Human-computer interaction-based ai robot intelligent control system and method
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- NINGBO SMART XINNENG TECHNOLOGY CO LTD
- Filing Date
- 2026-05-15
- Publication Date
- 2026-07-10
AI Technical Summary
AI robots are prone to entering a state of mechanical resonance during autonomous optimization control, which can cause end-effector vibration and robotic arm tremor, endangering safety.
By analyzing the historical operation records of AI robots, the risk of abnormal mechanical vibration is assessed, control strategies are optimized to reduce the risk of abnormality, and safety control is achieved through human-computer interaction.
It effectively reduces the probability of safety accidents caused by mechanical resonance, maintains the operating efficiency and compliance of AI robots, and achieves a balance between safety and performance.
Smart Images
Figure CN122353597A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of robot intelligent control technology, specifically to an AI robot intelligent control system and method based on human-computer interaction. Background Technology
[0002] With the continuous development of AI technology, its application in the field of industrial robots has become increasingly widespread. Traditional robots can only operate according to preset fixed parameters, while AI robots using AI technology can automatically optimize their motion trajectory, joint impedance characteristics, etc. in real time according to task objectives, thereby improving their working performance.
[0003] However, a potential problem exists in the actual automatic optimization of motion trajectory and joint impedance characteristics by AI robots. When the AI robot runs according to the preset control strategy, it will autonomously generate and execute a set of control strategies that it considers to be the optimal stiffness, damping, and motion parameters. However, when the AI robot runs according to the seemingly optimal control strategy, it may quietly enter a state of severe mechanical resonance under certain human-machine physical interaction conditions. Once the AI robot enters a state of severe mechanical resonance, it can easily cause the end tool to shake violently or the entire robotic arm to vibrate continuously. The amplitude can diverge to a dangerous level in a very short time, which can not only seriously damage the joint transmission and structural components of the AI robot, but may even directly impact the workers and pose a serious safety threat to them. Summary of the Invention
[0004] The purpose of this invention is to provide an AI robot intelligent control system and method based on human-computer interaction to solve the problems raised in the prior art.
[0005] To achieve the above objectives, the present invention provides the following technical solution: an AI robot intelligent control method based on human-computer interaction, the method comprising: Step S1: Obtain the historical operation records of the AI robot, analyze the degree of mechanical vibration abnormality of the AI robot in the historical operation records, and obtain the abnormal operation records; Step S2: Obtain the control strategy of the AI robot, and combine it with the marked abnormal operation records to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy, and obtain the risk control strategy; Step S3: Analyze the degree of abnormal risk contribution of control parameters in the risk control strategy, determine the abnormal control parameters, evaluate the effect of abnormal control parameters on the mechanical vibration abnormality correction of AI robot under different correction directions and correction scales, and perform strategy correction and optimization of risk control strategy to obtain the target control strategy of AI robot. Step S4: Control the operation of the AI robot according to the target control strategy, and use the AI robot to interact with the staff.
[0006] Further, step S3 includes: Obtain the risk control strategy of the AI robot in the current cycle, and obtain the characteristic historical control strategy of the risk control strategy; Calculate the data deviation Δv between a certain control parameter in the risk control strategy and the characteristic historical control strategy, set the risk direction variable F of a certain control parameter of the AI robot, where F ∈ {+1, -1}, and calculate the risk contribution value P = Δv·F of a certain control parameter in the risk control strategy; Set the risk contribution threshold p. When P > p, it is determined that there is a high-risk abnormal risk in a certain control parameter of the risk control strategy, and a certain control parameter is recorded as an abnormal control parameter. Otherwise, a certain control parameter is not processed, where p > 0; Obtain the comparison historical operation record of the AI robot, obtain each similar comparison historical control strategy of the risk control strategy, obtain the safety reference value of a certain abnormal control parameter in the risk control strategy, and obtain the reference difference d of a certain abnormal control parameter; Set the correction direction variable K of a certain abnormal control parameter. When d > 0, perform an increase operation on a certain abnormal control parameter, K = +1. When d ≤ 0, perform a decrease operation on a certain abnormal control parameter, K = -1; Obtain the preset step candidate value η´ of a certain abnormal control parameter from the platform, and calculate the correction step η of a certain abnormal control parameter; Calculate the initial correction value Y = X + η·K of a certain abnormal control parameter, obtain the initial correction values of each abnormal control parameter, and construct the initial correction strategy vector of the risk control strategy. Calculate the vibration abnormal risk value G of the AI robot under the initial correction strategy vector Y , , and obtain the vibration abnormal risk value G of the AI robot and the second risk threshold G2; Set the correction adjustment coefficient β, where 1 > β > 0. When G Y ≥ G2, re-obtain the initial correction values of each abnormal control parameter until G Y < G2, and obtain the initial correction values of each corrected abnormal control parameter for aggregation to obtain the target correction strategy vector; When G Y < G2, then record the initial correction strategy vector as the target correction strategy vector; According to the target correction strategy vector, perform strategy correction on the risk control strategy of the AI robot to obtain the target control strategy; In the above steps, by analyzing the degree of abnormal danger contribution of different control parameters in the control strategy, the key parameters that cause the abnormality of the AI robot, that is, the abnormal control parameters, are clearly identified. And through the real resonance records and operation records of the AI robot's own history, the abnormal control parameters in the risk control strategy are gradually adjusted, rather than the traditional one-time reconstruction, so as to achieve the maximum improvement in safety at the cost of the smallest performance sacrifice and realize the effective balance between performance and safety.
[0007] Further, step S2 includes: Obtain the control strategy generated by the AI robot in the current cycle, obtain the data of each control parameter from the control strategy, and construct the strategy vector V´ of the control strategy; Obtain the historical control strategy in the marked abnormal operation record of the AI robot, and construct the strategy vector V of the historical control strategy; Calculate the strategy similarity value S between the control strategy in the current cycle and the historical control strategy in the marked abnormal operation record, set the strategy similarity threshold S´. When S > S´, it is determined that the control strategy in the current cycle is similar to the historical control strategy in the marked abnormal operation record; when S ≤ S´, it is determined that the control strategy in the current cycle is not similar to the historical control strategy in the marked abnormal operation record; Obtain the total number C of several marked abnormal operation records similar to the control strategy of the AI robot, and calculate the vibration abnormal risk value G = C / C of the AI robot under the control strategy sum , where C sum represents the total number of marked abnormal operation records of the AI robot; Set the first risk threshold G1 and the second risk threshold G2. When G > G1, it is determined that the AI robot has a mechanical vibration abnormal risk under the control strategy, and the control strategy of the AI robot in the current cycle is recorded as the risk control strategy, where G1 > G2 > 0; When G2 ≤ G ≤ G1, obtain the cumulative sum S of the strategy similarity values between the control strategy in the current cycle and the historical control strategies in several marked abnormal operation records sum , set the cumulative threshold S α ; When S sum > S α it is determined that the AI robot has a mechanical vibration abnormal risk under the control strategy, and the control strategy of the AI robot in the current cycle is recorded as the risk control strategy; otherwise, it is determined that the AI robot does not have a mechanical vibration abnormal risk under the control strategy; When G < G2, it is determined that the AI robot does not have a mechanical vibration abnormal risk under the control strategy.
[0008] Further, step S1 includes: Set the sampling interval duration Δt and the sampling feature value N, obtain the historical operation record of the AI robot, and sample the vibration acceleration of the AI robot N times at each sampling interval duration from the historical operation record to obtain the set of vibration acceleration of the AI robot within the sampling interval duration in the historical operation record; Set the characteristic duration T and vibration threshold R, and calculate the root mean square value R of the vibration acceleration of the AI robot during the a-th sampling interval in the historical operation records. a When R a When the value is greater than R, the duration of the a-th sampling interval is recorded as the abnormal sampling interval duration; otherwise, the duration of the a-th sampling interval in the historical operation record is not processed. Obtain the total number n of several adjacent abnormal sampling intervals from the historical operation records, and calculate the duration T of mechanical vibration in the historical operation records. sum =n×△t, when T sum When T >, the AI robot's mechanical vibration is determined to be abnormal in the historical operation record, and the historical time period consisting of several abnormal sampling intervals is obtained to obtain the characteristic historical time period in the historical operation record. Extract the operation records of the AI robot within a specific historical period from the historical operation records to obtain the AI robot's marked abnormal operation records.
[0009] Furthermore, step S4 includes: Obtain the target control strategy of the AI robot in the current cycle, and extract the values of various control parameters from the target control strategy; The system acquires the preset sampling interval and vibration threshold R, sets the control parameters of the AI robot according to the values of the control parameters in the target control strategy, and uses the AI robot to interact with the staff. It also monitors the root mean square value of the vibration acceleration of the AI robot in each sampling interval in real time. When the root mean square value of the vibration acceleration of the AI robot in any sampling interval within the current cycle is greater than the vibration threshold R, the AI robot stops running and maintenance personnel are dispatched to inspect the AI robot.
[0010] To better implement the above methods, an AI robot intelligent control system is also proposed, which includes an anomaly analysis module, a control strategy risk assessment module, a strategy correction module, and an operation control module. The anomaly analysis module is used to acquire the historical operation records of the AI robot and analyze the degree of mechanical vibration anomalies in the historical operation records to obtain marked abnormal operation records. The control strategy risk assessment module is used to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy, and obtain the risk control strategy. The strategy correction module is used to correct and optimize the risk control strategy to obtain the target control strategy for the AI robot. The operation control module is used to control the operation of the AI robot according to the target control strategy and to enable the AI robot to interact with staff.
[0011] Furthermore, the anomaly analysis module includes an anomaly duration analysis unit and an anomaly recording analysis unit; The abnormal duration analysis unit is used to determine the degree of abnormality of vibration acceleration within the sampling interval of the AI robot in the historical operation record, and to determine the abnormal sampling interval duration of the AI robot. The anomaly analysis unit is used to acquire the duration of abnormal sampling intervals in the historical operation records of the AI robot, analyze the degree of abnormal mechanical vibration of the AI robot in the historical operation records, and obtain marked abnormal operation records.
[0012] Furthermore, the control strategy risk assessment module includes a strategy similarity analysis unit and a control strategy risk assessment unit; The strategy similarity analysis unit is used to analyze the similarity between the control strategy in the current period and the historical control strategies in the marked abnormal operation records, and to obtain the marked abnormal operation records that are similar to the control strategy. The control strategy risk assessment unit is used to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy in the current cycle, and to obtain the risk control strategy.
[0013] Furthermore, the strategy correction module includes an abnormal parameter determination unit and a strategy correction unit; The abnormal parameter determination unit is used to analyze the degree of contribution of abnormal risks to control parameters in risk control strategies and determine abnormal control parameters. The strategy correction unit is used to evaluate the effect of abnormal mechanical vibration correction on the AI robot under different correction directions and scales of abnormal control parameters, and to perform strategy correction and optimization on the risk control strategy to obtain the target control strategy of the AI robot.
[0014] Furthermore, the operation control module includes an operation control unit; The operation control unit is used to acquire the target control strategy of the AI robot in the current cycle, obtain the values of various control parameters from the target control strategy, set various control parameters of the AI robot according to the values of various control parameters in the target control strategy, and use the AI robot to interact with the staff.
[0015] Compared with existing technologies, the beneficial effects of this invention are: it solves the problem of mechanical resonance caused by AI robots in autonomous optimization. Traditional methods can only passively respond by suppressing resonance with filters or stopping the robot after it occurs. At this time, the vibration energy has already been released, which can easily cause harm to the staff who interact with the AI robot. However, this invention uses the historical operation records of the AI robot to judge the risk of abnormal mechanical vibration of the control strategy generated by the AI robot, and makes a fine balance between the safety and operation performance of the AI robot. This not only preserves the optimization results of the AI controller in terms of compliance and operating efficiency to the greatest extent, but also greatly reduces the probability of safety accidents caused by mechanical resonance of the AI robot, thus achieving safe control of the AI robot. Attached Figure Description
[0016] Figure 1 This is a flowchart of the strategy risk assessment process for the AI robot intelligent control method based on human-computer interaction according to the present invention. Figure 2 This is a flowchart of the modules of an AI robot intelligent control system according to the present invention. Detailed Implementation
[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0018] Example: Figures 1-2 As shown, the present invention provides a technical solution, an AI robot intelligent control method based on human-computer interaction, the method comprising: Step S1: Obtain the historical operation records of the AI robot, analyze the degree of mechanical vibration abnormality of the AI robot in the historical operation records, and obtain the abnormal operation records; Step S1 includes: Set the sampling interval duration Δt and the sampling feature value N, obtain the historical operation record of the AI robot, and sample the vibration acceleration of the AI robot N times at each sampling interval duration from the historical operation record to obtain the set of vibration acceleration of the AI robot within the sampling interval duration in the historical operation record; Set the characteristic duration T and vibration threshold R, and calculate the root mean square value R of the vibration acceleration of the AI robot during the a-th sampling interval in the historical operation records. a When R aWhen the value is greater than R, the duration of the a-th sampling interval is recorded as the abnormal sampling interval duration; otherwise, the duration of the a-th sampling interval in the historical operation record is not processed. Obtain the total number n of several adjacent abnormal sampling intervals from the historical operation records, and calculate the duration T of mechanical vibration in the historical operation records. sum =n×△t, when T sum When T >, the AI robot's mechanical vibration is determined to be abnormal in the historical operation record, and the historical time period consisting of several abnormal sampling intervals is obtained to obtain the characteristic historical time period in the historical operation record. For example, the duration of abnormal sampling intervals where there is an adjacency relationship is as follows: If the duration of the a-th sampling interval and the duration of the (a+1)-th sampling interval in the historical operation record are both recorded as abnormal sampling interval durations, then it is determined that there is an adjacent relationship between the abnormal sampling interval durations corresponding to the a-th sampling interval duration and the (a+1)-th sampling interval duration. Extract the operation records of the AI robot within a specific historical period from the historical operation records to obtain the AI robot's marked abnormal operation records.
[0019] Step S2: Obtain the control strategy of the AI robot, and combine it with the marked abnormal operation records to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy, and obtain the risk control strategy; Step S2 includes: Obtain the control strategy generated by the AI robot in the current cycle, extract the data of various control parameters from the control strategy, and construct the strategy vector V´ of the control strategy; For example, control strategies include data on various control parameters that affect the motion and force control of AI robots; For example, the control parameters include: the target trajectory, position, velocity, acceleration sequence, target stiffness, target damping, and target inertia of the AI robot joints; For example, the policy vector for constructing control strategies is constructed as follows: Data of various control parameters are obtained from the control strategy, and the data of the control parameters are read directly to obtain the corresponding values of the control parameters; The values of various control parameters are obtained from the control strategy, the values of various control parameters are normalized to obtain the characteristic values of each control parameter, and then they are aggregated in a preset order to obtain the strategy vector of the control strategy. Obtain the historical control policies from the marked abnormal operation records of the AI robot, and construct the policy vector V of the historical control policies; Calculate the policy similarity value S between the control policy in the current period and the historical control policy in the marked abnormal operation record, set the policy similarity threshold S´, when S > S´, it is determined that the control policy in the current period is similar to the historical control policy in the marked abnormal operation record, and when S ≤ S´, it is determined that the control policy in the current period is not similar to the historical control policy in the marked abnormal operation record; For example, the specific calculation formula for the policy similarity value S is: , Obtain the total number C of several marked abnormal operation records similar to the control policy of the AI robot, and calculate the vibration abnormal risk value G of the AI robot under the control policy as G = C / C sum , where C sum represents the total number of marked abnormal operation records of the AI robot; Set the first risk threshold G1 and the second risk threshold G2. When G > G1, it is determined that the AI robot has a mechanical vibration abnormality risk under the control policy, and the control policy of the AI robot in the current period is recorded as the risk control policy, where G1 > G2 > 0; When G2 ≤ G ≤ G1, obtain the cumulative sum S of the policy similarity values between the control policy in the current period and the historical control policies in several marked abnormal operation records sum , set the cumulative threshold S α ; When S sum > S α , it is determined that the AI robot has a mechanical vibration abnormality risk under the control policy, and the control policy of the AI robot in the current period is recorded as the risk control policy. Otherwise, it is determined that the AI robot does not have a mechanical vibration abnormality risk under the control policy; When G < G2, it is determined that the AI robot does not have a mechanical vibration abnormality risk under the control policy.
[0020] Step S3: Analyze the abnormal danger contribution degree of the control parameters in the risk control policy, determine the abnormal control parameters, evaluate the mechanical vibration abnormality correction effect of the abnormal control parameters under different correction directions and correction scales, and perform policy correction and optimization on the risk control policy to obtain the target control policy of the AI robot; Among them, step S3 includes: Obtain the risk control policy of the AI robot in the current period, and obtain the characteristic historical control policy of the risk control policy; For example, the specific acquisition process of the characteristic historical control policy of the risk control policy is: Obtain the maximum value S of the policy similarity values between the historical control policies of each marked abnormal operation record of the AI robot and the risk control policymax Get the maximum value S max The corresponding historical control strategies for marking historical anomalies are recorded as characteristic historical control strategies of risk control strategies. Calculate the data deviation Δv between a certain control parameter in the risk control strategy and the characteristic historical control strategy, set the hazard direction variable F for a certain control parameter of the AI robot, where F∈{+1,-1}, and calculate the hazard contribution value P=Δv·F for a certain control parameter in the risk control strategy; For example, the specific calculation process for the data deviation between a certain control parameter in the risk control strategy and the characteristic historical control strategy is as follows: The deviation between a control parameter in the risk control strategy and a historical control strategy is obtained by subtracting the eigenvalue of a control parameter in the strategy vector of the risk control strategy from the eigenvalue of a control parameter in the strategy vector of the historical control strategy. For example, the specific process for setting the danger direction variable F of a certain control parameter of an AI robot is as follows: When the value of a certain control parameter is smaller, the greater the risk of mechanical vibration in the AI robot, the danger direction variable F of that control parameter is -1, such as the target damping control parameter of the AI robot joint. When the value of a certain control parameter is larger, the risk of mechanical vibration in the AI robot is greater. Then the danger direction variable F of the control parameter is +1. Such as the human-machine contact stiffness of the AI robot end effector and the amplitude of end effector movement. Set a risk contribution threshold p. When p>p, it is determined that a certain control parameter in the risk control strategy has a high-risk abnormality and is recorded as an abnormal control parameter. Otherwise, no action is taken on the certain control parameter, where p>0. Obtain the historical operation records of the AI robot, obtain the historical control strategies of various similar comparisons of the risk control strategy, obtain the safety benchmark value of a certain abnormal control parameter in the risk control strategy, and obtain the benchmark difference value d of a certain abnormal control parameter. Set a correction direction variable K for a certain abnormal control parameter. When d>0, the abnormal control parameter is increased by K=+1. When d≤0, the abnormal control parameter is decreased by K=-1. For example, the specific process of obtaining the historical operation records of the AI robot is as follows: Obtain a historical running record of the AI robot. If the AI robot is not judged to have abnormal mechanical vibration in a certain historical running record, then record that historical running record as the reference historical running record of the AI robot. For example, the specific process for obtaining a similar historical control strategy to the risk control strategy of an AI robot is as follows: Obtain the policy similarity value S between the control policy in the current period and the historical control policy in a certain historical operation record. △ Obtain the policy similarity threshold S´, when S △ When >S´, it is determined that the control strategy in the current period is similar to a historical control strategy in a certain historical operation record, and the historical control strategy in the certain historical operation record is recorded as the similar historical control strategy of the risk control strategy. For example, the specific process for obtaining the safety baseline value of a certain anomaly control parameter in a risk control strategy is as follows: Obtain the mean value of the characteristic value of a certain abnormal control parameter in each similar historical control strategy, and record it as the safety benchmark value of a certain abnormal control parameter in the risk control strategy; For example, the specific process for obtaining the baseline difference value d of a certain abnormal control parameter is as follows: Subtracting the characteristic value of a certain abnormal control parameter in the risk control strategy from the safety benchmark value of a certain abnormal control parameter yields the benchmark difference value d of that abnormal control parameter. Obtain the preset candidate step size η´ of a certain abnormal control parameter from the platform, and calculate the correction step size η of the certain abnormal control parameter; For example, the specific formula for calculating the correction step size η of a certain abnormal control parameter is as follows: , Calculate the initial correction value Y = X + η·K for a certain abnormal control parameter, obtain the initial correction values for each abnormal control parameter, construct the initial correction strategy vector for the risk control strategy, and calculate the vibration abnormality risk value G of the AI robot under the initial correction strategy vector. Y Obtain the vibration abnormality risk value G and the second risk threshold G2 of the AI robot; For example, the specific construction process of the initial modified strategy vector for constructing a risk control strategy is as follows: The initial modified strategy vector of the risk control strategy is obtained by replacing the feature values of each abnormal control parameter in the strategy vector of the risk control strategy with the initial modified values of each abnormal control parameter. For example, the specific process of obtaining the vibration anomaly risk value of the AI robot under the initial correction strategy vector is as follows; When the strategy vector of the risk control strategy is the initial modified strategy vector, obtain the ratio between the total number C of several marked abnormal operation records similar to the control strategy of the AI robot and the total number of marked abnormal operation records of the AI robot. Set the correction adjustment coefficient β, where 1>β>0, when GY When G ≥ G2, re - obtain the initial correction values of each abnormal control parameter until G Y < G2, and obtain and collect the initial correction values of each corrected abnormal control parameter to obtain the target correction strategy vector; When G Y < G2, then record the initial correction strategy vector as the target correction strategy vector; According to the target correction strategy vector, perform strategy correction on the risk control strategy of the AI robot to obtain the target control strategy; For example, when G Y ≥ G2, re - obtain the initial correction values of each abnormal control parameter until until G Y < G2, the specific process is: When G Y > G, then divide the correction step size η of the initial correction value of each abnormal control parameter by β, and recalculate the initial correction values of each abnormal control parameter, and recalculate the vibration abnormal risk value G of the AI robot Y , when G ≥ G Y > G2, then multiply the correction step size η of the initial correction value of each abnormal control parameter by β, re - obtain the initial correction values of each abnormal control parameter, and recalculate the vibration abnormal risk value G of the AI robot Y ; For example, according to the target correction strategy vector, perform strategy correction on the risk control strategy of the AI robot to obtain the target control strategy, and the specific acquisition process is: Obtain the eigenvalue of each control parameter from the target correction strategy vector, and inverse - map the eigenvalue of each control parameter back to the physical parameter space to obtain the value of each control parameter; For example, the eigenvalue of the m - th control parameter in the target correction strategy vector is U m , obtain the maximum value U (m,max) and the minimum value U (m,min) ; Inverse - map the eigenvalue U m of the m - th control parameter back to the physical parameter space to obtain V M The specific calculation formula is: V M =(U (m,max) -U (m,min) )·U m +U (m,min) .
[0021] Step S4: Control the operation of the AI robot according to the target control strategy, and use the AI robot to interact with the staff; Among them, step S4 includes: Obtain the target control strategy of the AI robot in the current cycle, and extract the values of various control parameters from the target control strategy; The system acquires the preset sampling interval and vibration threshold R, sets the control parameters of the AI robot according to the values of the control parameters in the target control strategy, and uses the AI robot to interact with the staff. It also monitors the root mean square value of the vibration acceleration of the AI robot in each sampling interval in real time. When the root mean square value of the vibration acceleration of the AI robot in any sampling interval within the current cycle is greater than the vibration threshold R, the AI robot stops running and maintenance personnel are dispatched to inspect the AI robot.
[0022] To better implement the above methods, an AI robot intelligent control system is also proposed, which includes an anomaly analysis module, a control strategy risk assessment module, a strategy correction module, and an operation control module. The anomaly analysis module is used to acquire the historical operation records of the AI robot and analyze the degree of mechanical vibration anomalies in the historical operation records to obtain marked abnormal operation records. The control strategy risk assessment module is used to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy, and obtain the risk control strategy. The strategy correction module is used to correct and optimize the risk control strategy to obtain the target control strategy for the AI robot. The operation control module is used to control the operation of the AI robot according to the target control strategy and to enable the AI robot to interact with staff.
[0023] The anomaly analysis module includes an anomaly duration analysis unit and an anomaly analysis recording unit. The abnormal duration analysis unit is used to determine the degree of abnormality of vibration acceleration within the sampling interval of the AI robot in the historical operation record, and to determine the abnormal sampling interval duration of the AI robot. The anomaly analysis unit is used to acquire the duration of abnormal sampling intervals in the historical operation records of the AI robot, analyze the degree of abnormal mechanical vibration of the AI robot in the historical operation records, and obtain marked abnormal operation records.
[0024] The control strategy risk assessment module includes a strategy similarity analysis unit and a control strategy risk assessment unit. The strategy similarity analysis unit is used to analyze the similarity between the control strategy in the current period and the historical control strategies in the marked abnormal operation records, and to obtain the marked abnormal operation records that are similar to the control strategy. The control strategy risk assessment unit is used to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy in the current cycle, and to obtain the risk control strategy.
[0025] The strategy correction module includes an abnormal parameter determination unit and a strategy correction unit. The abnormal parameter determination unit is used to analyze the degree of contribution of abnormal risks to control parameters in risk control strategies and determine abnormal control parameters. The strategy correction unit is used to evaluate the effect of abnormal mechanical vibration correction on the AI robot under different correction directions and scales of abnormal control parameters, and to perform strategy correction and optimization on the risk control strategy to obtain the target control strategy of the AI robot.
[0026] The operation control module includes an operation control unit; The operation control unit is used to acquire the target control strategy of the AI robot in the current cycle, obtain the values of various control parameters from the target control strategy, set various control parameters of the AI robot according to the values of various control parameters in the target control strategy, and use the AI robot to interact with the staff.
[0027] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.
Claims
1. An AI robot intelligent control method based on human-computer interaction, characterized in that, The method includes: Step S1: Obtain the historical operation records of the AI robot, analyze the degree of mechanical vibration abnormality of the AI robot in the historical operation records, and obtain the abnormal operation records; Step S2: Obtain the control strategy of the AI robot, and combine it with the marked abnormal operation records to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy, and obtain the risk control strategy; Step S3: Analyze the degree of abnormal risk contribution of control parameters in the risk control strategy, determine the abnormal control parameters, evaluate the effect of abnormal control parameters on the mechanical vibration abnormality correction of AI robot under different correction directions and correction scales, and perform strategy correction and optimization of risk control strategy to obtain the target control strategy of AI robot. Step S4: Control the operation of the AI robot according to the target control strategy, and use the AI robot to interact with the staff.
2. The AI robot intelligent control method based on human-computer interaction according to claim 1, characterized in that, Step S3 includes: Obtain the risk control strategy of the AI robot in the current period, and obtain the historical control strategies with characteristics of the risk control strategy. Calculate the data deviation Δv between a certain control parameter in the risk control strategy and the characteristic historical control strategy, set the hazard direction variable F for a certain control parameter of the AI robot, where F∈{+1,-1}, and calculate the hazard contribution value P=Δv·F for a certain control parameter in the risk control strategy; Set a risk contribution threshold p. When p>p, it is determined that a certain control parameter in the risk control strategy has a high-risk abnormality and is recorded as an abnormal control parameter. Otherwise, no action is taken on the certain control parameter, where p>0. Obtain the historical operation records of the AI robot, obtain the historical control strategies of various similar comparisons of the risk control strategy, obtain the safety benchmark value of a certain abnormal control parameter in the risk control strategy, and obtain the benchmark difference value d of a certain abnormal control parameter. Set a correction direction variable K for a certain abnormal control parameter. When d>0, the abnormal control parameter is increased by K=+1. When d≤0, the abnormal control parameter is decreased by K=-1. Obtain the preset candidate step size η´ of a certain abnormal control parameter from the platform, and calculate the correction step size η of the certain abnormal control parameter; Calculate the initial correction value Y = X + η·K for a certain abnormal control parameter, obtain the initial correction values for each abnormal control parameter, construct the initial correction strategy vector for the risk control strategy, and calculate the vibration abnormality risk value G of the AI robot under the initial correction strategy vector. Y Obtain the vibration abnormality risk value G and the second risk threshold G2 of the AI robot; Set a correction adjustment coefficient β, where 1 > β > 0. When G Y ≥ G2, re-obtain the initial correction values of each abnormal control parameter until G Y < G2, and obtain the initial correction values of each corrected abnormal control parameter for aggregation to obtain a target correction strategy vector; When G Y <When G2, the initial correction strategy vector is denoted as the target correction strategy vector; Based on the target-modified strategy vector, the AI robot's risk control strategy is modified to obtain the target control strategy.
3. The AI robot intelligent control method based on human-computer interaction according to claim 1, characterized in that, Step S2 includes: Obtain the control strategy generated by the AI robot in the current cycle, extract the data of various control parameters from the control strategy, and construct the strategy vector V´ of the control strategy; Obtain the historical control policies from the marked abnormal operation records of the AI robot, and construct the policy vector V of the historical control policies; Calculate the policy similarity value S between the control policy in the current period and the historical control policy in the marked abnormal operation record. Set the policy similarity threshold S'. When S>S', it is determined that the control policy in the current period is similar to the historical control policy in the marked abnormal operation record. When S≤S', it is determined that the control policy in the current period is not similar to the historical control policy in the marked abnormal operation record. Obtain the total number C of several marked abnormal operation records similar to the control strategy of the AI robot, and calculate the vibration anomaly risk value G of the AI robot under the control strategy: G = C / C sum , where C sum This represents the total number of marked abnormal operation records of the AI robot; Set the first risk threshold G1 and the second risk threshold G2. When G > G1, it is determined that the AI robot has a risk of mechanical vibration abnormality under the control strategy, and the control strategy of the AI robot in the current cycle is recorded as the risk control strategy, where G1 > G2 > 0; When G2≤G≤G1, the sum of the similarity values S between the control strategy in the current period and the historical control strategies in the several marked abnormal operation records is obtained. sum Set the cumulative threshold S α ; When S sum >S α If the control strategy is not in effect, the AI robot is determined to have a risk of abnormal mechanical vibration under the control strategy. The control strategy of the AI robot in the current period is recorded as the risk control strategy. Otherwise, the AI robot is determined not to have a risk of abnormal mechanical vibration under the control strategy. When G < G2, it is determined that the AI robot does not have a risk of mechanical vibration abnormality under the control strategy.
4. The AI robot intelligent control method based on human-computer interaction according to claim 1, characterized in that, The step S1 includes: Set the sampling interval duration △t and the sampling eigenvalue N, obtain the historical operation record of the AI robot, and sample the vibration acceleration of the AI robot N times every sampling interval duration from the historical operation record to obtain the vibration acceleration set of the AI robot within the sampling interval duration in the historical operation record; Set the characteristic duration T and vibration threshold R, and calculate the root mean square value R of the vibration acceleration of the AI robot during the a-th sampling interval in the historical operation records. a When R a When the value is greater than R, the duration of the a-th sampling interval is recorded as the abnormal sampling interval duration; otherwise, the duration of the a-th sampling interval in the historical operation record is not processed. Obtain the total number n of several adjacent abnormal sampling intervals from the historical operation records, and calculate the duration T of mechanical vibration in the historical operation records. sum =n×△t, when T sum When T >, the AI robot's mechanical vibration is determined to be abnormal in the historical operation record, and the historical time period consisting of several abnormal sampling intervals is obtained to obtain the characteristic historical time period in the historical operation record. Intercept the operation record of the AI robot within the characteristic historical period from the historical operation record to obtain the marked abnormal operation record of the AI robot.
5. The AI robot intelligent control method based on human-computer interaction according to claim 1, characterized in that, 6. An AI robot intelligent control system, used to execute the AI robot intelligent control method based on human-computer interaction as described in any one of claims 1-5, characterized in that, 7. The AI robot intelligent control system according to claim 6, characterized in that, 8. The AI robot intelligent control system according to claim 6, characterized in that, The strategy similarity analysis unit is used to analyze the similarity between the control strategy in the current period and the historical control strategy in the marked abnormal operation record, and to obtain the marked abnormal operation record that is similar to the control strategy. The control strategy risk assessment unit is used to assess the degree of abnormal mechanical vibration risk of the AI robot under the control strategy in the current cycle, and to obtain the risk control strategy.
9. An AI robot intelligent control system according to claim 6, characterized in that, The strategy correction module includes an abnormal parameter determination unit and a strategy correction unit; The abnormal parameter determination unit is used to analyze the degree of abnormal risk contribution of control parameters in the risk control strategy and determine the abnormal control parameters. The strategy correction unit is used to evaluate the mechanical vibration anomaly correction effect of the AI robot under different correction directions and correction scales of the abnormal control parameters, and to perform strategy correction and optimization of the risk control strategy to obtain the target control strategy of the AI robot.
10. An AI robot intelligent control system according to claim 6, characterized in that, The operation control module includes an operation control unit; The operation control unit is used to acquire the target control strategy of the AI robot in the current cycle, obtain the values of various control parameters from the target control strategy, set various control parameters of the AI robot according to the values of various control parameters in the target control strategy, and use the AI robot to conduct human-computer interaction with the staff.