Anyone who has gotten a cringey response from ChatGPT knows that artificial intelligence is only as good as the data feeding it. But for AI systems that need to understand items in the physical world, such as buildings, farmland, power lines, construction sites or solar installations, there’s a powerful source of data, and that’s drones. Whereas the general public might be mocking AI and seeing drones as merely a hobbyist gadget or a delivery vehicle, there’s a symbiotic relationship between drones and AI that’s important to pay attention to.
Why AI needs drones
AI models that operate in the physical world need lots of data, both to train it and then ultimately to provide real-world insights.
An algorithm that detects roof damage for an insurance company needs to have seen tens of thousands of roofs — all in different conditions, but from consistent angles and at consistent resolutions. A system that monitors solar panel degradation needs a continuous stream of thermal imagery from operating installations. A construction site monitoring platform needs regular, georeferenced captures of every project it tracks.
Satellites generally don’t provide great data for those types of things, largely because the resolution isn’t high enough and the revisit rates aren’t fast enough for most applications. And of course, it can be tough — both practically and/or financially — to get this data from more traditional mechanisms like the ground, ladders or helicopters. Drones shift the economics to actually make this work, as they offer a scalable, repeatable, cost-effective way to put a precisely calibrated camera over a specific location, at a specific altitude, and on a defined schedule.
As an example of how important drones are to AI, FlyGuys, a nationwide drone data platform, delivered over 10 million images of assets in a single month for a single customer. Its pilots captured data at exactly 100 feet altitude, across all 50 states to feed an AI system that generated maintenance recommendations for building envelopes.
“FlyGuys is the data pipeline for AI,” said CEO Joe Stough in a prepared statement. “We capture and standardize consistent, high-volume data at scale, feeding our customers’ platforms where their AI models generate actionable insights.”
How AI can use data generated from a drone
Many people mock AI for wacky outputs, but its outputs are only as good as its inputs. Luckily, drone data can serve as a truly useful AI input.
Now there was an era in the drone industry before the AI boom where it felt like drone pilots needed to be capable of analyzing data themselves. Clients wanted the drone pilot to fly the mission and tell them what they see based on potentially thousands of photos.
However, AI has changed that workflow. In fact enterprise AI customers generally prefer to receive raw imagery rather than processed outputs. That might seem counterintuitive given that just a couple years ago the mandate for drone pilots was to offer the finished analysis.
Rraw data gives AI platforms the flexibility to run their own models, train on proprietary datasets, and generate custom outputs for their specific use cases. The drone operator’s job is to capture clean, consistent, well-documented imagery. What the AI does with it is someone else’s problem.
That has implications for what “quality” means in this context. For a real estate shoot, quality is about aesthetics, like nice golden hour light, smooth gimbal movement or a compelling composition. For an AI data pipeline, quality means something entirely different: correct altitude, correct overlap between frames, correct GPS metadata, correct sensor settings, captured on schedule, documented and delivered in the right format.
Ways AI depends on drones
Of all the AI-adjacent drone applications growing right now, digital twins may be the most consequential yet least understood outside of industry circles.
A digital twin is a detailed, three-dimensional virtual replica of a real-world asset or environment, built from captured data including imagery and LiDAR. Think of it as a living 3D model of a building, a factory, a construction site, or an energy installation that can be updated continuously as the physical asset changes.
Ways to use digital twins include a construction company that might compare the current state of a project against the design model and catch deviations before they become expensive problems. A facilities manager could conduct a virtual walkthrough of a building without traveling to the site. Or, a energy company can monitor the condition of remote infrastructure without deploying inspection teams.
All of those use cases requires continuous, precise, georeferenced data capture, which creates a niche for drones.
How drone hardware is evolving to serve AI
The convergence of drones and AI is changing not just what drones are used for but how they’re built, creating a need for improvements in certain aspects of their tech. For example, obstacle avoidance systems have evolved from basic proximity sensors to AI-powered vision systems. The best obstacle avoidance sensors can identify and navigate around specific types of hazards in real time, while also relying on machine learning to optimize flight paths for data capture efficiency rather than just following pre-programmed waypoints.
The cameras themselves are changing too. Thermal sensors, multispectral cameras, and LiDAR systems that once required large, expensive platforms are increasingly available on compact drones, making the data types that AI applications need accessible at a much wider range of price points and deployment contexts.
What this means for the drone industry
Sure, there are some extremely talented drone photographers out there, but the ones making serious money as aerial artists are rare.
Drone pilots and operators who position themselves as reliable, scalable providers of consistent, well-documented data rather than creators of compelling aerial imagery are likely the ones who will find repeatable work that benefits from this AI wave.
The post Drones and AI: here’s how the two technologies need each other appeared first on The Drone Girl.