De-Clouding the Point Cloud with Mobile LiDAR

De-Clouding the Point Cloud with Mobile LiDAR

A couple of years ago, we posted a blog titled LiDAR is Gaining Momentum” which is a great overview of LiDAR technology, “point clouds”, and how 3D mapping can be applied for a preliminary site assessment. With the increased popularity of autonomous vehicles and other mapping applications that utilize LiDAR, laser scanning has seen tremendous growth in the past few years. The sensor costs have come down and the quality has gone up to the point where it is a reliable, scalable field survey solution.  Our last blog focused on aerial LiDAR which is collected from a plane or helicopter. This blog post focuses on the advantages of mobile LiDAR data for design work.  

Overview of Mobile LiDAR
Mobile LiDAR utilizes scanners attached to a car or ATV like you may have seen atop a self-driving vehicle or Google Streetview car. The biggest difference between it and aerial LiDAR is that the “point cloud” density can be accurate enough for survey grade collections, with point-to-point distances able to be measured within 2cm accuracy. While not as fast as an airplane, a metro area that would take months for an engineer to walk and survey can be collected in days– literally as fast as the speed limit.

Turning Data into Information
Big Data may be coveted and raw “point cloud” data has plenty since it is generated by spraying 700,000 points per second. The challenge with achieving value from Big Data is turning it into usable information.  For LiDAR this key lies in “Feature Extraction.”

In GIS, the term “feature” refers to a unique object that is classified as either a point, a line or a polygon. For example, streetlights (point), sidewalks (line) and grass area (polygon) are all “features.” Features also have properties called attributes. A streetlight might have an attribute for its height, a sidewalk might have a material, and a grass area could have soil information.

By defining features, attributes and the extraction criteria, raw points can be translated into real world objects that a designer or engineer can use. Feature extraction can be done in many ways, which can be as simple as manually tracing points or as complex as running machine learning algorithms to recognize patterns. Ideally, extraction is automated in some fashion. For example, all streetlights look very similar in shape and orientation, so object recognition software could run through an entire metro area and classify anything that looks like a streetlight.

Mobile collection not only has higher accuracy, but driving a site provides the perfect opportunity to collect HD imagery as well. Like laser scanners, camera quality has grown dramatically and high-resolution photogrammetry can be accurate enough to produce measurements as shown in our blog describing cell tower audits via drone. The final product is 360 degree imagery which simulates a virtual site walk, all overlaid with LiDAR. Basically, imagine Google Streetview in HD where an engineer can measure any object in sight.

A measurement being taken from the edge of a street to a culvert within a “point cloud”

Utilizing for Design
Once features have been extracted we have a 3D model of our entire site with intelligent properties. Since the collection was geospatially accurate, it can also be integrated with any other GIS data such as streets, buildings, utilities, or geological data.  The field data should also be accurate enough to use as an input for automated design processing. Let’s imagine designing a smart lighting system for an entire city. The streetlight locations could be loaded and run through an automated design to calculate the fiber optic path and network components. This would be able to estimate the fiber cable footage and project cost.  More importantly, an engineer can use the information to lay out a proposed design right in the 3D or 2D model. When moving into CAD for construction drawings, 90% of the drafting leg work is already done. There is no need to draw streets, sidewalks, trees, catch basins, hydrants, etc. Even entering critical dimensions and offsets is unnecessary as they can be created on the fly.

As engineering projects gets larger and more complex, the ability to maintain quality is crucial. Skilled resources are in high demand and their time is valuable. Traditionally, a revisit was required if a dimension was missed in the field.  With mobile LiDAR it can be as simple as opening up and re-measuring from a desktop. An engineer can spend their time doing actual design and save the tedious work for the machines, all with unprecedented speed and accuracy.

Of course, everything we discussed is still above ground and we don’t have X-Ray vision just yet. The next revolutionary fielding technique, and possibly a blog post topic in the future, is Ground Penetrating Radar (GPR) which can be used to create a map of all material and utilities underground.  More to come...

About Patrick Hutto

Email address:

Patrick Hutto is the Director of Technology for the Network Design team at Foresite Group in Birmingham, Alabama. He has over 10 years of experience designing and delivering networks from high speed trading platforms to citywide FTTH build outs. Patrick has a strong technical and programming background and strives to find or develop solutions to increase productivity and quality on complex infrastructure projects.