Public safety

Why this matters

When a flash flood cuts roads and people are reported missing, rescue teams need an aerial view as fast as possible. Drones provide images within minutes without risking responders’ lives. But if a drone team flies without coordinating with the command post, the images may never reach those who need them, or the drone may get in the way of a rescue helicopter. Murphy’s Disaster Robotics (2014) documents dozens of real robot deployments in disasters and shows that technology helps when it fits the responders’ way of working.

The whole course follows one hypothetical case: a flash flood in a district, with one person missing in riverside forest; the incident command post asks the provincial drone team for support. All numbers are synthetic training data, not a real event. The Python code for every module can be downloaded from /downloads/uat-364/.

Drone roles across the disaster timeline

A real example: after the 2019 cyclones in Mozambique, WFP used drones with AI to count damaged houses, and the drone pilots were local staff that WFP had trained before the disaster. Preparing people before a disaster matters as much as the aircraft.

The incident command system

Incidents involving many agencies need a single command structure. The incident command system (ICS) in NIMS, third edition (FEMA, 2017), has an Incident Commander and four main sections: Operations, Planning, Logistics and Finance/Administration. The drone team is not an independent unit but a resource within this structure. Thailand’s framework is the National Disaster Prevention and Mitigation Plan 2021–2027 of the Department of Disaster Prevention and Mitigation, and work with urban search and rescue teams follows the UN INSARAG Guidelines 2020.

Placing the UAS unit under Operations and sending data to the situation unit under Planning is one example; agencies may arrange it differently. What matters is that everyone knows who tasks the drone team and who receives the data.

Resources for continuous missions

Overwatch of a flooded area may need a drone in the air continuously for hours. One battery flies for a limited time and takes longer to charge than to fly, so the team must work out batteries and aircraft before deploying.

import math

MISSION_MIN, FLIGHT_MIN, SWAP_MIN, CHARGE_MIN = 240, 25, 5, 60

sorties = math.ceil(MISSION_MIN / FLIGHT_MIN)
per_slot = math.ceil((FLIGHT_MIN + SWAP_MIN + CHARGE_MIN) / FLIGHT_MIN)
print(f"{sorties} sorties over {MISSION_MIN} min")
print(f"batteries in rotation per aircraft slot: {per_slot} (+1 spare = {per_slot + 1})")
print("no gap during battery swaps needs 2 aircraft alternating "
      f"-> {2 * (per_slot + 1)} batteries and a charger that keeps up")

10 sorties over 240 min
batteries in rotation per aircraft slot: 4 (+1 spare = 5)
no gap during battery swaps needs 2 aircraft alternating -> 10 batteries and a charger that keeps up

Preparing data and the team before a disaster

The knowledge unit on preparedness notes that post-event images are easier to interpret with pre-event baseline data carrying dates and coordinate systems, such as community boundaries, buildings, roads and staging areas. Old data may not match the current situation, so the age of every layer must be known.

from datetime import date

EVENT = date(2026, 9, 20)
layers = [  # layer name, data date, coordinate system
    ("village boundaries", date(2025, 11, 2), "EPSG:32647"),
    ("building footprints", date(2023, 5, 14), "EPSG:32647"),
    ("roads", date(2026, 6, 30), "EPSG:4326"),
    ("staging areas", None, None),
]
for name, d, crs in layers:
    issues = []
    if d is None:
        issues.append("date unknown")
    elif (EVENT - d).days > 365:
        issues.append(f"{(EVENT - d).days} days old")
    if crs is None:
        issues.append("CRS unknown")
    print(f"{name:<20} {'OK' if not issues else ', '.join(issues)}")

village boundaries   OK
building footprints  1225 days old
roads                OK
staging areas        date unknown, CRS unknown

Law and data protection

An emergency does not automatically exempt a team from permission. The team must check CAAT requirements and coordinate through the local command post (see UAT 313). Mission images may show faces, homes and the locations of victims. The ICRC Handbook on Data Protection in Humanitarian Action (3rd ed., 2024) has a chapter on drones, highlighting personal data, data security and community trust (details in module 5). NIST’s standard test methods for small drones can be used to design pilot skill training.

Module lab

Common mistakes

Summary

• Drones have roles before, during and after a disaster; preparing people matters as much as the aircraft
• The drone team is a resource under the incident command system; everyone must know who tasks it and who receives its data
• Continuous missions need a plan for aircraft, batteries and charging
• Baseline data must have known age and coordinate system, and emergencies do not waive permission or data protection

Check your understanding

1. A 3-hour mission with 20 minutes per battery needs how many sorties?
2. With 20 minutes of flight, 5 minutes to swap and 50 minutes to charge, how many batteries rotate per aircraft slot?
3. What are the four main ICS sections under NIMS?
4. Why does continuous overwatch need two drones?
5. Does an emergency exempt a team from flight permission?