Table 1 Comparison among strategies for segmenting and reconstructing signals
From: Static force field representation of environments based on agents’ nonlinear motions
Method used in | Segmentation/Reconstruction Type | Input features | Error measure | Output |
|---|---|---|---|---|
[40] | Batch (top-down) | Data from etch machine | ℓ 2 distance between line and data points | Identification of states in a HMM |
[41] | Online (Sliding window + top down) | Data points series | ℓ 2 distance between line and data points | Signal’s piecewise linear representation |
[42] | Online (Slope change threshold) | Data points series | Distance between reference slopes | Signal’s piecewise linear representation |
[43] | Batch (bottom-up) | High frequency data points | ℓ 2 distance between line and super-interval data | Identification of events |
[44] | Online (furthest candidate segmenting point) | Data points series | Maximum vertical distance to a line | Signal’s piecewise linear representation |
[44] | Online (Backward segmentation) | Data points series | Maximum vertical distance to a line | Signal’s piecewise linear representation |
[45] | Offline (nonlinear state estimation) | Aircraft sensor data | Root-mean-square error | Aircraft path reconstruction |
[46] | Offline | Interactive Multiple Mode filter data | Distance to predefined modes of flight | Identification of modes of flight |
Proposed work | Online (sliding window and angle distributions) | Velocity vectors from bank of filters | Mahalanobis distance between angle and distribution | Velocity vectors linked to an attractive field |