When it comes to the top disruptive technologies of the last 10 years, drones are definitely up there and now, with the advent of AI, there are even more game-changing benefits for commercial drone users.
If you missed our recent webinar where we explored the specifics of capturing building envelopes using drones, we offered up some advice on the best workflows to optimise the safety, data quality and efficiency of building inspections. We also covered AI-powered advancements for automating inspection tasks.
Our webinar is now available to watch on YouTube but if you’d rather read than watch, here’s a recap of the strategies and best practices for capturing high-quality data of structures and then we’ll look at where AI is set to disrupt the building inspection process further.
Understanding the Users of Drone Data
While it’s easy to be mesmerised by the sight of a drone taking flight, the real magic lies in the data these flying cameras capture. It's this data that drives value, making the process of capturing and utilising it efficiently essential.Â
First and foremost, it’s vital to consider who will be using the data captured by drones. This could include internal stakeholders, clients, and partners. And it’s not just humans you need to consider. At some stage, data may need to be integrated into CAD and BIM software. AI systems also rely on structured data to function well, so well-organised and well-captured data is crucial if you’d like to be able to utilise any artificial intelligence tools.
The Complexity of the Built World
Buildings and structures present a complex and unstructured environment for data capture. From varying materials on facades to the intricate geometries of bridges and towers, each type of structure demands a tailored approach. Here, we outline some practical strategies to help you to navigate your next structural inspection.
Strategy 1. Divide and Conquer
When capturing data for buildings, think of the process in manageable components rather than as a whole. This approach is especially important considering the limited battery life and storage capacity of drones. By dividing the building into smaller sections and capturing these independently, you ensure comprehensive coverage without overwhelming your resources.
Capturing Short Buildings
For buildings that are longer or wider than they are tall, a top-down approach combined with orbit captures around the building’s facade works best. You’ll want the drone to capture images from above and then perform orbits around the building to ensure all sides are well-documented. This method provides a holistic view without needing to capture every facade individually as the field of view from the orbit should be plenty to capture the sides of the building.
Capturing Tall Buildings
Tall buildings require an additional step to make sure you collect the most comprehensive dataset. You’ll still want to divide the building into different components but in addition to capturing the roof and then the facades, you’ll also want to include an orbit of the building which will tie the two together. You’ll want his orbit to be at a gimbal angle so that you can see the parts of the building which tie the roof and facades together. These overlapping images are crucial for creating a seamless and detailed 3D model.Â
Tower Inspections
Inspecting towers requires a different workflow altogether. Towers benefit from flying the drone in concentric circles around the structure. This method promotes image overlap and aligns with the tower’s geometry, ensuring high-quality data capture of the entire building surface. Don’t forget to maintain a downward gimbal angle as this will help to avoid complications from the sky’s uniform appearance.
Bridge Inspections
Although bridges don’t have a roof or facade, you can still approach them in the same way you would a building. Consider both the top deck and the supporting structures underneath. Overlap is extremely important for bridge inspections. A double grid flight path can effectively capture the bridge’s horizontal and vertical elements, ensuring comprehensive data for both 2D and 3D modelling.
Strategy 2. Multiple Datasets for Capturing Buildings with Drones
To get the best results for any building inspection, combining 2 different datasets is key to giving you all of the answers you need. Strategy 2 may take a little longer but it’ll be worth it.
Collect one dataset, as above, for your 3D model.
3D modelling requires images from a distance with significant overlap.
The next dataset will be your inspection images - these images will be close-up shots which will help you to detect potential problem areas. Using separate datasets for each purpose allows for detailed analysis without compromising on either front, giving you a full 3D model for context and detailed images to allow you to inspect, annotate and report any issues.
TOP TIP: Leveraging Photogrammetry
Photogrammetry is the basis for creating 3D models from drone-captured images. It requires good overlap, minimal gaps, and capturing from the right distance from the structure to ensure you get the most accurate and detailed 3D model. Understanding the fundamentals of photogrammetry will help you plan your mission more effectively and collect better inspection data on the day.
The Role of AI in Drone Data Analysis
Now you’ve collected all of your data, this is where things get interesting…. Drone building inspections have already shaved hours, even days off the amount of time it takes to comprehensively capture images of an entire building envelope, AI stands to make those time savings even more apparent.
We recently announced a new feature in Hammer Missions to allow users to automatically identify specific structural features or issues using Spector AI. By training the system to identify specific issues, such as water ponding on roofs or cracks in surfaces, the AI feature can automatically find problem areas and highlight them for the user to add notes and tags. While human input is still necessary, Spector AI provides a powerful tool for streamlining the data analysis process, saving engineers and building inspectors hours of manually combing through images for defects and problems.
AI is still on a learning curve and Spector is no different. As usage increases, Spector will learn to detect even more features. Have a specific use case in mind? Our team loves a challenge! Get in touch.
Conclusion
To get the most benefit from drone building inspections you need to start with a plan. Employing structured, strategic methods to the inspection process will help you to gather the high-quality data needed for a functional 3D model and the detailed inspection images for retrospective analysis.Â
By considering the end-users of this data, understanding the fundamentals of photogrammetry, and then utilizing the latest tools, you can create a streamlined, repeatable workflow for you and your team to rely on. Whether you’re inspecting a roof or modelling a skyscraper, these strategies will help you achieve optimal results, time after time.