Learning-Based Information Processing for Unmeasured Body-Part State Inference
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies lack an effective method for estimating the states of body parts of a user during walking, particularly for body parts not directly attached with sensors.
Innovation Solution
An information processing system that includes sensor devices attached to body parts of a user, which obtain sensor data and use learning models to estimate the states of other body parts, with an electronic device providing notifications based on these estimates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor devices are attached to all body parts to directly measure states, then measurement precision is improved, but device complexity and ease of operation deteriorate
Solution Approach 1:
The patent creates a virtual model (copy) of the user's body with estimated states for all body parts, rather than physically measuring each part. The learning model generates this virtual copy by inferring unmeasured body part states from measured sensor data, eliminating the need for sensors on every body part while maintaining comprehensive monitoring capability
Solution Approach 2:
The learning model acts as an intermediary that translates limited sensor measurements into comprehensive body state information. It mediates between the small number of actual sensor readings and the full set of body part states needed for complete posture analysis, filling in missing information through intelligent inference
2Measurement precision
If sensor devices are attached to all body parts to directly measure states, then measurement precision is improved, but ease of operation deteriorates
Solution Approach 1:
The system creates a virtual representation of the user's complete body state through learning model inference, allowing comprehensive monitoring without requiring the user to wear multiple sensors. This virtual copy enables full body awareness while simplifying the user experience to just wearing one or a few sensor devices
3Device complexity
If learning models are used to estimate states of unmeasured body parts, then device complexity is reduced, but measurement precision may deteriorate
Solution Approach 1:
The learning model is pre-trained with extensive gait data and biomechanical knowledge before deployment. This preliminary preparation enables the model to make accurate inferences in real-time applications, ensuring that the virtual estimates match the precision that would be obtained from direct physical measurements
Solution Approach 2:
The system continuously refines its estimates by incorporating feedback from actual sensor measurements and comparing predicted versus observed states. This feedback mechanism allows the learning model to adapt and improve its precision over time, maintaining accuracy while using fewer sensors
Data Source
AI summary
An information processing apparatus includes a controller. The controller obtains sensor data indicating movement of body parts of a user from at least one sensor device attached to the body parts of the user. The controller estimates, on the basis of the obtained sensor data and learning models, states of body parts of the user other than the body parts of the user to which the at least one sensor device is attached.