Wrist-Worn Hand Activity Sensing Using IMU and Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional activity sensing technologies primarily focus on locomotion-related body activities, neglecting fine-grained hand activities that are often independent of body movements, limiting their contextual sensitivity and applicability in informatics applications.
Innovation Solution
A system comprising a wrist-worn computational device with an inertial measurement unit (IMU) and machine learning components that capture and classify hand activity data using convolutional neural networks, enabling the detection of various hand activities without external infrastructure, and providing insights for personal informatics, health monitoring, and skill assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional activity sensing focuses on locomotion-related body activities, then the system can detect gross body movements, but it fails to capture fine-grained hand activities that are independent of body movements
Solution Approach 1:
The patent segments the activity sensing function by introducing a dedicated hand activity detection module that operates independently from the locomotion detection system. The IMU data processing is divided into separate pipelines: one for body-level activities and another for hand-level activities, allowing each to be optimized for its specific detection goals without interfering with the other.
Solution Approach 2:
The patent adds a new dimension of analysis by focusing on hand-wrist movements as a separate level of observation from whole-body locomotion. This dimensional shift enables the system to capture fine-grained hand activities (typing, scrolling, gesturing) that occur independently of body movement, effectively adding a micro-movement dimension to the existing macro-movement detection capability.
2Measurement precision
If the system uses machine learning algorithms to classify hand activities, then classification accuracy improves, but computational complexity and processing time increase
Solution Approach 1:
The patent performs preliminary feature extraction and data preprocessing before feeding data to the machine learning classifier. By pre-processing IMU data to extract relevant hand activity features and pre-training the classification model during manufacturing or initial setup, the system reduces real-time computational complexity while maintaining high classification accuracy during actual use.
Solution Approach 2:
The machine learning model is trained offline using labeled hand activity data, allowing it to learn complex patterns without requiring complex runtime computation. The pre-trained model then serves itself by automatically classifying hand activities in real-time with minimal processing, eliminating the need for complex real-time training algorithms.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system accurately senses and identifies fine-grained hand activities, enhancing contextual awareness and health monitoring capabilities, while preventing harmful patterns and promoting healthy habits, such as reducing the risk of repetitive strain injuries and improving skill evaluation.
Implementation Method 1
The IMU component obtains, from a wrist-worn computational device, hand activity data associated with a sustained series of hand motor actions
Data Source
AI summary
Systems and techniques for facilitating hand activity sensing are presented. In one example, a system obtains, from a wrist-worn computational device, hand activity data associated with a sustained series of hand motor actions in performance of a human task. The system also employs a machine learning technique to determine classification data indicative of a classification for the human task.