Hearing Instrument Activity Detection With Hierarchical Onboard Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Hearing instruments face limitations in battery and processing power due to existing activity detection methods that consume significant power and resources when transferring motion data to external devices for analysis.
Innovation Solution
Utilizing onboard computing devices within hearing instruments to apply a hierarchy of machine-trained activity models directly to motion data from internal sensors, reducing power consumption by local processing and sequential model application.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion data is transferred to external devices for activity detection, then activity detection can be performed, but power consumption and processing resource usage increase
Solution Approach 1:
The hearing instrument performs activity detection autonomously using onboard processing capabilities. The device processes motion data locally through machine-trained activity models without requiring external devices, thereby reducing power consumption while maintaining detection functionality.
Solution Approach 2:
The activity detection system is divided into hierarchical levels of complexity. Simple activities are detected using basic models that consume minimal power, while more complex activities can be detected using advanced models when additional processing resources are available. This segmented approach allows the system to adapt power consumption to the actual detection needs.
2Adaptability or versatility
If a single complex machine-trained activity model is used to detect multiple activities, then comprehensive activity detection is achieved, but processing power consumption increases
Solution Approach 1:
The activity detection system is divided into a hierarchy of multiple specialized models, each trained to detect specific activities or activity categories. Instead of using one complex model for all activities, the system segments the detection task into simpler sub-tasks handled by individual models, reducing the processing power required for each detection operation.
Solution Approach 2:
The system applies activity models selectively based on the detection context. Not all models are executed for every detection task - the system chooses the appropriate level of model complexity based on the situation, applying only the necessary processing power to achieve the required detection coverage.
3Adaptability or versatility
If multiple activity models are applied simultaneously to motion data, then comprehensive activity detection is achieved, but processing time and power consumption increase
Solution Approach 1:
The activity models are organized in a hierarchical structure where detection proceeds from simpler to more complex models. The system segments the detection process into stages, applying models in a predetermined sequence rather than simultaneously, which reduces the overall processing time while maintaining comprehensive detection coverage.
Solution Approach 2:
Simpler activity models are applied first as preliminary detection steps. If these models successfully identify an activity, the system can stop processing and return the result immediately, avoiding the time cost of applying more complex models. This preliminary action approach reduces average processing time while maintaining detection accuracy.
Data Source
AI summary
A computing system includes a memory and at least one processor. The memory is configured to store motion data indicative of motion of a hearing instrument. The at least one processor is configured to determine a type of activity performed by a user of the hearing instrument and output data indicating the type of activity performed by the user.