Mach9

Get the latest jobs and industry news

Welch Comer Engineers has been surveying the Inland Northwest for more than 40 years. Today, most of the firm’s LiDAR data comes from drones. For drone pilot Thomas Tucker, that meant evaluating Mach9 on the data he works with every day. Before bringing Mach9 into the firm’s workflow, he wanted to see how it handled an aerial dataset. We processed one of his datasets during the evaluation. The results convinced him to move forward with Mach9. Here’s what his workflow looks like now, from flying the mission to exporting to CAD: → Collect low and slow as the project requires The collection altitude and speed are chosen to maximize point cloud density, which leads to the best results downstream for extraction. After collection, the data is processed to create a point cloud and an ortho TIFF. → Point cloud and ortho both go into Mach9. For top-down work, he keeps the ortho in the background and dials point cloud visibility down. This makes gravel roads and hard-to-see pavement easier to distinguish than in a colorized cloud, while keeping enough of the cloud visible to work with the underlying 3D data. → Mach9’s automated extraction returns line work and points as a drafted first pass. On one of his quarter-mile datasets, processing takes about 5 minutes. The automated survey comes back in about 30 minutes. → He runs every vertex himself and approves what's right. On his most recent dataset, he mostly added a few vertices here and there. By his estimate, half of those additions came down to preference rather than correction. → He builds the TIN surface in the same window. He drapes the points onto the cloud and checks flagged crossing breaklines before exporting the surface as XML. …then straight out to CAD.

2020: waiting on someone to export a drawing of the site 2026: adding planners, engineers, operations and maintenance teams to the point cloud directly When a customer buys Digital Surveyor to speed up extraction, the 2nd conversation is almost always about who else needs the data. Survey team has it. Then a project manager asks to look at the project Then someone on the bridge and structures side wants to verify a bridge and sign clearances. So when I set up a workspace, we work through three things: → Access levels: Who edits, who views, who comments → Entry point: Whether someone can open it without specialist software → Interoperability: Whether the output lands in the systems they already run Whether it’s planning, asset strategy, or project managers, most of them will never need to go in and edit or anything of that nature. They open it, review the part they came for, get insights they need, leave comments, and integrate it as part of their workflows Access to the point cloud gives them a level of collaboration that they traditionally didn’t have.

I have 10+ years of experience in civil engineering. So I’ve seen firsthand how long engineers wait on survey data before they can start. AI and automation are starting to close that gap. This week, Seth Gulich, PSM and I will show you how. We'll be presenting at an American Council of Engineering Companies (ACEC) webinar on AI-assisted data extraction, from raw point cloud to stamped deliverable. We'll cover: - how AI and machine learning find and extract features from point clouds and imagery - a practical field-to-finish workflow, from data collection through CAD and GIS delivery - building QA/QC around AI-assisted work, with the licensed surveyor responsible for the final product We'll share what we've learned running this on real projects, then open it up for Q&A. Highly recommend staying for the QA/QC section. Building review around automated output is a different discipline than reviewing manual work, and the licensed surveyor is still responsible for what goes out the door. It’s on Wednesday, October 21, 1:30 pm ET. *Attendees can earn 1 PDH. Link in the comments.

On a recent statewide project, the person who got the most out of our data had never opened a point cloud before. He sat on the asset management side, and his question was whether a load could pass under a structure. Normally that question becomes a request to the survey team. The survey team goes back to the file, and the answer comes back days later in an email. We decided to give him view access and walkthrough training instead. He drove the capture trajectory, clicked structures, and pulled up the collected image at each one until he had answered his own question about bridge clearance. He was able to build his own picture of the corridor at his own pace, without needing to go back-and-forth with the survey team. The data holds more than the deliverable asks of it. Put it in front of someone with a different problem and they find their own way to an answer. This is why I treat sharing as its own workflow. Extraction is the problem we try to solve first. Getting the same data in front of planning, asset management, and the people who inherit a project after delivery is a separate problem with its own tools. Even if most of them will never draw a line in it, they might get further than anyone scoped for.

🛣️ Nobody asked us to LiDAR-scan the crookedest street in the world. We did it anyway. Lombard street's famous switchbacks were built in 1922 to reduce a steep 27% natural grade, and it's been one of San Francisco's top landmarks ever since. At over 400 ft long with about 100 ft of elevation change, it makes for a pretty cool dataset. Here's Lombard as a 3D point cloud in Digital Surveyor, down to the last hydrangea. Where should we point the scanner next? Drop a landmark 👇 #RealityCapture #MobileMapping #LiDAR #Surveying #lombardstreet