CN-121973194-A - Mechanical arm conveying stability regulation and control method and device

CN121973194ACN 121973194 ACN121973194 ACN 121973194ACN-121973194-A

Abstract

The invention discloses a method and a device for regulating and controlling the carrying stability of a mechanical arm, and relates to the field of storage logistics. The method comprises the steps of collecting multi-dimensional data of each timestamp in a current time window in the process of carrying out a task by a mechanical arm, processing the multi-dimensional data through a stability prediction model to predict risk probability of each timestamp in a future time window so as to determine risk probability of the future time window, and generating a parameter adjustment strategy in response to the risk probability of the future time window conforming to a preset regulation and control condition so as to adjust motion parameters of the mechanical arm through the parameter adjustment strategy. According to the embodiment, the prospective active safety control is realized by fusing the stability prediction model of the multi-dimensional data, the parameters of the mechanical arm can be early warned and automatically adjusted before the object falls or shakes violently, the interruption of a carrying task is avoided, and the continuity, safety and execution efficiency of carrying operation are improved.

Inventors

LIN HONGYU
WANG PANPAN
WANG HAOTIAN
HE TIAN

Assignees

北京京东远升科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260121

Claims (15)

1. The mechanical arm conveying stability regulation and control method is characterized by comprising the following steps of: In the process of carrying out a carrying task by the mechanical arm, collecting multi-dimensional data of each time stamp in the current time window; processing the multidimensional data through a stability prediction model to predict risk probabilities of all time stamps in a future time window, and further determining the risk probabilities of the future time window; and generating a parameter adjustment strategy in response to the risk probability of the future time window accords with a preset regulation and control condition, so as to adjust the motion parameters of the mechanical arm through the parameter adjustment strategy.
2. The method of claim 1, wherein generating a parameter adjustment policy comprises: constructing adjustment parameters of each timestamp in the future time window relative to the last timestamp in the current time window to generate one or more candidate parameter adjustment strategies; Sequentially updating the current time window and the future time window by adopting a sliding window strategy, and for a single candidate parameter adjustment strategy, re-predicting the risk probability of the updated future time window through a stability prediction model based on the multi-dimensional data of the updated current time window; selecting a risk probability adjacent to a preset risk threshold, determining a size relation between the risk probability and the preset risk threshold, and updating a parameter adjustment range according to the size relation and a candidate parameter adjustment strategy corresponding to the selected risk probability; and iteratively executing the processes of generating candidate parameter adjustment strategies, updating time windows, predicting risk probabilities and updating parameter adjustment ranges until the optimal parameter adjustment strategies are obtained.
3. The method of claim 2, wherein generating a parameter adjustment policy further comprises: for each timestamp in the future time window, determining an optimal parameter adjustment strategy in parallel; and sequentially splicing the optimal parameter adjustment strategies of all the time stamps in the future time window according to the sequence of all the time stamps to obtain the optimal parameter adjustment strategy of the future time window.
4. A method according to any one of claims 1-3, wherein the parameter is a speed parameter, and said adjusting the motion parameter of the mechanical arm by the parameter adjustment strategy comprises: determining a current speed of a last timestamp in the current time window; determining a target speed of each timestamp in the future time window according to the parameter adjustment strategy and the current speed; and generating a smooth transition section between the current speed and the target speed of the first time stamp in the future time window, so as to adjust the operation parameters of the mechanical arm through the smooth transition section and the target speeds of the time stamps in the future time window.
5. The method of claim 1, wherein the collecting multi-dimensional data for each timestamp in the current time window further comprises: Determining sampling frequency of each channel for collecting corresponding dimension data; For dimension data with sampling frequency larger than preset sampling frequency, performing downsampling processing on the collected dimension data by using the preset sampling frequency; and filling the latest effective dimension data into the multi-dimension data of each time stamp in the current time window for the dimension data with the sampling frequency smaller than or equal to the preset sampling frequency.
6. The method of claim 1, further comprising training a stability prediction model, the training process comprising: The method comprises the steps of acquiring multi-dimensional data and execution results of historical carrying tasks at all time stamps, wherein the multi-dimensional data comprise dynamic data and static data, the dynamic data comprise mechanical arm movement dimension data and tail end sensing dimension data, the static data comprise load object dimension data and grabbing configuration dimension data, and the execution results comprise at least one of vibration, slippage and falling of an object; Extracting a time sequence feature vector based on the dynamic data, extracting a static feature vector based on the static data, and fusing the time sequence feature vector and the static feature vector to obtain a global feature vector of each time window; And training a basic stability prediction model by taking the global feature vector of the time window as input and the execution result of each timestamp in the future time window which is positioned behind the time window and adjacent to the time window as output to obtain a trained stability prediction model.
7. The method of claim 6, wherein the data of different dimensions is acquired through different channels, and the extracting the timing feature vector based on the dynamic data comprises: Generating a plurality of time windows by adopting a sliding window strategy, determining all channels for collecting the dynamic data, and analyzing the characteristic representation of all the channels in each time window; analyzing the dependency relationship among the timestamps in the adjacent time windows through the characteristic representation of all the channels in each time window to obtain the hidden vector of each timestamp; and obtaining the time sequence feature vectors of all channels in each time window based on the time sequence feature vectors of all time stamps.
8. The method of claim 7, wherein said analyzing the characteristic representation of all channels at each time window comprises: respectively extracting time domain features and frequency domain features based on dynamic data acquired by each time stamp of a single channel in a single time window; Modeling the time domain signal to obtain a time domain feature vector, fusing the time domain feature vectors of all channels in the single time window to obtain a time domain feature matrix, converting the time domain feature matrix into a one-dimensional time domain vector, and Modeling the frequency domain signals to obtain frequency domain feature vectors, fusing the frequency domain feature vectors of all channels in the single time window to obtain a frequency domain feature matrix, and converting the frequency domain feature matrix into one-dimensional frequency domain vectors; And splicing the one-dimensional time domain vector and the one-dimensional frequency domain vector to obtain the characteristic representation of all channels in the single time window.
9. The method of claim 8, wherein extracting the time domain features and the frequency domain features based on the dynamic data collected by the individual channels in the individual time windows at the individual time stamps, respectively, comprises: determining time-varying information of dynamic data based on the dynamic data acquired by each time stamp of a single channel in the single time window to obtain a time domain signal; And determining fluctuation frequency according to the time domain signal, decomposing the time domain signal into a plurality of frequency domain signals according to different fluctuation frequencies, and constructing a spectrogram based on the plurality of frequency domain signals.
10. The method according to any one of claims 6-9, further comprising: And extracting frequency domain statistical characteristic data from the dynamic data to be added into the static data, wherein the frequency domain statistical characteristic data comprises at least one of main frequency, spectral energy distribution, spectral entropy and frequency band energy ratio.
11. The method according to any one of claims 6-9, wherein the training the base stability prediction model to obtain a trained stability prediction model comprises: training the basic stability prediction model based on a training set to obtain a trained stability prediction model; Inputting a test set into the stability prediction model, and predicting risk probabilities of a plurality of time stamps in a future time window; And comparing the predicted result with the real result in the test set, identifying a time stamp of the prediction error, and adjusting parameters of the stability prediction model according to the multidimensional data of the time stamp of the prediction error.
12. Mechanical arm handling stability regulation and control device, characterized by comprising: the acquisition module is used for acquiring multidimensional data of each timestamp in the current time window in the process of executing the carrying task by the mechanical arm; The prediction module is used for processing the multidimensional data through a stability prediction model so as to predict the risk probability of each timestamp in a future time window and further determine the risk probability of the future time window; and the regulation and control module is used for responding to the risk probability of the future time window to accord with a preset regulation and control condition and generating a parameter regulation strategy so as to regulate the motion parameters of the mechanical arm through the parameter regulation strategy.
13. An electronic device, comprising: One or more processors; storage means for storing one or more programs, When executed by the one or more processors, causes the one or more processors to implement the method of any of claims 1-11.
14. A computer readable medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the method according to any of claims 1-11.
15. A computer program product comprising a computer program which, when executed by a processor, implements the method according to any one of claims 1-11.

Description

Mechanical arm conveying stability regulation and control method and device Technical Field The invention relates to the field of warehouse logistics, in particular to a method and a device for regulating and controlling the conveying stability of a mechanical arm. Background In the handling operation of the mechanical arm, the prediction and control of the falling risk of the object are always key technical difficulties. Although the existing abnormality detection strategy based on the force sensor has certain practicability, the problem of obvious response lag exists. The traditional threshold protection mechanism mainly relies on rapid braking after an abnormal stress signal exceeds the limit to prevent falling, and the mode can only intervene when the abnormality occurs or is close to the occurrence, and belongs to a passive response strategy. In the scenes of rapid carrying, dynamic task switching and the like, the hysteresis is extremely easy to cause untimely control reaction, and the problems of error stopping, erroneous judgment or protection failure and the like are caused, so that the requirements of modern industry on efficient and stable carrying are difficult to meet. Disclosure of Invention In view of the above, embodiments of the present invention provide a method and an apparatus for adjusting and controlling the handling stability of a mechanical arm, which at least can solve the problem of response hysteresis in predicting the falling risk of an object in the prior art. In order to achieve the above object, according to an aspect of the embodiments of the present invention, there is provided a method for controlling handling stability of a robot arm, including: In the process of carrying out a carrying task by the mechanical arm, collecting multi-dimensional data of each time stamp in the current time window; processing the multidimensional data through a stability prediction model to predict risk probabilities of all time stamps in a future time window, and further determining the risk probabilities of the future time window; and generating a parameter adjustment strategy in response to the risk probability of the future time window accords with a preset regulation and control condition, so as to adjust the motion parameters of the mechanical arm through the parameter adjustment strategy. To achieve the above object, according to another aspect of the embodiments of the present invention, there is provided a robot handling stability control apparatus, including: the acquisition module is used for acquiring multidimensional data of each timestamp in the current time window in the process of executing the carrying task by the mechanical arm; The prediction module is used for processing the multidimensional data through a stability prediction model so as to predict the risk probability of each timestamp in a future time window and further determine the risk probability of the future time window; and the regulation and control module is used for responding to the risk probability of the future time window to accord with a preset regulation and control condition and generating a parameter regulation strategy so as to regulate the motion parameters of the mechanical arm through the parameter regulation strategy. In order to achieve the above object, according to still another aspect of the embodiments of the present invention, there is provided a robot handling stability control electronic device. The electronic equipment comprises one or more processors and a storage device, wherein the storage device is used for storing one or more programs, and when the one or more programs are executed by the one or more processors, the one or more processors realize any one of the mechanical arm conveying stability regulation and control method. To achieve the above object, according to still another aspect of the embodiments of the present invention, there is provided a computer readable medium having stored thereon a computer program which, when executed by a processor, implements any one of the above-described robot arm handling stability regulation methods. To achieve the above object, according to still another aspect of an embodiment of the present invention, there is provided a computer program product. The computer program product of the embodiment of the invention comprises a computer program, and the program is executed by a processor to realize the mechanical arm conveying stability regulating and controlling method provided by the embodiment of the invention. According to the scheme provided by the invention, one embodiment of the scheme has the advantages that normal machine vibration and high-risk signals can be effectively distinguished by fusing multidimensional data and designing a stability prediction model, and false alarm or missing alarm situations are reduced. The method realizes proactive active safety control, can early warn and automatically adjust the motion parameters of the mec