--- license: cc-by-4.0 task_categories: - object-detection - audio-classification tags: - drone-detection - counter-uas - acoustic-array - ptz-tracking - multimodal - gps-truth - low-altitude pretty_name: "nightjar field flight 2026-08-04 #3 — verification of the proportional zoom climb" --- # nightjar flight — 2026-08-04, third sortie Short verification sortie. Same rig and site as the other 2026-08-04 sessions. GPS truth: **1,997 valid UTC stamps of 1,998**. | | | |---|---| | airborne | 200 s | | max height | 14 m | | max slant range | 142 m | | median elevation angle | **6°** | | path flown | 357 m | ## Why this session exists It verifies a two-stage fix to the cued zoom climb. The cue's containment focal (~7.07 mm from FOV ≥ 2·k·σ) had been computed on every cue and never commanded; when that was fixed, the climb then ran at full lens speed into a 1 mm deadband and **overshot**. This session carries the proportional-speed correction. Measured against the 7.07 mm ceiling: | SEARCH focal | previous sortie | **this sortie** | |---|---|---| | median | 10.88 mm | **5.93 mm** | | p90 | 11.80 mm | 10.92 mm | | max | **20.03 mm** | **10.92 mm** | The long overshoot tail is gone and the median now sits just under the ceiling instead of 54% above it. Zoom also climbs on tracks properly here: TRACK median 2.37×, p90 9.90×, max 20.27×, with 39.7% of the flight above 2× and 6.9% above 5×. ## What this session does NOT show ⚠ **At 14 m altitude and a 6° median elevation angle the drone was on the treeline**, against foliage rather than sky, and never got past 142 m. Detection read 31% inside 50 m and 0% beyond — same as every other sortie that day — and the tracked median range was 19 m. So this data verifies camera *behaviour* only. It says nothing about the 50–150 m detection problem, and it sits at the null of a horizontal array's elevation response (sensitivity goes as sin(el)), so it is also a poor test of acoustic elevation. ## Cue gate 8 cue candidates, 3 chirp-rejects, **all 8 at the relaxed 10.0 dB threshold** — nightjar was never in TRACK when a cue was evaluated. A state-dependent presence gate (14.6 dB while tracking, 10.0 while searching) was live; `audio/cue_log.jsonl` now stamps `thr_db` and `nj_state` on every candidate so each cue is attributable to the threshold that produced it. ## Contents / gotchas ``` audio/uma16_.wav UMA-16, 16 ch, S16_LE audio/sb_.wav SB-POLARIS, 8 ch, S32_LE audio/cue_log.jsonl cue gate log (thr_db, nj_state per candidate) video_segments.tar 1080p native + 768x432 substream, 60 s MPEG-TS blackbox/state.jsonl tracker FSM ~2.9 Hz blackbox/frames.tar substream stills f.jpg dji/*.csv, *.txt decoded + original FlightRecord (CUSTOM.dateTime = true UTC) ``` UMA-16 0-based channel 9 is a dead electronic-floor channel — exclude it, run on 15. SB-POLARIS has 3 live capsules of 8; detect them by RMS (live ~−70 dBFS, dead ~−190) rather than hardcoding. ⚠ **The arrays use different sample widths** — UMA S16_LE, SB S32_LE; reading both as int16 scrambles SB channel identity and destroys its spectrum. Calibration as flown: `az_offset_deg` **301.37**, `invert_az` true (re-solved from the JBL beacon that afternoon, 0.40° drift). Cue producer: cheap daemon (`SUBHUNT=0`). Site notch 100–200 Hz. Start pose pan 178.9 / tilt 45, pointed at the JBL beacon. ## Elevation datum (fit_el_datum, 2026-08-16) The rig height in this day's drone-altitude frame (the missing constant that floored close-range elevation truth at ~25 deg estate-wide) is SOLVED from this repo's own blackbox TRACK/LOCK poses vs the eval-cache GPS — no acoustics, no field survey. - **z_rig = -0.20 m**, el = -1 x tilt + 87.43 deg - north offset -115.59 deg (pan sign -1), az gate residual 1.5 deg - med |el residual| **1.11 deg** over 71 camera-on-drone rows, horiz 19-30 m - per band: 0-30m: 1.11 deg (n=71) Artifact: `joshruby/acoustic-knowledge` -> `v3/assets/el_datum/nightjar-flight-20260804-3.json`. Tool: `sirch613/subhunt-v2` `v3/fit_el_datum.py`. Derive el truth as atan2(hgt - z_rig, horiz) — NEVER from raw hgt.