device also includes a reflection module, which is used to:

[0048] Determine the target history record data for performing the same task from the memory stream database based on the semantic information of the speech text;

[0049] Performing reflection processing on the target historical record data using the large language model according to a preset time period to obtain a reflection result corresponding to the execution task;

[0050] The target history record data is updated based on the reflection result to obtain an updated memory stream database.

[0051] Furthermore, the device also includes a growth module, which is used to:

[0052] Generate a growth-type reply content and a growth-type action instruction corresponding to the user's voice information based on the updated memory stream database, so that the execution component of the robot performs the task based on the growth-type action instruction;

[0053] Collecting image data when the robot performs a task, and obtaining a corresponding execution status judgment result based on the image data;

[0054] Importance analysis is performed based on preset importance analysis rules and the execution status judgment result to obtain a growing importance value corresponding to the user voice information.

[0055] Furthermore, the growth module is also used for:

[0056] Obtaining a preset importance threshold, and comparing the growth importance value with the importance threshold;

[0057] If the growth-type importance value is greater than the importance threshold, the corresponding growth-type reply content is obtained; and information is extracted from the growth-type reply content to obtain a growth-type reply summary;

[0058] The relevant historical record data in the memory stream database is updated based on the growing response summary to obtain an updated memory stream database.

[0059] According to another aspect of the present invention, there is provided a storage medium, wherein the storage medium stores at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the control method of the growing robot as described above.

[0060] According to another aspect of the present invention, there is provided a computer device, comprising a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0061] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the control method of the growing robot as described above.

[0062] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0063] The present invention provides a control method and device, storage medium, and computer equipment for a growing robot. Compared with the prior art, the present invention enables the robot to acquire unique memory and reflection capabilities by combining a large language model and a memory stream database. Especially in the medical field, the robot in this field is more similar to the human thinking process and is convenient for processing complex task instructions. The present invention can provide a higher level of intelligent dialogue, improve the interactivity between users and robots, and improve user experience.

[0064] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0066] Figure 1 A schematic flow chart of a control method for a growing robot provided by an embodiment of the present invention is shown;

[0067] Figure 2 A schematic flow chart of another control method for a growing robot provided by an embodiment of the present invention is shown;

[0068] Figure 3 A schematic flow chart of another control method for a growing robot provided by an embodiment of the present invention is shown;

[0069] Figure 4 A schematic flow chart of another control method for a growing robot provided by an embodiment of the present invention is shown;

[0070] Figure 5 A schematic structural diagram of a control device for a growing robot provided by an embodiment of the present invention is shown;

[0071] Figure 6 A schematic structural diagram of another control device for a growing robot provided by an embodiment of the present invention is shown;

[0072] Figure 7 A schematic structural diagram of a computer device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0073] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0074] The embodiment of the present invention provides a control method for a growing robot, such as Figure 1 As shown, the method includes:

[0075] 101. Receive voice information from a user, and convert the voice information into text to obtain voice text;

[0076] In the embodiment of the present invention, the current execution end receives the user's voice information, which may be a command spoken by the user to the robot, or a communication dialogue between the user and the robot, etc., which is not specifically limited in the embodiment of the present invention. In the elderly care robot control scenario in the medical field, the voice information of the command type spoken by the user to the robot may be "put the water cup back on the table", etc., and the communication dialogue between the user and the robot may be "what food cannot be eaten with soy milk", etc., which is not specifically limited in the embodiment of the present invention.

[0077] In the embodiment of the present invention, after receiving the user's voice information, the current execution end converts the user's voice information into text to obtain a voice text. The text conversion process can adopt the existing ASR (Automatic Speech Recognition) speech-to-text technology, which is not specifically limited in the embodiment of the present invention.

[0078] 102. Retrieving a matching target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a text to be processed;

[0079] In an embodiment of the present invention, the current execution end retrieves the matching target record from the memory stream database based on the voice text obtained in step 101. The memory stream database stores long-term memory data, which can be used by the robot for reflection. The matching target record has similar semantics to the user's voice text. For example, in the elderly care robot control scenario in the medical field, the target record matching the voice text "Put the water cup back on the table" may include "Please put the water cup back on the table", "Put the water cup back on the table", etc. The embodiment of the present invention does not make specific limitations. In an embodiment of the present invention, the current execution end also integrates the obtained target record with the user's voice text to obtain a text to be processed.

[0080] 103. Use a large language model to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, generate action instructions corresponding to the execution steps, so that the execution component of the robot performs the task based on the action instructions.

[0081] In the embodiment of the present invention, since the target record matched with the speech text contains not only the speech text information, but also the corresponding reply content or task execution status, the current execution end can generate reply content corresponding to the text to be processed by using a large language model. Among them, the large language model can be implemented by BERT, GPT-4, LaMDA, LLaMA, etc., which is not specifically limited in the embodiment of the present invention.

[0082] Since the voice information received in step 101 can be a command spoken by the user to the robot, or a communication dialogue between the user and the robot, the generated reply content may include execution steps, or may only include the robot's reply content, etc., which is not specifically limited in the embodiment of the present invention. After receiving the reply content, the current execution end also needs to judge and process the reply content. If the reply content does not include execution steps, the reply content is directly fed back to the user; if the reply content includes execution steps, action instructions corresponding to the execution steps need to be generated. When each execution component of the robot receives an action instruction, it executes the task in the order of the instruction.

[0083] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to obtain target record data with a higher matching degree from the memory stream database, another control method for a growing robot is provided, in which the memory stream database stores historical record data with timestamps and importance values, such as Figure 2 As shown, the step of retrieving a matching target record from a memory stream database based on the voice text includes:

[0084] 201. Perform semantic recognition processing on the speech text to obtain semantic information; and obtain semantically relevant records from the historical record data based on the semantic information;

[0085] In an embodiment of the present invention, the current execution end performs semantic recognition processing on the speech text obtained in step 101 to obtain semantic information of the user. The semantic recognition processing can be implemented using a semantic recognition model. Semantic recognition models include RNN, LSTM, BiLSTM, etc., which are not specifically limited in the embodiment of the present invention. After obtaining the semantic information, the current execution end also screens the historical record data in the memory stream database based on the obtained semantic information to obtain semantic-related records related to the semantic information.

[0086] It should be noted that the data volume of semantically relevant records obtained through semantic correlation is likely to be still very large, so the semantically relevant records need to be further screened.

[0087] 202. Obtain the timestamp and the importance value of the semantically related record, and score the semantically related record based on the timestamp and the importance value to obtain a score value of each of the semantically related records;

[0088] 203. Determine the K semantically related records with the highest scoring values ​​as the target records.

[0089] In an embodiment of the present invention, since the historical record data also includes timestamp and importance value information, when filtering the semantically related records, the current execution end can score the semantically related records based on the acquired timestamp and importance value to obtain the score value of each semantically related record. Among them, the scoring process can be performed according to the preset scoring rules, such as determining the time sequence of each semantically related record according to the timestamp, and the semantically related record closer to the current time has a higher score when scoring; sorting from large to small according to the importance value, and the larger the importance value is, the higher the score is, etc., which is not specifically limited in the embodiment of the present invention. After the scoring process, the current execution end determines the K semantically related records with the highest scoring values ​​as the target records. Among them, the value of K can be set as needed, such as setting the K value to 2, 3, 4, 5, etc., which is not specifically limited in the embodiment of the present invention.

[0090] Furthermore, as a refinement and extension of the specific implementation of the above embodiment, in order to analyze and process the execution of the robot, another control method of the growing robot is provided, such as Figure 3 As shown, the method also includes:

[0091] 301. Collect image data when the robot performs a task, and perform image recognition processing on the image data to obtain a recognition result corresponding to the image data;

[0092] In an embodiment of the present invention, the current execution end collects image data when the robot performs a task, such as in the elderly care robot control scenario in the medical field, collecting images of the robot putting the water cup back on the table when performing a task, etc., which is not specifically limited in the embodiment of the present invention. Afterwards, the current execution end performs image recognition processing on the collected image data to obtain a recognition result corresponding to the image data. For example, the water cup and the table are identified from the image data and labeled accordingly. The specific image recognition processing can be implemented using YOLO vision, and can also be implemented using a VGG model, a ResNet model, etc., which is not specifically limited in the embodiment of the present invention.

[0093] 302. Based on the recognition result, the execution status of the robot is judged to obtain an execution status judgment result.

[0094] In the embodiment of the present invention, the current execution end judges and processes the execution status of the robot based on the recognition result obtained in step 302, such as judging whether the returned position is correct, whether the picked-up item is correct, etc., which is not specifically limited in the embodiment of the present invention.

[0095] It should be noted that, in order to use the execution status as data for reflection processing so that the robot can provide more human-like memory and reflection functions, after the step of judging the execution status of the robot based on the recognition result and obtaining the execution status judgment result, the method further includes:

[0096] 303. After integrating the voice text, the reply content and the execution status judgment result, a record data is obtained;

[0097] In the embodiment of the present invention, the current execution end also integrates the user's voice text, the reply content generated by the large language model, and the execution status judgment result after the robot completes the task, and obtains a record data. The record data can reflect the information of the entire task in the three stages of receiving commands, generating execution processes, and executing results.

[0098] 304. Obtaining a preset importance analysis rule, and performing importance analysis processing on the execution status judgment result based on the importance analysis rule to obtain an importance value corresponding to the record data;

[0099] In the embodiment of the present invention, the current execution end also obtains a preset importance analysis rule, and performs importance analysis on the execution judgment results obtained from step 301 to step 302 based on the obtained importance analysis rule, and obtains an importance value corresponding to the recorded data integrated in step 303. For example, whether the object is picked up correctly is analyzed and processed to obtain a first degree value; then, the walking path of the robot is analyzed and processed to obtain a second degree value; the placement of the object is analyzed and processed to obtain a third degree value; the time spent on the task execution process is analyzed and processed to obtain a fourth degree value; finally, the first degree value, the second degree value, the third degree value and the fourth degree value obtained above are weighted and summed to obtain the corresponding importance value, etc., which is not specifically limited in the embodiment of the present invention.

[0100] 305. Add a corresponding timestamp and the importance value to the record data, and save the record data with the timestamp and the importance value added to the memory stream database.

[0101] In an embodiment of the present invention, the current execution end obtains the timestamp when the task is executed, and adds the obtained timestamp and the importance value obtained in step 304 to the corresponding record data, and saves the record data with the added timestamp and importance value as historical record data in the memory stream database, so as to perform reflection processing based on the memory stream database with long-term memory.

[0102] Further, as a refinement and extension of the above-mentioned specific implementation methods, in order to enable the robot to provide more human-like memory and reflection functions, another control method for the growing robot is provided, such as Figure 4 As shown, the method also includes:

[0103] 401. Determine target history record data for performing the same task from the memory stream database based on semantic information of the speech text;

[0104] In an embodiment of the present invention, the current execution end determines the target historical record data for performing the same task from the memory stream database based on the semantic information of the voice text. For example, in the elderly care robot control scenario in the medical field, the historical record data corresponding to the task of the robot putting a water cup back on the table is determined as the target historical record data, or the historical record data corresponding to the task of the robot turning off the bedroom light is determined as the target historical record data, etc. The embodiment of the present invention does not make specific limitations.

[0105] 402. Performing reflection processing on the target historical record data using the large language model according to a preset time period to obtain a reflection result corresponding to the execution task;

[0106] In the embodiment of the present invention, the current execution end uses the large language model of step 103 to reflect on the target historical record data according to a preset time period to obtain a reflection result corresponding to the execution task. For example, when the target historical record data corresponding to the task of the robot putting the cup back on the table in step 401 is reflected, the voice text information in the target historical record data is input into the large language model for reflection, thereby abstracting a higher level of problem. It should be noted that prompt words can be used for guidance during reflection, and the embodiment of the present invention does not make specific limitations.

[0107] 403. Update the target history record data based on the reflection result to obtain an updated memory stream database.

[0108] In the embodiment of the present invention, the current execution end adds the reflection result obtained in step 402 as new information to the target historical record data determined in step 401, updates the target historical record data, and obtains an updated memory stream database.

[0109] It should be noted that, in order to reflect on the answer content based on the reflection on the question, the step of updating the target historical record data based on the reflection result, after obtaining the updated memory stream database, the method further includes:

[0110] 404. Generate a growth-type reply content and a growth-type action instruction corresponding to the user voice information based on the updated memory stream database, so that the execution component of the robot performs the task based on the growth-type action instruction;

[0111] 405. Collect image data when the robot performs a task, and obtain a corresponding execution status judgment result based on the image data;

[0112] 406. Perform importance analysis based on preset importance analysis rules and the execution status judgment result to obtain a growing importance value corresponding to the user voice information.

[0113] In the embodiment of the present invention, the current execution end uses the method of step 103 to regenerate the growth-type reply content and growth-type action instructions corresponding to the user's voice information based on the memory stream database updated from step 401 to step 403, so that the execution component of the robot performs the task based on the sequence of the growth-type action instructions; then, the same method as steps 301 to 302 is used to collect image data when the robot performs the task, and the corresponding execution status judgment result is obtained based on the collected image data. Then, the same method as step 304 is used to perform importance analysis based on the preset importance analysis rules and the execution status judgment result obtained in step 405 to obtain the growth-type importance value corresponding to the user's voice information.

[0114] Further, as a refinement and extension of the specific implementation of the above embodiment, in order to reflect on the answer content based on the reflection on the question, another control method of the growing robot is provided, and the method also includes:

[0115] Obtaining a preset importance threshold, and comparing the growth importance value with the importance threshold;

[0116] If the growth-type importance value is greater than the importance threshold, the corresponding growth-type reply content is obtained; and information is extracted from the growth-type reply content to obtain a growth-type reply summary;

[0117] The relevant historical record data in the memory stream database is updated based on the growing response summary to obtain an updated memory stream database.

[0118] In an embodiment of the present invention, the current execution end obtains a preset importance value, and compares the growth importance value obtained from step 404 to step 406 with the preset importance value; when the growth importance value is greater than the preset importance threshold, it indicates that the completion degree of the robot in executing the task meets the preset requirements, and then the corresponding growth reply content is obtained, and the growth reply content is subjected to information extraction processing to obtain a growth reply summary. For example, the execution step summary is extracted from the growth reply content, etc., which is not specifically limited in the embodiment of the present invention. After obtaining the growth reply summary, the current execution end needs to add the obtained growth reply summary to the memory stream database, that is, update the relevant historical record data in the memory stream database based on the growth reply summary to obtain an updated memory stream database. Through reflection, the embodiment of the present invention can abstract high-level problems from direct memory and put the answers back into the memory stream. When reflecting next time, not only direct memory can be retrieved, but also the information of reflection can be retrieved for deeper thinking.

[0119] The embodiment of the present invention provides a control method for a growing robot. Compared with the prior art, the present invention enables the robot to acquire unique memory and reflection capabilities by combining a large language model and a memory stream database. Especially in the medical field, the robot in this field is more similar to the human thinking process and is convenient for processing complex task instructions. The present invention can provide a higher level of intelligent dialogue, improve the interactivity between users and robots, and improve user experience.

[0120] As the above Figure 1 The embodiment of the present invention provides a control device for a growing robot, such as Figure 5 As shown, the device comprises:

[0121] The receiving and converting module 51 is used to receive the user's voice information and convert the voice information into text to obtain the voice text;

[0122] A retrieval and integration module 52 is used to retrieve a matching target record from a memory stream database based on the voice text, and integrate the target record with the voice text to obtain a text to be processed;

[0123] The generation and execution module 53 is used to use a large language model to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, action instructions corresponding to the execution steps are generated so that the execution component of the robot performs the task based on the action instructions.

[0124] Furthermore, the memory stream database stores historical record data with timestamps and importance values; the retrieval and integration module 52 is also used to:

[0125] Performing semantic recognition processing on the speech text to obtain semantic information; and acquiring semantically relevant records from the historical record data based on the semantic information;

[0126] Obtaining the timestamp and the importance value of the semantically related record, and scoring the semantically related record based on the timestamp and the importance value to obtain a scoring value for each of the semantically related records;

[0127] The K semantically related records with the highest scoring values ​​are determined as the target records.

[0128] Furthermore, if Figure 6 As shown, the device further includes an execution judgment module 54, which is used to:

[0129] Collecting image data when the robot performs a task, and performing image recognition processing on the image data to obtain a recognition result corresponding to the image data;

[0130] The execution status of the robot is judged based on the recognition result to obtain an execution status judgment result.

[0131] Furthermore, if Figure 6 As shown, the device further includes a data storage module 55, which is used to:

[0132] After integrating the voice text, the reply content and the execution status judgment result, a record data is obtained;

[0133] Obtaining a preset importance analysis rule, and performing importance analysis processing on the execution situation judgment result based on the importance analysis rule to obtain an importance value corresponding to the recorded data;

[0134] A corresponding timestamp and the importance value are added to the record data, and the record data to which the timestamp and the importance value are added are saved in the memory stream database.

[0135] Furthermore, if Figure 6 As shown, the device further includes a reflection module 56, which is used to:

[0136] Determine the target history record data for performing the same task from the memory stream database based on the semantic information of the speech text;

[0137] Performing reflection processing on the target historical record data using the large language model according to a preset time period to obtain a reflection result corresponding to the execution task;

[0138] The target history record data is updated based on the reflection result to obtain an updated memory stream database.

[0139] Furthermore, if Figure 6 As shown, the device further comprises a growth module 57, which is used for:

[0140] Generate a growth-type reply content and a growth-type action instruction corresponding to the user's voice information based on the updated memory stream database, so that the execution component of the robot performs the task based on the growth-type action instruction;

[0141] Collecting image data when the robot performs a task, and obtaining a corresponding execution status judgment result based on the image data;

[0142] Importance analysis is performed based on preset importance analysis rules and the execution status judgment result to obtain a growing importance value corresponding to the user voice information.

[0143] Furthermore, the growth module 57 is also used for:

[0144] Obtaining a preset importance threshold, and comparing the growth importance value with the importance threshold;

[0145] If the growth-type importance value is greater than the importance threshold, the corresponding growth-type reply content is obtained; and information is extracted from the growth-type reply content to obtain a growth-type reply summary;

[0146] The relevant historical record data in the memory stream database is updated based on the growing response summary to obtain an updated memory stream database.

[0147] The embodiment of the present invention provides a control device for a growing robot. Compared with the prior art, the present invention enables the robot to acquire unique memory and reflection capabilities by combining a large language model and a memory stream database. Especially in the medical field, the robot in this field is more similar to the human thinking process and is convenient for processing complex task instructions. The present invention can provide a higher level of intelligent dialogue, improve the interactivity between users and robots, and improve user experience.

[0148] According to one embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the control method of a growing robot in any of the above method embodiments.

[0149] Figure 7 A schematic diagram of the structure of a computer device provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the computer device.

[0150] like Figure 7 As shown, the computer device may include: a processor (processor) 602 , a communication interface (Communications Interface) 604 , a memory (memory) 606 , and a communication bus 608 .

[0151] The processor 602 , the communication interface 604 , and the memory 606 communicate with each other via a communication bus 608 .

[0152] The communication interface 604 is used to communicate with other devices such as clients or other servers.

[0153] The processor 602 is used to execute the program 610, and specifically can execute the relevant steps of the control method of the above-mentioned growing robot.

[0154] Specifically, the program 610 may include program codes, which include computer operation instructions.

[0155] The processor 602 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present invention. The one or more processors included in the computer device may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0156] The memory 606 is used to store the program 610. The memory 606 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0157] The program 610 may be specifically configured to enable the processor 602 to perform the following operations:

[0158] Receive the user's voice information and convert the voice information into text to obtain voice text;

[0159] Retrieving a matching target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a text to be processed;

[0160] A large language model is used to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, action instructions corresponding to the execution steps are generated so that the execution component of the robot performs the task based on the action instructions.

[0161] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0162] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention. The non-Company software tools or components appearing in the embodiments of the present invention are only examples and do not represent actual use.

Claims

1. A control method for a growing robot, characterized in that: include: Receive the user's voice information and convert the voice information into text to obtain voice text; Retrieving a matching target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a text to be processed; A large language model is used to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, action instructions corresponding to the execution steps are generated so that the execution component of the robot performs the task based on the action instructions.

2. The method according to claim 1, characterized in that The memory stream database stores historical record data with timestamps and importance values; The retrieving a matching target record from a memory stream database based on the voice text includes: Performing semantic recognition processing on the speech text to obtain semantic information; and acquiring semantically relevant records from the historical record data based on the semantic information; Obtaining the timestamp and the importance value of the semantically related record, and scoring the semantically related record based on the timestamp and the importance value to obtain a scoring value for each of the semantically related records; The K semantically related records with the highest scoring values ​​are determined as the target records.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: Collecting image data when the robot performs a task, and performing image recognition processing on the image data to obtain a recognition result corresponding to the image data; The execution status of the robot is judged based on the recognition result to obtain an execution status judgment result.

4. The method according to claim 3, characterized in that After the execution status of the robot is judged based on the recognition result and the execution status judgment result is obtained, the method further includes: After integrating the voice text, the reply content and the execution status judgment result, a record data is obtained; Obtaining a preset importance analysis rule, and performing importance analysis processing on the execution situation judgment result based on the importance analysis rule to obtain an importance value corresponding to the recorded data; A corresponding timestamp and the importance value are added to the record data, and the record data to which the timestamp and the importance value are added are saved in the memory stream database.

5. The method according to claim 1, characterized in that The method further comprises: Determine the target history record data for performing the same task from the memory stream database based on the semantic information of the speech text; Performing reflection processing on the target historical record data using the large language model according to a preset time period to obtain a reflection result corresponding to the execution task; The target history record data is updated based on the reflection result to obtain an updated memory stream database.

6. The method according to claim 5, characterized in that After the target historical record data is updated based on the reflection result to obtain an updated memory stream database, the method further includes: Generate a growth-type reply content and a growth-type action instruction corresponding to the user's voice information based on the updated memory stream database, so that the execution component of the robot performs the task based on the growth-type action instruction; Collecting image data when the robot performs a task, and obtaining a corresponding execution status judgment result based on the image data; Importance analysis is performed based on preset importance analysis rules and the execution status judgment result to obtain a growing importance value corresponding to the user voice information.

7. The method according to claim 6, characterized in that The method further comprises: Obtaining a preset importance threshold, and comparing the growth importance value with the importance threshold; If the growth-type importance value is greater than the importance threshold, the corresponding growth-type reply content is obtained; and information is extracted from the growth-type reply content to obtain a growth-type reply summary; The relevant historical record data in the memory stream database is updated based on the growing response summary to obtain an updated memory stream database.

8. A control device for a growing robot, characterized in that: include: The receiving and converting module is used to receive the user's voice information and convert the voice information into text to obtain the voice text; A retrieval and integration module, used for retrieving a matching target record from a memory stream database based on the voice text, and integrating the target record with the voice text to obtain a text to be processed; A generation and execution module is used to use a large language model to generate reply content corresponding to the text to be processed. If the reply content contains execution steps, action instructions corresponding to the execution steps are generated so that the robot's execution component performs the task based on the action instructions.

9. A storage medium, wherein at least one executable instruction is stored in the storage medium, and the executable instruction executes an operation corresponding to the control method of a growing robot as described in any one of claims 1-7.

10. A computer device, comprising a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the control method of the growing robot as described in any one of claims 1-7.

Images