Dynamic Data Selection Using Voting and Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for selecting data in compute systems operating in physical environments are ineffective and inefficient, particularly in real-time scenarios, as they fail to accurately determine the best data to use for decision-making processes.
Innovation Solution
The implementation of a voting mechanism and data fusion techniques to select between incongruent data, where the system determines whether to fuse data or use a voting mechanism based on the context, ensuring accurate data selection for tasks such as determining object depth or activating heating elements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current data selection techniques are used, then the system can operate with simple data processing, but the accuracy and reliability of data selection deteriorates in real-time physical environments
Solution Approach 1:
The patent implements dynamic data selection by continuously evaluating the congruence of data from multiple sensors and adapting the selection strategy in real-time. The system determines whether to use voting mechanisms or data fusion based on current environmental conditions and data quality, making the data processing approach flexible and context-dependent rather than static
Solution Approach 2:
The system changes processing parameters by switching between different data selection methods (voting vs. fusion) based on data congruence metrics. When data from multiple sensors congruent, the system uses voting mechanisms; when incongruent, it employs data fusion techniques. This parameter change allows the system to optimize accuracy for different operational scenarios
2Measurement precision
If data fusion is used to select between incongruent data, then data selection accuracy improves, but processing time and computational load increases
Solution Approach 1:
The system applies partial data fusion by first checking for data congruence using simpler voting mechanisms. Only when data is determined to be incongruent does the system proceed to more computationally intensive fusion processes. This partial action approach minimizes processing time for the majority of cases while maintaining accuracy for problematic data scenarios
Solution Approach 2:
The patent implements preliminary data congruence checking before committing to full data fusion processes. By pre-evaluating whether data from multiple sensors agrees, the system can avoid unnecessary computational overhead of data fusion when simple voting suffices, thus reducing overall processing time while maintaining accuracy when needed
3Productivity
If a voting mechanism is used to select data, then processing efficiency improves, but reliability deteriorates when data from sensors are incongruent
Solution Approach 1:
The system uses feedback from data congruence evaluation to dynamically switch between voting mechanisms and data fusion approaches. When voting produces incongruent results, the feedback triggers a transition to more reliable fusion techniques, ensuring that reliability is maintained while preserving processing efficiency for congruent data scenarios
Solution Approach 2:
The patent makes the data selection mechanism dynamic by adapting between voting and fusion based on real-time data quality assessment. This dynamic approach allows the system to maintain high processing efficiency through voting when appropriate, while automatically transitioning to more reliable fusion methods when data incongruence is detected, thus balancing efficiency and reliability
Data Source
AI summary
The present disclosure generally relates select data (e.g., sensor data and/or depth values generated using sensor data). In some examples, techniques for determining whether to use sensor fusion are provided. In some examples, techniques for determining whether to use a voting mechanism are provided. In some examples, techniques for responding to different types of decisions are provided.