Meeting Log 04: Sensor Fusion Heuristics and False Positive Suppression
Bimodal Logic Gate Calibration and Disturbance Differentiation
Meeting Details
Date: 19 March 2026
Attendees: Ketan Clint Pinto, Mr. Roshan Renji
Subject: Bimodal Logic Gate Calibration and Disturbance Differentiation
As the project transitioned into its final developmental phases, the focus shifted entirely to system reliability and the elimination of false-positive alerts. The Wi-Fi CSI engine demonstrated exceptional sensitivity, accurately detecting high-velocity impacts. However, during a controlled demonstration, the dropping of a high-mass inanimate object (a backpack) triggered a critical alert. The supervisor emphasized that in a real-world geriatric care environment, a system prone to non-biological false alarms would be deemed clinically unacceptable. This necessitated the immediate refinement of the master logic loop to fully incorporate the mmWave radar data.
The session involved a rigorous analysis of heuristic weighting and logical gate design. The confidence engine was algorithmically recalibrated to prevent Wi-Fi variance from autonomously triggering an alert. The logic was restructured to assign a maximum confidence interval of 70% to high-intensity kinetic disturbances. To achieve the 100% threshold required to initiate the emergency protocol, the engine must receive immediate biological verification from the LD2410C radar. Specifically, the radar must report a "State 2 (STATIONARY)" status, confirming the presence of micro-Doppler respiratory shifts at floor level. When the inanimate object drop test was repeated, the Wi-Fi variance spiked, but the radar correctly reported an "EMPTY" state. The confidence metric remained capped at 70%, and the alert was successfully suppressed. The supervisor approved this bimodal confirmation logic as the definitive solution to the false-positive dilemma.