Multi-Sensor Recognition Fusion Control for Accuracy Mismatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor fusion systems often result in degraded object recognition due to significant differences in accuracy between individual sensor results, particularly when one sensor produces inaccurate data with noise or malfunctions.
Innovation Solution
An information processing system that controls the fusion process based on the similarity and accuracy of recognition results from multiple sensors, adjusting or skipping fusion when necessary to maintain or enhance recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sensor fusion is performed to improve object recognition accuracy, then completeness of detection is improved, but recognition accuracy may degrade when sensors produce significantly different or inaccurate results
Solution Approach 1:
The system dynamically changes the fusion parameter (degree of fusion) based on the consistency of recognition results from multiple sensors. When recognition results are consistent, a higher degree of fusion is applied to improve accuracy. When results differ significantly, the fusion degree is reduced or suspended to prevent accuracy degradation, thus adapting the fusion process to current sensor performance conditions.
Solution Approach 2:
The fusion degree is made dynamic rather than fixed. The system continuously adjusts the fusion degree based on real-time comparison of recognition results from different sensors. This dynamic adjustment allows the system to optimize the balance between utilizing multiple sensor inputs and maintaining recognition accuracy when sensor data quality varies.
2Adaptability or versatility
If multiple sensors are fused to enhance detection completeness, then coverage is improved, but system complexity increases
Solution Approach 1:
The system simplifies the fusion process by dynamically adjusting the fusion degree parameter. Instead of always performing complex multi-sensor fusion, the system adapts the fusion level based on result consistency, reducing computational complexity when full fusion is not beneficial while maintaining comprehensive detection coverage.
3Loss of information
If sensor fusion is always performed to maximize information utilization, then data completeness is improved, but processing time increases
Solution Approach 1:
The system skips or reduces fusion processing when recognition results from multiple sensors are already consistent. By comparing results first and only performing full fusion when necessary (i.e., when results differ), the system avoids unnecessary processing time while still ensuring information completeness when it matters most.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
PROBLEM To provide an information processing system and an information processing method capable of inhibiting a degradation of object recognition results caused by fusing the results. SOLUTION Provided is an information processing system that includes: first obtainer 120 that obtains a first recognition result of an object based on sensor data from first sensor 201; second obtainer 130 that obtains a second recognition result of an object based on sensor data from second sensor 202 different from first sensor 201; first determiner 140 that performs a first determination of determining a degree of similarity between the first recognition result and the second recognition result; fusion controller 160 that controls, in accordance with a result of the first determination, a fusion process of fusing the first recognition result and the second recognition result; and outputter 180 that outputs at least one of the first recognition result, the second recognition result, and a third recognition result, in accordance with the result of the first determination. The third recognition result is a fusion of the first recognition result and the second recognition result.