Equipment
Quadrotor UAVs
A modern quadrotor is a tightly coupled chain: flight controller → ESC → motors → propellers → airframe → IMU → back to the controller. Under normal flight, current, RPM, frame vibration, and IMU response all move together with consistent timing. The faults that matter most — PID instability, ESC degradation, propeller imbalance, sensor drift — appear first as a break in that relationship, not as a single out-of-range reading.
On a healthy airframe, motor command, current, vibration, and IMU response track each other with a stable timing relationship. As anomalies emerge, the coupling weakens before any one channel crosses a threshold. The signature is the loss of agreement — not the value of any single reading.
Each signal class on a quadrotor already has well-established analysis methods. EchoSignal doesn’t replace them — it runs them as domain plugins on top of the general engine, then models normal across all of them together. Different methods, combined into one coupling baseline.
Time–frequency methods such as STFT and wavelet analysis capture the non-stationary, transient features that flight maneuvers and prop-wash produce — the kind that fixed-band thresholds miss.
Cross-correlation, magnitude-squared coherence, and causality measures quantify how strongly — and in which direction — motor electricals, frame vibration, and IMU response track each other. The relationship is the baseline.
ARX, output-error, and state-space models predict the expected vibration and IMU response from the PWM and current inputs. Residuals between predicted and observed signals are tracked over time — persistent growth flags a structural change.
Read one at a time, each telemetry channel is a known problem with mature tools. But on a quadrotor the meaningful anomaly is a break in the cross-channel relationship — the coupling between motor command, current, vibration, and IMU that no single method, and no operator watching strip charts, can reliably hold in view.
That combined coupling anomaly is exactly what EchoSignal surfaces — and brings to your airframe team to judge.
On a multi-rotor platform, recurring in-flight instability events were proving hard to attribute. Telemetry channels, examined one at a time, looked acceptable on each affected flight. By modeling normal across the combined PWM, motor-current, airframe-vibration, and IMU signals together, EchoSignal surfaced the cross-channel coupling losses that preceded each event, and brought them to the airframe team to confirm root cause.
Outcome figures are placeholders, to be filled from your records once cleared for publication.
Bring your own telemetry — from your own airframes, in your own mission profiles — and see what the combined baseline surfaces. The best proof runs on your own data.