Reliability Calculation for Object Estimation Using Error Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for estimating the number of people in areas within a space face challenges in accurately calculating reliability due to assumptions of normal error distribution, leading to potential overestimation and deviations from actual values when error distributions are not normal.
Innovation Solution
An information processing device with a derivation unit, range calculation unit, and reliability calculation unit that uses an error model to derive constraint expressions from sensor observation values, calculates possible ranges of people, and determines reliability based on these ranges, independent of the estimation value, to account for non-normal error distributions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If normal distribution assumption is used for error calculation, then calculation simplicity is improved, but reliability calculation accuracy deteriorates when error distribution is non-normal
Solution Approach 1:
The patent changes the parameter of error distribution from assumed normal distribution to empirically determined actual distribution based on sensor characteristics. This involves collecting sensor error data, analyzing its statistical properties, and using the actual distribution parameters (mean, standard deviation, skewness, kurtosis) to calculate reliability, thereby resolving the contradiction between calculation simplicity and accuracy.
2Device complexity
If statistical model with normal distribution is used, then estimation value calculation is simplified, but deviation from actual measured value increases
Solution Approach 1:
The patent modifies the statistical model by changing the error distribution parameters from theoretical normal distribution to actual empirical distribution derived from sensor data. This involves determining the true statistical characteristics (mean, variance, skewness, kurtosis) of sensor errors and incorporating these actual parameters into the estimation model, thereby improving measurement precision while maintaining model tractability.
Data Source
AI summary
An information processing device is disclosed. In the information processing device, a derivation unit derives, for each of sensors, a constraint expression obtained by substituting an observation value for a term of an observation value variable in an error model. The observation value indicates the number of present objects observed by the sensors. The error model represents correspondence between: the number of present objects in an observation range represented by the observation value variable and an error term related to an assumed detection error range of the sensor, and the number of present objects in an observation range represented by a variable indicating the number of present objects in an observation unit space. A range calculation unit calculates a possible range of the number of present objects. A reliability calculation unit calculates reliability of an estimation value of the number of present objects for each of the areas.