← Back to the chapter · Next detail →

Technical detail · Sensor Fusion Architectures

Fusion input-output signatures

Those levels become easier to audit when each one is written as an input-output signature. The ladder diagram in Figure 3 names the five signatures and attaches one concrete transformation to each.

Five stacked cards show fusion signatures: DAI-DAO turns a noisy frame into a cleaner frame, DAI-FEO turns a raw frame into labelled blobs with pixel sizes, FEI-FEO joins two feature sets, FEI-DEO turns a feature vector into a fault label, and DEI-DEO votes two verdicts into one, with a closing note that lower rows ship fewer bytes but erase replay detail.
The five input-output signatures range from data-in/data-out filtering to decision-in/decision-out voting. Later rows produce more compact outputs with less detail available for replay.

The top rows of Figure 3 stay in the data domain: DAI-DAO turns a noisy frame into a cleaner frame with a Gaussian or median filter, and DAI-FEO hands segmentation a raw frame and receives blobs with pixel sizes. Below them, FEI-FEO joins video and audio features for one activity window, FEI-DEO turns a feature vector into a fault label, and DEI-DEO lets two detector verdicts vote. The closing rule condenses the architecture decision into one line — retain as much input detail as the links can afford and as the audit duty requires — because a compact output alone cannot reconstruct the detail discarded on the way down.

← Back to the chapter · Next detail →