Method and system for predicting the next effective worker action and providing timely prompts

Abstract

The present invention belongs to the field of artificial intelligence technology, and discloses a method and system for predicting the next effective worker action and giving timely prompts, collecting and analyzing historical videos of a worker's workstation, using a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of the action, identifying the action currently being performed by the worker, and trying to match the inherent rhythm, prompting or warning the worker's next action; by using the skilled actions of experienced workers to quickly train and warn new workers of their operating actions. The present invention uses Transformer as the architecture basis, retains the spatiotemporal characteristics between worker action frames, and autoregressively predicts the future for a longer period of time. It can predict the next action of the worker and predict several consecutive actions that the worker may perform; by predicting the worker's action steps, using the skilled actions of experienced workers to quickly train and warn new workers of their operating actions, thereby achieving the purpose of standardizing the worker's operating standards.

Description

Technical Field

[0001] The present invention belongs to the field of artificial intelligence technology, and in particular relates to a method and system for predicting the next effective worker action and providing timely prompts. Background Art

[0002] Compared to traditional action recognition, anticipation is often more challenging. It requires classifying the current spatiotemporal visual pattern into a single action category. Predicting the worker's next effective action is multimodally distributed, and action recognition usually avoids temporal reasoning by leveraging instantaneous contextual clues. Anticipation inherently requires modeling the progression of past behaviors and predicting the future. For example, the presence of a plate of food with a fork indicates that it may be eating behavior, but anticipating the same action requires recognizing and reasoning about previous behavior sequences. Modeling long-term context is crucial to the anticipation effect.

[0003] Right now, it’s important for AI systems to be able to predict people’s future actions for applications ranging from self-driving cars to augmented reality. When someone is building a desk, they might wonder whether to attach a leg or a drawer next. Based on past experience, a bystander can suggest the right course of action. But this type of prediction is a challenging task for AI, requiring both predicting a multimodal distribution of future activities and modeling the progression of past actions.

[0004] Predicting the next effective action of a worker belongs to a long-term predictive reasoning task. The commonly used method is to use a standard architecture to extract frame-level features, and then use clustering and cyclic models. Most of these models only aggregate features in the time range, and rarely consider modeling the sequential time evolution of the video on the frame. For example, cyclic models such as LSTM are often used for prediction, but its non-parallel nature makes it difficult to model long-distance time dependencies. Therefore, the present invention adopts a Transformer-based attention architecture. The present invention uses a processed worker action dataset to train the model and learn to predict action labels that match real future features.

[0005] In industrial production, most assembly lines require manual operation. This specific scenario has the characteristics of a single work background for the actors, high repeatability of actions, clear processes and little possibility of change, which increases the possibility of achieving operational standardization.

[0006] Through the above analysis, the problems and defects of the existing technology are as follows: Existing action prediction is a challenging task for AI, which requires both predicting the multimodal distribution of future activities and modeling the progress of past actions. Summary of the invention

[0007] In view of the problems existing in the prior art, the present invention provides a method and system for predicting the next effective worker action and providing timely prompts.

[0008] The present invention is implemented as follows: a method for predicting the next effective worker action and giving timely prompts, the method for predicting the next effective worker action and giving timely prompts comprises:

[0009] Use cameras to collect and analyze historical videos of a worker's workstation, use a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of movements, identify the current worker's movements, and try to match the inherent rhythm to prompt or warn the worker of the next move; at the same time, use the skilled movements of experienced workers to quickly train and warn new workers of their operating movements.

[0010] Furthermore, the method for predicting the next effective worker action and providing timely prompts includes the following steps:

[0011] Step 1: Use a camera to collect real-time video data as a data set and perform preprocessing;

[0012] Step 2: label the video data set preprocessed in step 1; for the labeled operation segment of the data set, a seconds, sample the previous part of the frame and the end frame;

[0013] Step 3: Input the sampled fragments into the network model TFP built on the basis of transformer for training;

[0014] Step 4: By training the model weights through three loss functions, TFP spontaneously learns to focus on the spatial features corresponding to hands and objects, capturing the sequential nature of action videos;

[0015] Step 5: Use the model to process unlabeled worker work video data, pause the output video once every second, and display the predicted future actions in the context of the current worker's actions.

[0016] Furthermore, the TFP in step three uses a two-stage architecture consisting of a backbone network that operates on a single frame or clip, and a head architecture that operates on frame-level features to predict future features and operations.

[0017] Furthermore, the steps of TFP operation in step 3 include:

[0018] (1) Given a video clip T frames V, the backbone network extracts features of each frame;

[0019] (2) Split the frames into non-overlapping blocks, embed them using a feed-forward network, attach special classification tags, and apply multi-layer multi-head self-attention; share weights between frames and use features corresponding to the head classification tags;

[0020] (3) After obtaining the features extracted by the backbone network, the head network uses the Causal Transformer Decoder to predict the future features of each input frame;

[0021] (4) After processing all previous features and including them, a linear classifier is used to decode the predicted features into a distribution over semantic operation classes, and the final prediction is used as the model output for the next operation expected task.

[0022] Furthermore, the video clip T frame V in step (1) is 1 ,…,X T}, the backbone network extracts the features of each frame, represented as {z 1 ,…,z T}, z t =B(X t );

[0023] In the step (3), N 1 ,N 2 ,...,N t =D(Z 1 ,Z 2 ,...,Z t ), where N t is the predicted future feature Z corresponding to the frame feature t ;

[0024] The distribution Y of the semantic operation classes in step (4) t =θ(Z t ).

[0025] Further, the LOSS method in step 4 includes:

[0026] Classify the features of the last frame of a video clip to predict the labeled future action; regress the intermediate frame features to the features of subsequent frames to train the model to predict; train the model to classify the intermediate actions.

[0027] Another object of the present invention is to provide a system for predicting the next effective worker action and giving timely prompts using the method for predicting the next effective worker action and giving timely prompts, the system for predicting the next effective worker action and giving timely prompts comprising:

[0028] The data acquisition and preprocessing module is used to use the camera to collect real-time video data as a data set and perform preprocessing;

[0029] The dataset annotation module is used to annotate the preprocessed video dataset; for the labeled operation segment of the dataset, the label is set before the action starts. a seconds, sample the previous part of the frame and the end frame;

[0030] The sampling segment training module is used to input the sampled segments into the network model TFP built on the basis of transformer for training;

[0031] Model weight training module, which is used to train model weights through three loss functions. TFP spontaneously learns to focus on spatial features corresponding to hands and objects, capturing the sequential nature of action videos;

[0032] The action prediction module is used to use the model to process unlabeled worker work video data, pause the output video once every second, and obtain the predicted future action displayed in the context of the current worker action.

[0033] Another object of the present invention is to provide a computer device, the computer device comprising a memory and a processor, the memory storing a computer program, and when the computer program is executed by the processor, the processor performs the following steps:

[0034] Use cameras to collect and analyze historical videos of a worker's workstation, use a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of movements, identify the current worker's movements, and try to match the inherent rhythm to prompt or warn the worker of the next move; at the same time, use the skilled movements of experienced workers to quickly train and warn new workers of their operating movements.

[0035] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the following steps:

[0036] Use cameras to collect and analyze historical videos of a worker's workstation, use a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of movements, identify the current worker's movements, and try to match the inherent rhythm to prompt or warn the worker of the next move; at the same time, use the skilled movements of experienced workers to quickly train and warn new workers of their operating movements.

[0037] Another object of the present invention is to provide an information data processing terminal, which is used to implement the system for predicting the next effective worker action and providing timely prompts.

[0038] In combination with the above technical solutions and the technical problems solved, please analyze the advantages and positive effects of the technical solutions to be protected by the present invention from the following aspects:

[0039] First, in view of the technical problems existing in the above-mentioned prior art and the difficulty of solving the problems, the technical solutions to be protected by the present invention and the results and data during the research and development process are closely combined to analyze in detail and deeply how the technical solutions of the present invention solve the technical problems, and some creative technical effects brought about after solving the problems. The specific description is as follows:

[0040] The present invention provides a method for predicting the next correct operation of a worker's work action based on computer vision. The prediction method uses a camera to collect and analyze historical videos of a worker's workstation, uses a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of its movements, identifies the current action of the worker, and tries to match its inherent rhythm, so as to prompt or warn the worker of the next action. The method can quickly train and warn new workers of their operating actions by using the skilled actions of experienced workers.

[0041] The present invention provides a computer vision-based method for predicting the next correct operation of a worker's work action. In view of the great progress made in the development of transformer in natural language processing and image modeling, and the ability to capture the connection between data well, the present invention adopts an end-to-end attention video prediction model, which better understands long-term dependencies than previous methods, such as how workers' past work steps indicate what they will do next.

[0042] The present invention uses Transformer as the architecture basis, retains the spatiotemporal features between worker action frames, and can be autoregressively derived to predict the longer future, not only predicting the next action of the worker, but also predicting several consecutive actions that the worker may perform. By predicting the worker's action steps through this method, the skilled actions of experienced workers can be used to quickly train and warn new workers of their operating actions, thereby achieving the purpose of standardizing workers' operating standards.

[0043] The present invention discloses a method for predicting the next correct operation of a worker's work action based on computer vision, which involves image processing technology. The main points of the technical solution are: 1) collecting and analyzing historical videos of a worker's workstation; 2) using a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of its action;

[0044] 3) Identify the action that the current worker is doing and try to match its inherent rhythm; 4) Prompt or warn the worker of the next action. This method can quickly train and warn new workers of their operating actions by using the skilled actions of experienced workers.

[0045] Second, considering the technical solution as a whole or from the perspective of the product, the technical effects and advantages of the technical solution to be protected by the present invention are described in detail as follows:

[0046] For other environments with a large number of highly repetitive operations similar to assembly line operation scenarios, the present invention uses AI to predict the worker's next correct operation, thereby achieving the role of early warning and regulation.

[0047] Third, as auxiliary evidence of the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0048] The expected benefits and commercial value of the technical solution of the present invention after transformation are as follows: The present invention can be applied to factory assembly line operations, ① it can serve as an early warning for workers' operations, standardize the action flow of assembly line operations, thereby improving workers' operating efficiency and effectively preventing problems in workers' operations. ② the present invention can be used for operational training of new workers, improve worker training efficiency, and reduce personnel training expenses. ③ the present invention only needs the corresponding camera on the assembly line to complete data collection, which is easy to deploy and scale. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0050] Figure 1 It is a flow chart of a method for predicting the next effective worker action and providing timely prompts provided by an embodiment of the present invention.

[0051] Figure 2 It is a block diagram of a system structure for predicting the next effective worker action and providing timely prompts provided by an embodiment of the present invention;

[0052] In the figure: 1. Data collection and preprocessing module; 2. Data set annotation module; 3. Sampling segment training module; 4. Model weight training module; 5. Action prediction module.

[0053] Figure 3 It is a schematic diagram of prediction of action frames by a head network provided by an embodiment of the present invention.

[0054] Figure 4 It is a schematic diagram of an expected reference for a sampling operation provided by an embodiment of the present invention.

[0055] Figure 5 It is a schematic diagram of the operation of the model provided by the embodiment of the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0057] In view of the problems existing in the prior art, the present invention provides a method and system for predicting the next effective worker action and providing timely prompts. The present invention is described in detail below in conjunction with the accompanying drawings.

[0058] 1. Explanatory Examples In order to enable those skilled in the art to fully understand how to implement the present invention, this section provides an illustrative example that expands and describes the technical solution of the claims.

[0059] like Figure 1 As shown, the method provided by the embodiment of the present invention for predicting the next effective worker action and providing timely prompts includes the following steps:

[0060] S101, using a camera to collect real-time video data as a data set and perform preprocessing;

[0061] S102, annotating the video data set preprocessed in S101; for the labeled operation segment of the data set, before the start of the action a seconds, sample the previous part of the frame and the end frame;

[0062] S103, inputting the sampled fragments into the network model TFP built on the basis of transformer for training;

[0063] S104, by training the model weights with three loss functions, TFP spontaneously learns to focus on the spatial features corresponding to hands and objects, capturing the sequential nature of action videos;

[0064] S105, using the model to process unlabeled worker work video data, pausing the output video once every second to obtain the predicted future actions displayed against the background of the current worker actions.

[0065] like Figure 2 As shown, the system provided by the embodiment of the present invention for predicting the next effective worker action and providing timely prompts includes:

[0066] The data acquisition and preprocessing module 1 is used to use a camera to collect real-time video data as a data set and perform preprocessing;

[0067] The data set annotation module 2 is used to annotate the preprocessed video data set; for the labeled operation segment of the data set, the label is set before the action starts. aseconds, sample the previous part of the frame and the end frame;

[0068] The sampling segment training module 3 is used to input the sampled segments into the network model TFP built on the basis of transformer for training;

[0069] Model weight training module 4 is used to train the model weights through three loss functions. TFP spontaneously learns to focus on the spatial features corresponding to hands and objects, capturing the sequential nature of action videos;

[0070] The action prediction module 5 is used to use the model to process unlabeled worker work video data, pause the output video once every second, and obtain the predicted future action displayed in the context of the current worker action.

[0071] The schematic diagram of the prediction of the action frame by the head network provided by the embodiment of the present invention is as follows: Figure 3 shown.

[0072] The expected reference schematic diagram of the sampling operation provided by the embodiment of the present invention is as follows: Figure 4 shown.

[0073] The schematic diagram of the model operation provided by the embodiment of the present invention is as follows Figure 5 shown.

[0074] The purpose of the present invention is to provide a method for predicting the next correct operation of a worker's work action based on computer vision. The prediction method uses a camera to collect and analyze historical videos of a worker's workstation, uses a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of its movements, identifies the current action of the worker, and tries to match its inherent rhythm, so as to prompt or warn the worker of the next action. This method can quickly train and warn new workers of their operating actions by using the skilled actions of experienced workers.

[0075] The above technical objectives of the present invention are achieved through the following technical solutions: a computer vision-based method for predicting the next correct operation of a worker's work action. In view of the great progress made in the development of transformers in natural language processing and image modeling, and the ability to capture the connection between data well, the present invention adopts an end-to-end attention video prediction model, which better understands long-term dependencies than previous methods, such as how workers' past work steps indicate what they will do next.

[0076] The steps are as follows:

[0077] 1) Use a camera to collect real-time video data as a data set and preprocess it;

[0078] 2) Label the video dataset preprocessed in step 1), and then for the labeled operation segment of the dataset, a Seconds, the present invention samples the previous part of the frame and the end frame.

[0079] 3) Input the sampled fragments into the network model TFP (Transformer Prediction) built on the basis of transformer for training.

[0080] 4) By training the model weights through three loss functions, TFP will spontaneously learn to focus on the spatial features corresponding to hands and objects, capturing the sequential nature of action videos.

[0081] 5) The model is used to process unlabeled video data of workers at work. The output video is paused once every second, and the predicted future actions are shown against the background of the current worker's actions.

[0082] Furthermore, in step 3), TFP uses a two-stage architecture consisting of a backbone network that operates on a single frame or clip, followed by a head architecture that operates on frame-level features to predict future features and operations.

[0083] Furthermore, the specific steps of TFP operation in step 3) are:

[0084] A) Given a video clip T frame V = {X 1 ,…,X T}, the backbone network extracts the features of each frame, represented as {z 1 ,…,z T}, z t =B(X t );

[0085] B) Split the frames into non-overlapping blocks, embed them with a feed-forward network, attach a special classification tag, and apply multiple layers of multi-head self-attention. Then, the invention shares weights between frames and uses features corresponding to the head classification tag;

[0086] C) After obtaining the features extracted by the backbone network, the head network uses the Causal Transformer Decoder to predict the future features of each input frame. 1 ,N 2 ,...,N t =D(Z 1 ,Z 2 ,...,Z t ), where N t is the predicted future feature Z corresponding to the frame feature t ;

[0087] D) After processing all previous features and including it, a linear classifier is used to decode the predicted features into a distribution over semantic action classes i.e. Y t =θ(Z t ). The final prediction, Y t The model output is used as the expected task for the next operation.

[0088] Furthermore, the specific method of LOSS described in step 4) includes the following:

[0089] First, we classify the features of the last frame of a video clip to predict the labeled future action; second, we regress the intermediate frame features to the features of subsequent frames to train the model to predict what will happen next; third, we train the model to classify the intermediate actions. By jointly optimizing these three losses, our model predicts future actions more realistically. These additional losses make the model more suitable for long-range reasoning because they provide additional supervision to the model. It also shows that its performance can be improved by incorporating more upper and lower frames.

[0090] In summary, the present invention has the following beneficial effects:

[0091] 1. The Transformer architecture is used to preserve the spatiotemporal features between worker action frames, which can be autoregressively derived to predict the longer future. It can not only predict the worker’s next action, but also predict several consecutive actions that the worker may perform.

[0092] 2. This method can be used to predict the steps of workers' movements, so that the skilled movements of experienced workers can be used to quickly train and warn new workers of their operating movements, thereby achieving the purpose of standardizing workers' operating standards.

[0093] 2. Application examples. In order to prove the creativity and technical value of the technical solution of the present invention, this section is an application example of the claimed technical solution on a specific product or related technology. The video collected by the camera at the worker's workstation is preprocessed, the preprocessed data set is annotated and sampled, and the sampling result is divided into 16×16 non-overlapping patches as the input frame, each patch is flattened into a 256d vector, and they are linearly projected to 768d, which is the feature dimension used in the entire encoder. Although the present invention does not need to classify each frame separately, the present invention still adds a learnable token embedding to the patch feature, and its output will be used as the frame-level embedding input of the head. Then, the present invention adds the learned position embedding to each patch feature. The present invention selects the frame-specific spatial position encoding so that the same backbone model with shared weights can be applied to each frame. The embedded features then pass through multiple decoder layers, each of which consists of masked multi-head attention, LayerNorm (LN) and multi-layer perceptron (MLP), and the final output then passes through another LN to obtain future frame embeddings, thereby completing the next step of action prediction task.

[0094] 3. Evidence of the effects of the embodiments. The embodiments of the present invention have achieved some positive effects during the development or use process, and indeed have great advantages over the prior art. The following content is described in conjunction with the data, charts, etc. of the test process.

[0095] The present invention adopts an end-to-end transformer architecture to process video data sets collected by cameras. Compared with the traditional model architecture, this architecture introduces a self-attention mechanism and combines a bidirectional encoder and a causal decoder, which provides good results for action prediction tasks. The concurrent structure also improves the efficiency of predicting the worker's next action.

[0096] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0097] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A method for predicting the next effective worker action and providing timely prompts, characterized in that: The method for predicting the next effective worker action and giving timely prompts includes: using a camera to collect and analyze historical videos of a worker's workstation, using a machine learning model with a self-attention mechanism to learn and automatically analyze the rhythm of the action, identifying the action currently being taken by the worker, and trying to match the inherent rhythm, so as to prompt or warn the worker of the next action; at the same time, by using the skilled actions of experienced workers to quickly train and warn new workers of their operating actions; The method for predicting the next effective worker action and providing timely prompts comprises the following steps: Step 1: Use a camera to collect real-time video data as a data set and perform preprocessing; Step 2: label the video data set preprocessed in step 1; for the labeled operation segment of the data set, sample the first part of the frames and the end frame ta seconds before the start of the action; Step 3: Input the sampled fragments into the network model TFP built on the basis of transformer for training; Step 4: By training the model weights through three loss functions, TFP spontaneously learns to focus on the spatial features corresponding to hands and objects, capturing the sequential nature of action videos; Step 5: Use the model to process unlabeled worker work video data, pause the output video once every second, and obtain the predicted future actions displayed against the background of the current worker's actions; The steps of TFP operation in step 3 include: (1) Given a video clip T frames V, the backbone network extracts features of each frame; (2) Split the frames into non-overlapping blocks, embed them using a feed-forward network, attach special classification tags, and apply multi-layer multi-head self-attention; share weights between frames and use features corresponding to the head classification tags; (3) After obtaining the features extracted by the backbone network, the head network uses the Causal Transformer Decoder to predict the future features of each input frame; (4) After processing all previous features and including them, a linear classifier is used to decode the predicted features into a distribution over semantic operation classes, and the final prediction is used as the model output for the next operation expected task; The video clip T frame V in step (1) is T }, the backbone network extracts the features of each frame, represented as {z1,…,z T }, zt=B(X t ); In step (3), N1, N2, ..., N t =D(Z1,Z2,...,Z t ), where N t is the predicted future feature Zt corresponding to the frame feature; The distribution Y of the semantic operation classes in step (4) t =θ(Z t ).

2. The method for predicting the next effective worker action and providing timely prompts as claimed in claim 1, characterized in that: The TFP in step 3 uses a two-stage architecture consisting of a backbone network that operates on a single frame or clip, and a head architecture that operates on frame-level features to predict future features and operations.

3. The method for predicting the next effective worker action and providing timely prompts as claimed in claim 1, characterized in that: The LOSS method in step 4 includes: Classify the features of the last frame of a video clip to predict labeled future actions; regress intermediate frame features to features of subsequent frames to train the model to predict; train the model to classify intermediate actions.

4. A system for predicting the next effective worker action and providing timely prompts for implementing the method for predicting the next effective worker action and providing timely prompts as described in any one of claims 1 to 3, characterized in that: The system for predicting the next effective worker action and providing timely prompts includes: The data acquisition and preprocessing module is used to use the camera to collect real-time video data as a data set and perform preprocessing; The dataset annotation module is used to annotate the preprocessed video dataset; for the labeled operation segment of the dataset, the previous part of the frame and the ending frame are sampled ta seconds before the start of the action; The sampling segment training module is used to input the sampled segments into the network model TFP built on the basis of transformer for training; Model weight training module, which is used to train model weights through three loss functions. TFP spontaneously learns to focus on spatial features corresponding to hands and objects, capturing the sequential nature of action videos; The action prediction module is used to use the model to process unlabeled worker work video data, pause the output video once every second, and obtain the predicted future action displayed in the context of the current worker action.

5. A computer device, characterized in that: The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for predicting the next effective worker action and providing timely prompts as described in any one of claims 1 to 3.

6. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the method for predicting the next effective worker action and providing timely prompts as claimed in any one of claims 1 to 3.

7. An information data processing terminal, characterized in that: The information data processing terminal is used to implement the system for predicting the next effective worker action and providing timely prompts as described in claim 4.

Images