Best Practices with 3D modeling of communication towers (cell towers) with drones

Who should use this Guide?

Do you have a drone and would like to 3D model telecom assets? Are you an asset owner and would like to know about possible use cases? Are you a structural engineer seeking ways to improve your processes? Do you work in telecom and just want to understand how 3D modeling works? Keep reading and see if we can help answer any questions you may have relating to 3D modeling cell towers.

It’s no doubt that 3D modeling of infrastructure is omnipresent in today’s world.  In engineering, construction sites have been the primary vehicle for introducing 3D technology. Whether using ground-based lidar or aerial photogrammetry from a drone, we see more and more 3D modeling on job sites. Telecom, on the other hand, has been one infrastructure industry that has seen a slow adoption. Some companies have started to look at the technology but have not fully embraced or recognized the usefulness, efficiencies, and cost savings of 3D modeling. The businesses that have adopted 3D modeling are finding significant benefits.  

In this guide, Ryka will share with you some essential steps on how to produce clean, accurate, and presentable 3D models. To begin, we will assume you have a basic understanding of what your drone is capable of.   Images that you capture should contain GPS and IMU data and have the fundamental principle of photogrammetry.

Software Selection for processing telecom 3D Point Clouds

Pix4d is hands down the best choice for working with telecom assets. The software has become refined in the past few years providing many toolsets for vertical infrastructure. The ability to refine your measurements with 2D images is an essential aspect of creating accurate final products. Pix4D has provided this tool as a built-in function to their software. Towers have many intricate parts and having the ability to retune your measurement lines is insurance that you measure the correct item accurately.

Processing the datasets in Pix4D should be relatively easy if you start with a quality dataset with accurate GPS points. However, GPS is not the only form of scale you should rely on to extract measurements. The best way, in my opinion, is to establish a form of control. The control you select can be verified measurements, tower members, and/or CAD drawings to match precise measurements.   These known measurements can be applied as scales within Pix4D. More info on setting up scales within a project here.

Understanding the Science of Photogrammetry

Photogrammetry is the primary method Ryka uses to create 3D models of cell towers. We choose photogrammetry because of the ecosystem of software tools available to support image processing and the workflows associated with creating effective deliverables.

As defined by the American Society for Photogrammetry and Remote Sensing (ASPRS)  photogrammetry is a science and an art.  But how does photogrammetry work? Photogrammetry is a method in which you use overlapping photos that have known locations to triangulate points between images. In figure 1, you can visualize the overlap between images; this is the backbone of producing accurate and useful models.

Figure 1 Overlapping Images from a drone
Figure 1 Overlapping Images from a drone

Achieving the needed photo overlap to accurately 3D model a tower is a challenging task. Vertical infrastructure needs symmetrical data, and the overlap should be consistent to help the software identify key points between images. Figure 2 is the key point identifier in the Pix4D processing report.

Figure 2. Viewing the overlap
Figure 2. Viewing the overlap

This image is an excellent way to visualize the structure of your key points. If the key points are minimal between images, reevaluate the data collection and consider re-shooting the tower.

Camera Gimbal position for mapping cell towers

 For cell towers, we will typically be collecting low oblique imagery at an angle of around 45 degrees below the horizon. Rarely will we use nadir imagery unless creating 2D image products (Orthomosaics & Ground Images), then, in this case, we will fly the gimbal in a Nadir position. We recommend avoiding high oblique imagery and avoid capturing any sky in your photographs. This prevents challenges in processing and will cause the workflows to be slightly more problematic.  If capturing the sky is necessary, add manual tie points or annotate images within Pix4D. Collecting at least one dataset with low oblique imagery will help you build a foundation, then to get the upward detail add high oblique imagery as needed. Manual tie points will help the overall results when using multiple angles and are always recommended. Figure 3 shows the difference between Nadir, Low Oblique, and High Oblique angles.

Figure 3 Differences between Nadir, Low Oblique, and High Oblique
Figure 3 Differences between Nadir, Low Oblique, and High Oblique

If you would like to learn more about airborne photogrammetry in depth, this DOCUMENT is a great resource for aerial applications.

Planning a Photogrammetry Flight with your Drone

Planning the flight for your drone is the second most critical step. Without good quality data, none of this would be possible. If you have attempted to process towers in the past with minimal results, look back on your data and ensure you are following the principles of photogrammetry. Each step in the flight plan should be well thought out. Images and flight lines should have good continuity while avoiding taking duplicate photos in the same location. Images should be free of blur and in focus. It is a good practice to start using manual focus. Autofocus could cause a Bokeh effect, causing problems in processing.  If a human eye can notice flaws in the image, a computer will discern and amplify those flaws.

Two flight methods

Every Cell tower has slight construction differences. Thus flight plans must be adjusted to meet the needs of every varying site. This discussion reviews two types of flight methods: the column flight lines, and the orbital flight lines.

Orbital Flight lines


Orbiting consist of flying in a circular spiral making smooth verticle transitions around the tower. It’s important that each 360-degree orbit is at the same altitude to retain an 85% overlap and side lap. After the 360 is completed; move down about 10ft  to start the next 360. Fly these until every bit of the tower is covered. Certain drones have tools to automate and assist with this flight plan. In the figure below, an orbital flight plan is shown.

Orbital Flight Lines

Column Flight lines

Column flight lines come in handy when flying around towers in limited space, often close to obstacles and obstructions.  Trees, guyed wires, and other structures may prevent flying orbitals. In this case, select a column flight plan. In figure 4,  a column flight plans used to avoid the guyed towers.  With a column flight plan, it is harder to maintain the image overlap. Each column fight line should neighbor other adjacent columns as close as possible to preserve good side lap. Adding a camera operator and visual observer will help avoid any accidents and remain safe on site.

Spiral Flight Lines 2

Maintaining Flight Overlap & Data Continuity

It’s not always possible to keep image overlap at 85%; it’s very challenging for even the most skilled pilots to fly column flight lines as these must be flown manually. In the northwest, we have many tree and obstructions that we have to navigate around. When dealing with hundreds of images, it can be very challenging to verify overlap in the field. You may be able to find tools to help analyze, but in our experience, these are not always accurate since they don’t know the structures distance from your images center point (LAT, LON, ALT).  If you need to stop the flight for any reason (flight time, distractions, etc.) make sure to start from where you left off to avoid missing data.

SUMMARY

This discussion briefly covered software selection, photogrammetry, gimbal positioning and flight paths. With the final 3D model in mind, each area presents challenges for accurate data capture and processing. Processing oblique imagery requires substantial computing power and good graphics cards to support high-fidelity 3D models. There is no set amount of photos needed but expect to collect 500+ to make a complete model. Collecting more data is always better than having to revisit a site.  Unnecessary images can be filtered out to make an entirely accurate model.

 

So, What are the uses and how do you apply them to the telecommunications industry?

Usually, the first question asked when viewing a 3D model of a cell tower is how does this benefit the client? It’s an important question. However, there is no one answer as many customers today, use the data in different ways.  The 3D model of the tower is likely not the client’s final product. Understand the client’s needs and goals before designing an execution plan for the field which will not only benefit your team but also help refine your process and figure out what works best on site.

Some of Ryka’s current uses for 3D modeling a cell tower:

  1. Mount designing
  2. Structural Analysis
  3. Inspection
  4. Data Cataloging
  5. Visual Design
  6. As-builts
  7. Unclimbable Towers
  8. Enhanced Photo Simulations

Ryka has the required certifications, flight skill set, operational experience and capability to develop execution plans for any of these uses. If you are interested in learning more about the services Ryka UAS offers or have further questions about modeling cell towers, you can reach us at info@rykauas.com.

 


Also published on Medium.