Despite a recent rise in the potential wages for drone pilots, artificial intelligence threatens to replace operators entirely. UK Future Flight Challenge member, Adrian Westwood, looks at the future of drone pilots and how AI could impact future opportunities.
Following on from the National Careers Service report that indicated drone pilots are earning up to £65k a year, is it time for a reality check, or, should I say, reality cheque, and is time running out for drone pilots all together?
Drone pilots will become drone chauffeurs and porters, and the more experienced ones might remain as spotters. Very few sectors will be immune from autonomous UAVs. Rather than £65k, drone pilots will more likely be minimum wage roles.
UAVs are a disruptive technology which, in turn, will be disrupted. Within five years, UAVs will execute geo-spatial mapping, surveying and inspection missions fully autonomously, with no need for a pilot.
They will take off, fly a mission, return, download the data, recharge or change their batteries in portable docking areas, and be ready for the next mission…controlled by software and AI from a remote control centre hundreds, if not thousands of miles away…cue existing military UASs.
The human element
On your PfCO courses you are reminded the weakest link in the Unmanned Aerial System is the human element – the pilot. As drone technology becomes more reliable, consistent and safe, we, humans, become the variable factor in the chain of operations. We even have the I’M SAFE mnemonic to recite.
Platforms are being designed not to require humans in the loop. However, for regulatory reasons the old ways are still holding out, as we do not have the proving grounds to fully exploit existing technology, let alone experiment with the new rigs, software and airspace management systems.
The possibilities are being discussed and strategic relationships being built, and the UAVs and computer architecture are being designed and built, as we speak, with that future in mind. Additionally, complementary research and development on the use of AI will make the analysis of datasets autonomous, as well. Some pilots might hold out in congested areas.
The current drone pilot network system is like Uber taxis. The client wants a drone pilot, contacts one of the listing companies and organisations, the mission requirements are distributed, client and pilot are matched, and the flight executed. The same is true of direct contact between client and operator. This will all be computerised and autonomous.
Software platforms that connect UAV operators with companies that need their services, and which generate plans for autonomous flights and process the collected imagery have already been designed. Improved payload sensors, real-time detect and avoid systems, geo-fencing, comprehensive situational awareness software and remote electronic identification, UAV design and improved battery flight times mean BVLOS will be standard.
Fully-autonomous UAV missions will evolve as CAA regulations evolve. BVLOS flights will become more relevant to autonomous mission planning, mapping and UAV flights. Hopefully, the rising economic value of UAVs to many industry sectors will influence the legislators to instruct the CAA to reduce its restrictions on some BVLOS.
Clients are buying commercially-valuable insights extracted from data collected autonomously. We all know UAVs, rotors, fixed-wing and hybrid VTOL, are ideal for large-scale repeated and structured spatial data collection. UAVs collect the data by repeatedly and accurately flying the same path in a 3D space but offering 4D (time) visual insight. This enables our clients to identify changes and patterns. Clients want end-to-end solutions, quickly, efficiently, to specification and within budget. Fully-autonomous drones will be ready to operate within the next five years.
Autonomous mission plans for each of a construction company’s multiple projects will repeatedly collect 360-degree images, ortho images, DEM, DTM, image series, and other data types and deliver them to the client through a web interface. In addition to monitoring construction, this autonomous data collection will be used to monitor existing infrastructure, including mobile phone masts, electricity pylons and lines, power stations, railway tracks, pipelines, bridges and roads.
The new workflow, from request to delivery of data, follows clearly defined stages. It starts with the end-user setting up a cloud-based account to arrange data collections on specific sites, infrastructure and assets at a specified frequency and to store their data.
Whenever a collection is requested, the system automatically analyses the relevant airspace to ensure it can be flown legally and safely and to preview any other planning issues. It will check weather conditions and decide on camera and gimbal settings, heights and angles of approach for the collection of data at the optimum resolution and clarity.
Next, it generates an autonomous flight plan for the project and asset in question. This is saved for repeated use. The autonomous flight plan is then sent to the mobile phone app, which the UAV pilots use to receive their mission information and parameters.
On site, they do their standard pre-flight tests, upload the mission, click start and monitor the UAV for safety as it takes off and begins to capture the pre-defined dataset for that project. The UAV pilot can change batteries and the UAV will resume autonomously where it left off, unless one of the recently-developed charging docking stations is used.
All imagery data is being tracked remotely and, as the UAV is flying, metadata is being written into all the imagery captured. After the capture stage, a certain amount of data can be uploaded straight from the UAV via the mobile phone app, up to the cloud to be sorted, organised and processed.
For larger datasets, the drone pilot will upload the data through an online portal. There is no sorting required, which is a big issue in data management as UAV operations scale up. The drone pilot just takes the SD card, uploads it, and AI auto-sorts the data to produce the deliverables requested by the client.
At the back end, AI processes the images – for example, stitching them into 360-degree photos, creating orthomosaics, or DEM/DTM, tiling them for viewing on the web, or generating point clouds and 3D meshes – whatever the end client wants. These various analytical tools will plug in autonomously at the beginning of the process based on the requested mission plan.
In the final step, the platform delivers the requested insights to the user via the same interface they used to place their request. The data will be saved to be compared to repeated mission data, giving 4D insight.
With the speed at which UAV technology and command-and-control software and airspace management systems are being developed, it will not be long before the drone pilot is defunct. There will be the need for the chauffeur, the porter and the spotter as fully-autonomous UAVs will be flying well before driverless cars are common on the ground. Airspace 2025 will be very different.
Adrian Westwood operates Drone Reconnaissance Services in Ken, Surrey and Sussex, offering aerial survey, mapping and inspection services. UK Future Flight Challenge is a consortium seeking to explore how the unmanned aviation industry will develop over the next five years.