User-Specific Sensor Weight Calibration for Wearable Void Interpolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor systems face challenges in accurately synthesizing data at locations between and outside sensors, particularly due to the use of discrete sensor arrays which result in low accuracy estimates through traditional interpolation and extrapolation techniques, and fail to account for individual user-specific differences in stance and gait.
Innovation Solution
A method and system for determining user-specific weights using initial estimation weights, which are modified based on training data from specific users to improve the accuracy of synthesized sensor data at void locations between and outside sensors, by calculating aggregate force values and optimizing these weights to minimize error.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional interpolation and extrapolation techniques are used to synthesize sensor data at void locations, then the sensing unit complexity is reduced, but the accuracy of synthesized sensor data deteriorates
Solution Approach 1:
The patent applies parameter changes by transitioning from generic estimation weights to user-specific weights that are individually calibrated for each user. The system determines initial estimation weights and then modifies them based on training data from the specific user performing predefined actions, thereby optimizing the accuracy of synthesized sensor data for that particular user while maintaining the simplified discrete sensor array configuration.
2Ease of manufacture
If discrete sensor arrays are used, then the cost and complexity of the sensing unit are reduced, but the accuracy of sensor data estimates at interstitial and external locations deteriorates
Solution Approach 1:
The system maintains the cost-effective discrete sensor array configuration while improving estimation accuracy through user-specific weight calibration. By determining initial estimation weights and refining them with user-specific training data, the patent achieves high-fidelity sensor data synthesis at void locations without requiring a dense sensor array, thus maintaining ease of manufacture while improving measurement precision.
3Device complexity
If generic estimation weights are used for synthesizing sensor data, then the system is simpler to implement, but it fails to account for individual user variations in stance and gait
Solution Approach 1:
The patent implements dynamics by transitioning from static generic estimation weights to dynamic user-specific weights that are calibrated for each individual user. The system collects training data from users performing predefined actions and uses this data to determine optimized estimation weights specific to each user, enabling the system to adapt to individual variations in stance and gait while maintaining a relatively simple implementation framework.
Data Source
AI summary
A system, method and computer program product for determining user-specific weights usable to synthesize sensor data at void locations in a sensor array. Initial estimation weights are determined for the void locations. Sensor readings are obtained from a plurality of sensors while a user performs predefined actions. The sensors are arranged in a first pattern that maps the sensors to respective locations on the wearable device. Synthesized sensor readings are determined based on the sensor readings and the initial estimation weights. User-specific weights are determined by modifying the initial estimation weights using an aggregate force value determined from the obtained sensor readings. The user-specific weights can be used to determine synthesized sensor readings at void locations for the user. The sensor array can be mounted to a carrier device such as a wearable device worn by the user or fitness equipment used by the user.