Rehabilitation Motion Inference Using Segmented Exercise Elements
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing exercise assisting apparatuses face difficulties in generating form models and inference models due to the need for a large quantity of training data, especially when a person performs exercises not previously learned by the model.
Innovation Solution
The apparatus divides motion information into motion element information based on regularity and uses a pre-trained model to learn time-series antecedent dependency, generating an inference model capable of inferring motions from target motion information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a form model is generated through machine learning using training data indicating motions of a person doing exercise, then the model can accurately infer motions for that specific exercise, but it becomes difficult to easily generate a new form model when the person performs a different exercise type not previously learned
Solution Approach 1:
The patent segments motion information into multiple motion element information pieces based on the regularity of motion patterns. By dividing complex motion data into smaller, structured elements (such as breaking down exercise motions into sequential phases or key movement components), the system can more efficiently learn and generalize across different exercise types without requiring complete retraining for each new exercise.
Solution Approach 2:
The patent performs preliminary processing of motion data by extracting motion element information and establishing temporal relationships before actual model training. This preliminary structuring of data including time-series antecedent dependency creates a foundation that enables faster and easier generation of inference models for new exercise types, as the framework is already prepared to accommodate different motion patterns.
2Measurement precision
If a large quantity of training data indicating motions of a person doing a specific exercise is collected to generate a form model, then the model achieves high accuracy for that exercise, but the process becomes time-consuming and difficult to adapt to new exercises
Solution Approach 1:
By segmenting motion information into motion element information based on motion regularity, the system reduces the amount of raw data that needs to be processed and stored. Instead of requiring large quantities of complete motion sequences for each exercise type, the segmented approach allows the model to learn from smaller, structured motion elements that can be efficiently stored and rapidly processed when generating inference models for new exercises.
Solution Approach 2:
The patent changes the parameter representation of motion data from raw, unstructured motion sequences to structured motion element information with explicit temporal relationships. This parameter transformation enables the model to achieve high accuracy with less training data, as the structured representation captures essential motion characteristics more efficiently, reducing both data quantity requirements and model generation time.
3Productivity
If motion information is processed without considering time-series antecedent dependency, then the processing is simpler and faster, but the inference model cannot accurately capture the flow of changes in motion
Solution Approach 1:
The patent segments motion information into discrete motion element information pieces while maintaining temporal relationships between them. This segmentation approach allows the system to process motion data in manageable units that can be efficiently handled computationally, while the structured representation preserves the sequential nature of motion, enabling accurate capture of motion flow without excessive processing complexity.
Solution Approach 2:
The patent introduces motion element information as an intermediary representation between raw motion data and the final inference model. This intermediary structure captures temporal antecedent dependencies in a computationally efficient format, serving as a bridge that allows the model to learn time-series relationships without the full computational burden of processing raw continuous motion data, thus maintaining both speed and accuracy.
Data Source
AI summary
An exercise assisting apparatus 1 includes: a division unit 312 configured to divide sample motion information 3231 indicating motions of a sample person doing exercise into a plurality of pieces of motion element information 3235, according to regularity of the motions of the sample person; and a generation unit 314 configured to generate an inference model by causing a pre-trained model 322 to learn time-series antecedent dependency of the plurality of pieces of motion element information, the inference model inferring motions of a target person, based on target motion information 3211 indicating the motions of the target person doing exercise.