Crew, decisions and debrief

Example 1 The team’s PAVE worksheet

The team scores each item 0–3 and sets limits in advance (the scoring method is the team’s, not the FAA’s).

pave = {
    "Pilot": {"slept < 6 h": 2, "first flight at this site": 1},
    "Aircraft": {"motor replaced yesterday, one test flight": 2},
    "enVironment": {"gusts near limit": 2, "people near take-off point": 1},
    "External pressures": {"mayor wants images by 10:00": 3, "media present": 1},
}
LIMIT_TOTAL, LIMIT_ITEM = 8, 3
total = sum(v for items in pave.values() for v in items.values())
for area, items in pave.items():
    print(f"{area:<19} {sum(items.values())}  {', '.join(items)}")
worst = [(a, k) for a, items in pave.items() for k, v in items.items() if v >= LIMIT_ITEM]
print(f"total {total} (limit {LIMIT_TOTAL}); items at maximum: {worst}")
if total > LIMIT_TOTAL or worst:
    print("do not start as planned: remove or reduce risks, or postpone")

Pilot               3  slept < 6 h, first flight at this site
Aircraft            2  motor replaced yesterday, one test flight
enVironment         3  gusts near limit, people near take-off point
External pressures  4  mayor wants images by 10:00, media present
total 12 (limit 8); items at maximum: [('External pressures', 'mayor wants images by 10:00')]
do not start as planned: remove or reduce risks, or postpone

The total exceeds the limit, and the worst item is external pressure. A good response is not “be more careful” but to remove the risks that can really be reduced: have a well-rested pilot fly instead, cordon off the take-off area, test-fly the aircraft whose motor was just replaced, and tell the client in advance that images may arrive later than requested.

Example 2 Merging three sources into one timeline

Use an event recorded by every source, “arm”, as the reference and shift each source’s times (simulated data).

from datetime import datetime, timedelta

GPS_EPOCH, LEAP, TZ = datetime(1980, 1, 6), 18, 7
fc_arm = (2440, 354718000)                     # GWk, GMS of the arm event in the flight controller log
arm_utc = GPS_EPOCH + timedelta(weeks=fc_arm[0], milliseconds=fc_arm[1]) - timedelta(seconds=LEAP)
arm_local = arm_utc + timedelta(hours=TZ)
print("arm (from FC log):", arm_local.strftime("%Y-%m-%d %H:%M:%S"), "local")

sources = {   # source: (arm time on that clock, [(time on that clock, event)])
    "GCS laptop": ("09:32:05", [("09:36:50", "video link freeze reported"), ("09:38:12", "RTL commanded")]),
    "pilot phone": ("09:31:02", [("09:35:40", "gust felt at take-off point"), ("09:37:00", "decision to abort")]),
}
timeline = [(arm_local, "FC", "arm")]
for name, (arm_txt, events) in sources.items():
    arm_src = datetime.combine(arm_local.date(), datetime.strptime(arm_txt, "%H:%M:%S").time())
    offset = arm_local - arm_src
    print(f"{name}: clock offset {offset.total_seconds():+.0f} s")
    for txt, ev in events:
        t = datetime.combine(arm_local.date(), datetime.strptime(txt, "%H:%M:%S").time()) + offset
        timeline.append((t, name, ev))
for t, src, ev in sorted(timeline):
    print(t.strftime("%H:%M:%S"), f"{src:<12}", ev)

arm (from FC log): 2026-10-15 09:31:40 local
GCS laptop: clock offset -25 s
pilot phone: clock offset +38 s
09:31:40 FC           arm
09:36:18 pilot phone  gust felt at take-off point
09:36:25 GCS laptop   video link freeze reported
09:37:38 pilot phone  decision to abort
09:37:47 GCS laptop   RTL commanded

Ordered by corrected times, the real sequence is gust, video freeze, abort decision, and RTL a few seconds later. The slow part was the decision after the video froze, which took over a minute. With raw times, it would look as if the team decided quickly but the GCS operator was slow to command RTL, which would draw the lesson about the wrong person.