Lessons from Autonomy: Drones, Robotics and Human-Machine Teaming

■ Autonomous and remotely operated systems offer emergency services something new at scale: machines can go where people shouldn't have to. ■ Start with the dull, dirty and dangerous tasks. Match the level of autonomy to the risk of the task, and keep people in command.

■ Airspace integration, resilience in degraded conditions and community trust are as important as the technology. ■ An autonomous system is only as trustworthy as its command link. Hijacking, spoofing and jamming risks mean security must be engineered in, not bolted on.

■ This edition adds the full autonomy toolkit, use cases in detail, a practical autonomy scale, how to build a drone and robotics program, privacy principles, an illustrative flood-and-fire scenario, procurement security questions and a staged roadmap. Every year, firefighters, paramedics, police officers and SES volunteers are sent into situations where the risk to life is real: active fire fronts, collapsing structures, floodwater, toxic smoke and hazardous materials. Autonomous and remotely operated systems offer something powerful.

Machines can go where people shouldn't have to.

25,000+ Chula Vista drone-as-first-responder missions 2024–25 autonomous helicopter firefighting demonstrations 50+ years since Australia last designed and built a military aircraft before the MQ-28 A drone in flight. Drones are now routine for mapping, thermal imaging and search. Source: Photo by Jason Mavrommatis, Unsplash Why autonomy matters now Three trends are making autonomy a practical option for emergency services rather than a future aspiration.

The technology has matured. Drones that once needed specialist pilots and fair weather are now reliable, affordable and capable of thermal imaging, mapping and long flights. Sensors, batteries, communications and onboard computing have all improved rapidly.

Autonomous vehicles have moved from research projects to everyday operations in mines and on farms. The risks to people are growing. Longer and more intense fire seasons, more frequent floods and more complex urban environments expose crews to danger more often.

Every task a machine can safely take on is a task that doesn't put a person in harm's way. The workforce is stretched. With volunteer numbers under pressure and career staff deployed more often, autonomy can extend what each person can do: one operator supervising several drones, or a robot doing the reconnaissance a crew would otherwise do.

None of this means autonomy will replace firefighters, paramedics, police or SES volunteers. It means machines can take on more of the dull, dirty and dangerous work, while people focus on judgement, care and command. Where autonomy is today Across defence, mining, agriculture, logistics and emergency services, autonomy is moving quickly: ■ Drones are routine for mapping, thermal imaging, search and rescue, and damage assessment.

Approvals for beyond-visual-line-of-sight (BVLOS) operations are expanding: CASA ran new broad-area BVLOS approval

pathways as a trial from October 2025 to October 2026, and named emergency services among the expected beneficiaries. ■ "Drone as first responder" programs in the United States launch drones automatically to emergency calls, often arriving before ground crews and streaming live video to dispatchers. Chula Vista's program has passed 25,000 missions.

■ Optionally piloted and autonomous firefighting helicopters have been demonstrated in the US. Sikorsky and Rain showed an autonomous Black Hawk finding and suppressing test fires in 2024, followed by live-fire trials in California in 2025. ■ Firefighting robots are used in high-risk structure fires, tunnels and hazmat incidents.

■ Autonomous ground vehicles operate at scale in Australian mines and agriculture. ■ Human–machine teaming concepts in defence, such as the MQ-28 Ghost Bat developed by Boeing Australia, show how crewed and uncrewed systems can operate together. It is the first military aircraft designed and built in Australia in more than 50 years.

The autonomy toolkit "Autonomy" covers a wide range of systems with different capabilities, costs and risks: ■ Small drones for close-range reconnaissance, thermal imaging, search and mapping, usually flown within visual line of sight. ■ Larger remotely piloted aircraft with longer endurance, able to fly beyond visual line of sight to map fire edges, monitor floods or relay communications over large areas. ■ Optionally piloted and autonomous helicopters and aircraft, still emerging, which could one day carry out suppression or resupply missions in conditions too dangerous for crewed flight.

■ Ground robots for entry into burning or unstable structures, tunnels and hazardous materials sites. ■ Remotely operated or autonomous heavy plant, such as bulldozers and excavators, for containment lines and clearing debris. ■ Uncrewed boats for flood reconnaissance and search in fast-moving water.

■ Fixed autonomous sensors, such as AI-enabled cameras and environmental sensors, that watch continuously and raise alerts. Each has a different maturity level. Small drones are in routine use.

Ground robots and remotely operated plant are proven but not yet widespread in emergency services. Autonomous suppression aircraft are at the demonstration stage. A sensible program matches its ambition to the maturity of each tool.

The Australian context Australian emergency services are already significant users of drones. Fire, police, SES and ambulance agencies in most jurisdictions operate drone programs for mapping, search, reconnaissance and assessment, and drones have become a familiar sight at major fires and floods. Many agencies have trained career staff and volunteers as remote pilots, and some have begun working towards more complex operations, such as flying beyond visual line of sight.

What is often missing is integration. Drone teams in different agencies may use different equipment, procedures and data formats, and their imagery doesn't always reach the people who could use it. The next step is less about buying more drones and more about connecting what already exists.

Lesson 1: Start with the "dull, dirty and dangerous" Defence and industry adopted autonomy first for tasks that were dull (long surveillance), dirty (contaminated environments) or dangerous (high risk to life). Those tasks deliver value quickly, carry less public sensitivity, and let organisations build skills and trust before taking on harder problems. For fire services: the early priorities are clear: ■ Persistent overnight surveillance of fire edges and flood levels.

■ Reconnaissance into smoke, hazmat or structurally unsafe areas. ■ Search for missing persons across large areas and in floodwaters. ■ Resupply to isolated crews and communities.

■ Ignition operations for planned burns and backburning. ■ Rapid damage assessment after the event, to speed recovery and insurance. Use cases in detail ■ Overnight fire edge monitoring.

Thermal drones watch fire edges and containment lines through the night, detecting breakouts and spot fires while crews rest. Mature and high value. ■ Rapid damage assessment.

After a fire or flood, drones map affected areas quickly, speeding up assistance to residents and decisions about safe re-entry. Mature and high value. ■ Search and rescue.

Thermal drones search large areas, including floodwaters and bushland, faster than ground teams. Mature, and already saving lives. ■ Hazardous materials reconnaissance.

Drones and ground robots assess sites before crews enter, using cameras and gas sensors. Proven, with growing adoption. ■ Structural fire support.

Ground robots and drones provide views inside and above burning buildings and can deliver water in the most dangerous areas. Proven in some services. ■ Planned burn ignition.

Drones can carry out aerial ignition for planned burns in difficult terrain, reducing risk to crews and crewed aircraft. Emerging, with trials under way in several countries. ■ Communications relay.

Drones or tethered systems can carry radio or mobile network equipment to restore coverage over a fireground. Emerging. ■ Resupply.

Drones can deliver small loads of medical supplies, batteries or equipment to isolated crews or communities. Emerging. Lesson 2: Human–machine teaming, not replacement The most effective autonomy programs don't replace people.

They extend what people can do. One operator might supervise several drones. A crew might send a robot ahead to assess conditions before entering.

A commander might task an uncrewed aircraft to watch a flank overnight. For fire services: autonomy should be designed around crews and commanders, giving them more reach, better information and greater safety, with humans keeping control of critical decisions. The measure of success is not how many machines are deployed.

It is how many people come home safely and how much earlier risk is seen.

Designing for teams Human–machine teaming succeeds or fails on design. Lessons from defence and industry include: ■ Keep workload manageable. An operator monitoring several drones must not be overwhelmed by video feeds.

Systems should highlight what matters and filter the rest. ■ Make intentions visible. Operators need to see what the machine is about to do, not just what it is doing now.

■ Design for interruption. Emergencies are noisy and chaotic. Interfaces must be easy to pick up again after a distraction.

■ Share the picture. Imagery and data from autonomous systems should flow into the common operating picture, so the whole team benefits, not just the operator. ■ Define roles clearly.

Who tasks the drone, who flies it, who interprets what it sees, and who decides what to do about it? Training for teaming Working alongside autonomous systems is a skill that has to be learned. Training should cover not just how to operate the machine, but how to work with it as part of a team: ■ Understanding capabilities and limits, so crews know when to rely on a system and when not to.

■ Interpreting outputs, such as thermal imagery, which can be misleading without experience. ■ Communicating with operators, using clear, standard language for tasking and reporting. ■ Responding to failures, such as a lost link or a drone coming down near crews.

■ Practising together in exercises, so crews, operators and incident teams are familiar with each other before a real event. Simulation can help. Many drone platforms include training modes, and virtual environments allow teams to practise complex scenarios, such as night operations near aircraft, without risk.

Lesson 3: Levels of autonomy and calibrated trust Industries use defined levels of autonomy, from fully manual to fully autonomous, and match the level to the risk of the task. Trust has to be calibrated: people shouldn't over-trust a system in conditions it wasn't designed for, or under-trust it when it is reliable. A practical autonomy scale for emergency services The automotive industry describes driving automation in six levels, from no automation to full automation.

A similar scale can help emergency services decide what is appropriate for each task: ■ Level 0: Manual. A person controls the machine at all times. ■ Level 1: Assisted.

The machine helps with specific functions, such as holding position or avoiding obstacles, while a person flies or drives. ■ Level 2: Supervised automation. The machine follows a planned route or pattern; a person monitors and can intervene immediately.

■ Level 3: Conditional autonomy. The machine carries out a task on its own within defined conditions and alerts a person when it needs help.

■ Level 4: High autonomy. The machine completes tasks within a defined area without intervention, while people set goals and monitor outcomes. ■ Level 5: Full autonomy.

The machine operates in any conditions without human input. This is not appropriate for emergency services today. Most current emergency services use sits at levels 1 and 2.

Mapping and monitoring flights can move to level 3 or 4 as regulation and technology allow. Anything that delivers suppression or operates near people should stay at lower levels, with a person firmly in control. Machines into harm's way.

People in command. Source: Darren Covington For fire services: be clear about which tasks suit high autonomy (routine mapping flights), which suit supervised autonomy (suppression drops, robotic entry) and which must stay under direct human control. Train crews to understand both what the systems can do and where their limits are, and design interfaces that make the system's confidence and limits visible.

“The measure of success isn't how many machines we deploy. It's how many people come home safely.” Lesson 4: Airspace integration is critical During major fires, unauthorised drones regularly ground firefighting aircraft. As Australian fire agencies keep telling the public: "If you fly, we can't." A coordinated environment with crewed aircraft, agency drones and possibly autonomous suppression aircraft needs strong airspace management.

For fire services: invest in drone traffic management, real-time deconfliction between crewed and uncrewed aircraft, and close coordination with CASA and Airservices Australia. Work with the community to reinforce that unauthorised drones put lives at risk.

What good airspace management looks like ■ A single airspace picture showing crewed aircraft, agency drones and temporary restricted areas, available to air operations staff and drone operators. ■ Clear procedures for when drones may fly near aircraft operations, and who authorises them. ■ Electronic conspicuity, so aircraft and drones can be seen by each other's systems.

■ Rapid detection of unauthorised drones, with clear processes for pausing aircraft operations and locating the operator. ■ Close coordination with CASA and Airservices Australia, including for beyond-visual-line-of-sight approvals and emergency airspace restrictions. The community's role Most unauthorised drone flights near fires are made by curious residents or people wanting to help, not by people intending harm.

Public education before and during the season, clear messaging on social media and partnerships with drone clubs and retailers can reduce the problem significantly. A firefighting helicopter works above a fire in New South Wales. Crewed aircraft, agency drones and future autonomous systems must share the same airspace safely.

Source: Photo by Darren Covington Lesson 5: Swarms and scale Defence and research programs are developing swarms: many low-cost autonomous systems working together. For fire services: swarms of drones could cover large areas for early detection after dry lightning, map flood extent across whole catchments, or search for missing persons far faster than single aircraft. Scale also changes the economics, because losing one low-cost unit doesn't end the mission.

Practical considerations for swarms Swarms are promising but not yet ready for routine emergency use. Practical questions include how many systems one operator can safely supervise, how swarms share airspace with crewed aircraft, how they communicate when networks are degraded, and how they are recovered and maintained in the field. Early steps

could include coordinated multi-drone searches under a single operator, before more autonomous swarm behaviours are introduced. Lesson 6: Resilience in degraded environments Military and mining autonomy is designed to keep working when GPS, communications or visibility are degraded. For fire services: autonomous systems must work in smoke, heat, wind, darkness and with communications down, which are exactly the conditions where they're most needed.

Procurement should test systems in those conditions, not on a clear day. Testing in realistic conditions Procurement should require evidence that systems work in the conditions where they will be used: ■ Smoke and heat, which can blind cameras and affect electronics and batteries. ■ High winds and turbulence, common near large fires and storms.

■ Darkness and rain, when many operations are most needed. ■ Loss of GPS or communications, including in steep terrain and near damaged infrastructure. ■ Long operations, with battery swaps, maintenance and operator fatigue.

Field trials during the off-season, in partnership with suppliers and research organisations, can provide this evidence before systems are relied on in a real event. Interoperability across agencies Police, SES, ambulance and fire agencies often fly drones at the same event. Without coordination, they can duplicate effort, compete for airspace and produce imagery that others can't use.

Simple steps make a big difference: ■ Shared tasking through the incident management structure, with one air operations function coordinating all agency drones. ■ Common data formats and naming, so imagery from any agency can be loaded into the shared picture. ■ Mutual recognition of pilot accreditation between agencies, so pilots can support each other's operations.

■ Joint training and exercises, including with aviation operators. Sustaining operations in long campaigns Long campaign events test autonomy programs in ways short trials do not. Batteries need charging in the field, often without mains power.

Equipment suffers from heat, ash and dust. Pilots get tired. Spare parts run short.

Agencies should plan for campaign operations: mobile charging and generators, spare aircraft and batteries, maintenance support in the field, and rostering for pilots just as for crews. Lesson 7: Regulation, ethics and community acceptance Autonomy raises real questions about safety assurance, liability, privacy and public trust. Drones that can see into backyards and robots that operate near people need clear rules.

For fire services: build clear governance, test rigorously, be transparent with the community about how systems are used and what data they collect, and involve frontline crews, volunteers and unions in design and deployment. Privacy principles for drone imagery Drones capture large amounts of imagery, often including private homes, people and vehicles. To keep community trust, agencies should: ■ Collect only what is needed for the operational purpose.

■ Tell the community how drones are used and what happens to the data. ■ Store imagery securely in Australia, with access limited to authorised people. ■ Retain imagery only as long as needed, with clear deletion rules.

■ Avoid secondary uses unrelated to emergency management unless authorised by law. ■ Handle personal information in line with privacy law and the agency's privacy policy. An industrial robot arm.

Mining and manufacturing have already proved robotics at scale. Source: Photo by Homa Appliances, Unsplash Building an autonomy program Buying drones is easy. Building a safe, sustainable and effective program is harder.

The key components: ■ People. Remote pilots need the appropriate CASA licences and approvals, such as a remote pilot licence, and the agency needs an appropriate operator certificate for more complex operations. Pilots also need training in fireground safety, incident management and data handling.

■ Procedures. Operations manuals, risk assessments, maintenance procedures and emergency procedures for lost links and crashes. ■ Maintenance and logistics.

Batteries, spare parts, charging in the field and repair processes, especially during long campaigns. ■ Data management. How imagery is uploaded, processed, shared, stored and deleted.

■ Integration. How drone teams are tasked through the incident management structure, and how their outputs reach the common operating picture. ■ Volunteers.

Many volunteers have drone skills from work or hobbies. With proper training and accreditation, they can significantly expand capacity. Ethics: where should the line be?

As autonomous systems become more capable, emergency services will face real ethical questions. It is better to answer them deliberately, in advance, than under pressure during an event. ■ Decisions about life.

Machines can find people, but decisions about rescue priorities, triage and evacuation must remain with trained people, who are accountable for them. ■ Accountability. When an autonomous system contributes to a poor outcome, it must be clear who was responsible for its use.

"The system decided" is never an acceptable answer. ■ Surveillance creep. Capabilities acquired for emergencies can be tempting to use for other purposes.

Clear limits, published to the community, protect trust. ■ Fairness. Autonomous detection and monitoring should protect all communities, including remote and disadvantaged ones, not just those where data is easiest to collect.

■ Workforce impact. Autonomy should be used to make work safer and more effective, with a clear commitment to retraining and redeployment rather than replacing people. An ethics framework for autonomy, developed with the workforce, unions, volunteers and the community, and reviewed regularly, gives leaders a firm basis for these decisions.

The economics of autonomy Autonomy can look expensive, but its costs should be compared with the costs it avoids: ■ Injuries and deaths avoided by keeping people out of the most dangerous tasks. ■ Earlier detection, which keeps fires smaller and cheaper to suppress. ■ Faster damage assessment, which speeds up recovery and reduces the cost of disruption to communities.

■ More effective use of crewed aircraft, by giving them better information and using drones for tasks that don't need a crewed aircraft. ■ Extended workforce capacity, with one operator supervising several systems. Ongoing costs must be counted honestly too: training and licensing, maintenance, batteries, data storage, software subscriptions, security and replacement.

Programs that budget only for purchase often struggle to sustain themselves. Learning from early adopters Agencies in Australia and overseas that have built successful drone and robotics programs share several lessons: ■ Start small and prove value with a few well-chosen use cases, then scale.

■ Embed the capability in operations, with drone teams tasked through the incident management structure, rather than operating as a separate specialist unit. ■ Standardise equipment and procedures across the agency, so pilots and data can move between regions. ■ Share imagery widely, so the whole response benefits from what autonomous systems see.

■ Engage the community early, so drones are seen as a help rather than an intrusion. Making the data useful Autonomous systems generate huge volumes of video, imagery and sensor data. Without a plan, much of it is never used.

Agencies that get value from it do several things well: ■ Process at the edge. Onboard or vehicle-based computing can identify hotspots, people or hazards in real time, sending alerts rather than hours of raw video. ■ Standardise outputs, such as fire edge maps, flood extents and damage assessments, so they can be loaded straight into the common operating picture.

■ Tag and catalogue imagery by time, location and incident, so it can be found later for after-action reviews, recovery and research. ■ Share selectively, giving partner agencies the products they need while protecting sensitive imagery. ■ Feed the models.

Imagery of real fire behaviour and flood extents is valuable for improving prediction models, if it is captured and labelled consistently. Volunteers and autonomy Volunteers can play a major role in autonomy programs. Many already hold drone licences through work in agriculture, construction, surveying or media, and bring skills that agencies would otherwise have to train from scratch.

Volunteer drone teams could provide surge capacity during campaigns, local knowledge in their own areas, and a pathway for younger members interested in technology. To make this work, agencies need accreditation pathways that recognise existing licences and skills, agency-owned equipment so volunteers are not out of pocket, clear insurance and liability arrangements, and the same security and privacy rules for volunteer teams as for career staff. Barriers and how to overcome them ■ Regulation.

Beyond-visual-line-of-sight approvals can take time. Engage CASA early and build a track record with simpler operations. ■ Battery life and weather limits.

Plan operations around them, and test systems in realistic conditions before relying on them. ■ Skills. Pilots, maintainers and data specialists are in demand.

Combine career, volunteer and contracted capability. ■ Integration. Drones that operate outside the incident structure create confusion.

Task them through it. ■ Trust. Crews and communities need to see benefits.

Share early successes and be transparent about limits.

Partnerships with industry and research Australia has a strong base of expertise in autonomy: mining and agriculture companies that run autonomous fleets, defence industry developing uncrewed systems, universities and research organisations working on robotics and AI, and a growing number of drone businesses. Emergency services can draw on that expertise rather than building everything themselves. Useful forms of partnership include joint trials in realistic conditions, access to test ranges and facilities, research into how crews and autonomous systems work best together, and arrangements for industry to provide surge capability, such as drone mapping teams, during major events.

Each partnership should include clear agreements on safety, security, data ownership and privacy from the outset. Illustrative scenario: a fire followed by flood The following scenario is illustrative. Night one.

A fire burns through a valley. Two thermal drones monitor the containment lines overnight under a single operator, detecting three spot fires that are attacked by a standby crew before they spread. Crews on the line get much-needed rest.

Day two. A remotely operated dozer builds a containment line on a steep slope considered too dangerous for a crewed machine. The operator works from a safe location with live video from a drone overhead.

Day three. Air operations staff see an unauthorised drone near a helicopter operating area on the airspace picture. Aircraft are held for 20 minutes while Police locate the operator.

Day six. Heavy rain falls on the burnt catchment. A drone with a long-range camera monitors debris flow risk above a road, and an uncrewed boat searches a flooded creek for a reported missing person, while swiftwater technicians stay on the bank until the location is confirmed.

Day eight. Drone mapping of the burnt and flooded area is completed in a day, allowing assistance to reach residents sooner. In each case, machines did the dull, dirty or dangerous work, and people made the decisions.

Cyber security at the core: securing the command link Cyber security at the core: secure by design, not bolted on. Source: Darren Covington An autonomous system is only as trustworthy as its command link and its data. A drone or robot that can be hijacked, spoofed or fed false data is not a safety tool.

It is a hazard. Defence designs uncrewed systems on the assumption that adversaries will try to jam, spoof or take control of them, and emergency services should do the same. Key security requirements for emergency services autonomy: ■ Encrypted, authenticated command and control.

Only authorised operators can task or control a system, and every command is authenticated. ■ Resilience to GPS jamming and spoofing. Systems detect navigation interference and fail safe, for example by holding position or returning home.

■ Trusted supply chains. Hardware, firmware and software come from vetted suppliers, are kept patched, and don't send data to destinations the agency doesn't control. ■ Data sovereignty and privacy.

Imagery and video are stored securely in Australia, retained only as long as needed, and accessible only to authorised people. ■ Safe failure modes. Every system has a defined, tested behaviour when it loses its link, its navigation or its confidence.

■ Security in procurement. Cyber security and data handling are scored criteria in every autonomy purchase, alongside performance and price. Threats to autonomous systems ■ Hijacking or unauthorised control through weak or stolen credentials.

■ GPS jamming or spoofing, sending a system off course or causing it to crash.

■ Interception of video and data, exposing sensitive imagery. ■ Malicious firmware or software updates, inserted through the supply chain. ■ Data sent to unexpected destinations, such as overseas servers, without the agency's knowledge.

Security questions to ask in procurement ■ Where is data stored, processed and transmitted, and who can access it? ■ How are command links encrypted and authenticated? ■ How does the system behave when it loses its link or navigation?

■ How are software and firmware updates signed, tested and delivered? ■ What security testing has been done, and by whom? ■ How will the supplier notify us of vulnerabilities, and how quickly will they fix them?

Building security into the program Securing autonomous systems is not only a procurement task. It needs ongoing attention across the life of the program: ■ Asset register: know every drone, robot and controller, its software version and who is responsible for it. ■ Account management: individual accounts for pilots, multi-factor authentication for fleet management platforms, and prompt removal of access when people leave.

■ Patching: apply firmware and software updates promptly, after testing, through a controlled process. ■ Secure data handling: encrypted storage and transfer of imagery, with no automatic uploads to services the agency doesn't control. ■ Incident response: a plan for what to do if a system behaves unexpectedly or is suspected of being compromised, including grounding the fleet if necessary.

■ Regular review: periodic security testing of the fleet and its supporting systems. The next decade Looking ahead ten years, several developments could reshape how emergency services use autonomy: ■ Routine beyond-visual-line-of-sight operations, with drones launched automatically from stations or vehicles to confirm reports of fire or flood before crews arrive. ■ Persistent aerial monitoring of high-risk landscapes on bad fire days, combining drones, crewed aircraft and satellites.

■ Optionally piloted aircraft carrying out some suppression and resupply missions at night or in heavy smoke, under strict safety rules. ■ Ground robots as standard equipment for structural and hazardous materials incidents. ■ Remotely operated plant working on containment lines in terrain too dangerous for crewed machines.

■ AI on board, helping systems recognise fire edges, people and hazards, and sending only what matters to operators.

Each of these will bring new regulatory, ethical, security and workforce questions. Agencies that build sound foundations now, with clear autonomy levels, strong security, good data practices and trusted relationships with regulators and communities, will be best placed to adopt them safely. The prize Put simply: fewer people in harm's way, faster detection, persistent awareness and better outcomes for communities.

Autonomy won't replace the courage and judgement of emergency services people. It can make sure that courage is used where it matters most, with less risk to their lives. Putting it into practice: a staged roadmap First 90 days ■ Review current drone use across the agency, including licences, equipment, data handling and security.

■ Choose three dull, dirty or dangerous use cases for structured trials. ■ Agree autonomy levels and human roles for each. First year ■ Run trials in realistic conditions, with independent evaluation.

■ Establish airspace management procedures with air operations, CASA and Airservices Australia. ■ Write security, privacy and data sovereignty requirements into procurement. ■ Launch a public education campaign on unauthorised drones.

Years two to three ■ Scale proven use cases across regions, including volunteer drone teams. ■ Introduce beyond-visual-line-of-sight operations for mapping and monitoring. ■ Trial remotely operated heavy plant and ground robots.

■ Integrate all autonomous system outputs into the common operating picture. Measuring success ■ People kept out of harm's way: hours of hazardous work carried out by machines rather than people. ■ Detection: spot fires and changes detected by autonomous systems, and how quickly.

■ Speed of assessment: time to complete damage assessments after events. ■ Airspace safety: incursions and near misses involving drones and aircraft. ■ Security and privacy: compliance with procurement requirements and data handling rules.

What leaders should do next ■ Pick three "dull, dirty, dangerous" use cases and run structured trials this season.

■ Define autonomy levels for each task, with clear human roles and accountability. ■ Invest in airspace management and deconfliction with CASA and Airservices Australia. ■ Write security, privacy and data sovereignty into every autonomy procurement.

■ Engage the community early on how drones and robots will be used and governed. Questions for boards and executives ■ Which of our most dangerous tasks could be safely done by machines? ■ Do we have clear rules about the level of autonomy allowed for each task?

■ Where does our drone imagery go, who can see it, and how long do we keep it? ■ How would we know if one of our autonomous systems had been compromised? ■ How are we working with CASA, Airservices Australia and the community on safe airspace?

Glossary ■ BVLOS: beyond visual line of sight, flying a drone further than the pilot can see it. ■ Human–machine teaming: people and autonomous systems working together, with people in command. ■ Optionally piloted aircraft: an aircraft that can fly with or without a pilot on board.

■ RPAS: remotely piloted aircraft system, the regulatory term for drones. ■ Spoofing: sending false signals, such as fake GPS, to deceive a system. ■ Swarm: a group of autonomous systems that coordinate their actions.

Further reading ■ Civil Aviation Safety Authority, guidance on remotely piloted aircraft and BVLOS operations. ■ SAE International, J3016: Taxonomy and Definitions for Terms Related to Driving Automation Systems. ■ Australian Signals Directorate, Essential Eight Maturity Model.

■ Office of the Australian Information Commissioner, Australian Privacy Principles. Where do you see autonomy making the biggest difference first in emergency services? #EmergencyManagement #FireServices #Autonomy #Drones #Robotics #HumanMachineTeaming #CyberSecurity #BVLOS #Innovation #Safety #Australia Darren Covington writes on intelligence-driven emergency services, climate resilience, technology and cyber security in Australia.

Photo credits: Jason Mavrommatis / Unsplash; Darren Covington; Homa Appliances / Unsplash. Unsplash photos are used under the Unsplash Licence. Other photos and graphics © Darren Covington.