Automated flight in SITL and the field

Example 1 Which leg took unusually long?

Sortie 1 from Module 3 flies at 8 m/s and hovers 60 seconds at each point to take images. Actual times are read from the log as the time to reach each waypoint (simulated values). The team investigates any leg more than 20% slower than planned.

import math

pts = {"H": (0, 0), "P1": (420, 150), "P2": (900, -200), "P10": (1100, -900)}
legs = [("H", "P1"), ("P1", "P2"), ("P2", "P10"), ("P10", "H")]
actual_s = [118, 138, 170, 245]                  # seconds per leg from the log, including the hold at the end point
SPEED, HOLD = 8.0, 60
for (a, b), act in zip(legs, actual_s):
    d = math.dist(pts[a], pts[b])
    plan = d / SPEED + (HOLD if b != "H" else 0)
    diff = (act - plan) / plan
    flag = "  <- investigate" if diff > 0.20 else ""
    print(f"{a:>3}->{b:<3} {d:5.0f} m  plan {plan:5.1f} s  actual {act:3d} s  {diff:+.0%}{flag}")

  H->P1    446 m  plan 115.7 s  actual 118 s  +2%
 P1->P2    594 m  plan 134.3 s  actual 138 s  +3%
 P2->P10   728 m  plan 151.0 s  actual 170 s  +13%
P10->H    1421 m  plan 177.7 s  actual 245 s  +38%  <- investigate

The last leg, from P10 back to take-off, is much slower than planned while the others are close to plan. Possible causes are a headwind, or RTL using a different speed from the mission speed. Check wind direction and speed in the log before concluding. This information makes the energy plan for the next sortie more accurate.

Example 2 Stepping through events with the configured values (teaching model)

params = {"FS_GCS_CONTINUE_AUTO": True, "BATT_FS_LOW_ACT": "RTL", "BATT_FS_CRT_ACT": "LAND", "FS_EKF_ACTION": "LAND"}
events = [(300, "GCS link lost"), (420, "GCS link restored"), (540, "battery low"), (600, "EKF variance high")]
RANK = {"AUTO": 0, "RTL": 1, "LAND": 2}         # higher is more conservative (model assumption)

mode = "AUTO"
for t, ev in events:
    want = mode
    if ev == "GCS link lost":
        want = "AUTO" if params["FS_GCS_CONTINUE_AUTO"] else "RTL"
    elif ev == "battery low":
        want = params["BATT_FS_LOW_ACT"]
    elif ev == "EKF variance high":
        want = params["FS_EKF_ACTION"]
    new = want if RANK[want] >= RANK[mode] else mode
    print(f"t={t:>3}s {ev:<20} requested {want:<4} -> mode {new}")
    mode = new

t=300s GCS link lost        requested AUTO -> mode AUTO
t=420s GCS link restored    requested AUTO -> mode AUTO
t=540s battery low          requested RTL  -> mode RTL
t=600s EKF variance high    requested LAND -> mode LAND

A short GCS link drop does not stop the mission, because it is set to continue. But when the battery is low the aircraft returns to launch, and when the EKF goes wrong on the way back it lands where it is, which could be in the middle of the flooded area. The team must know in advance that this outcome is possible and prepare a recovery team. Before the real job, test the same sequence of events in SITL to see whether the result matches expectations.