Object Position Estimation Using Reference Anchors and Dissimilarity Matrices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for estimating the positions of multiple objects, such as loudspeakers or microphones, are cumbersome and prone to errors due to the need for precise distance measurements between objects, which becomes increasingly difficult as the number of objects increases, leading to inefficient and imprecise position estimation.
Innovation Solution
An object position estimating apparatus that generates characteristic vectors based on measurements of closeness to reference points in the real space, calculates norms between these vectors to create a dissimilarity matrix, and applies MultiDimensional Scaling to estimate object positions, allowing for easier and more accurate positioning without direct distance measurements between objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If distances between all pairs of position estimation objects are measured to derive a distance matrix, then position estimation can be performed using Multidimensional Scaling, but the number of measurements increases significantly and measurement errors accumulate as the number of objects increases
Solution Approach 1:
The patent introduces reference objects as intermediaries between the position estimation objects. Instead of directly measuring distances between all pairs of objects, the system measures distances from each object to the reference objects. These reference object distances serve as mediator data that is then used to compute relative positions through coordinate transformation, significantly reducing the number of required measurements.
Solution Approach 2:
The patent transforms the measurement problem from direct pairwise distance measurements in object space to measurements in a reference-based coordinate system. By measuring distances to reference objects and using coordinate transformation (Multidimensional Scaling), the system estimates positions in a different dimensional framework, avoiding the need for complete pairwise measurement matrices.
2Quantity of substance
If the number of position estimation objects increases, then more comprehensive position information is obtained, but the possibility of estimation error due to measurement error grows
Solution Approach 1:
Reference objects serve as stable intermediaries that provide consistent measurement benchmarks. By measuring all objects against the same reference objects, the system creates a unified coordinate framework that maintains reliability even as the number of estimated objects increases. The reference objects act as anchors that prevent error accumulation across multiple objects.
3Loss of time
If direct distance measurements between objects are performed, then position relationships are obtained, but the process becomes increasingly difficult and time-consuming as the number of objects increases
Solution Approach 1:
The patent segments the measurement task into two distinct phases: first, measuring distances from multiple reference objects to all position estimation objects; second, computing relative positions through coordinate transformation. This segmentation reduces the measurement burden from O(M²) pairwise measurements to O(M×N) reference-based measurements, where M is the number of objects and N is the number of reference objects.
Solution Approach 2:
The system performs preliminary measurements by establishing reference object positions first. These reference measurements are prepared in advance and serve as the foundation for subsequent position estimations. By having the reference framework established beforehand, the actual object positioning becomes a simpler transformation problem rather than requiring complete pairwise measurements.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An object position estimating apparatus which estimates positions of M objects in a real space (M being an integer not less than 2), including: a characteristic vector generating unit operable to generate, for each of M objects, a characteristic vector, the characteristic vector including as components measurements of the object measured on N scales (N being an integer not less than 3), each of N scales measuring closeness to each of N reference points in the real space; a dissimilarity matrix deriving unit operable to calculate a norm between the characteristic vectors of two objects for every pair from among M objects and to derive a dissimilarity matrix with M rows and M columns, the dissimilarity matrix including as elements the calculated norms; and an estimation unit operable to estimate positions of M objects in the real space based on the dissimilarity matrix and to output a estimation result.