(858) 350-4324 [email protected]

Project Overview:

In March of 2019, MODUS AI  conducted a demonstration and feasibility study to assess the efficiency of drone-based LiDAR vs. traditional survey methods for the Virginia Department of Transportation in the Warrenton area of Northern Virginia.
The purpose of this project is to replace the existing at-grade intersection of Route 15/17/29, Route 15/17/29 Business, and Route 880 (Lord Fairfax Road) with a grade-separated interchange to improve safety and capacity. Traffic conditions at this location include a high volume of northbound left turns into the town of Warrenton, a large (12 percent) volume of truck traffic, and local traffic associated with Lord Fairfax Community College, the Fauquier County landfill, and homes nearby. The Eastern Bypass currently carries approximately 43,500 vehicles each day. Traffic counts on Route 15/17/29 Business and Lord Fairfax Road are 11,000 and 2,000 vehicles per day, respectively. Total cost of the project is estimated at $27 million, and is being contracted to Shirley Contracting Company of Lorton, Virginia, with a completion date of late 2020.

Project Scope:

The project data analysis was performed using Blue Marble’s Global Mapper software.  The LiDAR mission was collected using the DJI M600 with  Velodye VLP-16 sensor integration. The scope of the project included four deliverable files:

  1. Ground classification LAS
  2. DEM in BIN format – dem file extension
  3. DEM in TIN format – dem file extension
  4. Contours in 0.1’ in dxf format.

We also created a 1-foot contour map for contrast comparison.

Equipment Selection and Why Drone LiDAR Was Selected:

The construction crew needed to have accurate maps/topography base knowledge before starting work. They already had basic layout drawings of the project but needed an assessment of how much volume cut and fill would have to be removed before they began work.

This type of work can be potentially be completed by six different methods; traditional survey crews, a road mobile scanner, a terrestrial scanner, drone photogrammetry, aerial LiDAR, or drone LiDAR. A brief introduction on the strengths and weaknesses of each method will help demonstrate why drone-based LiDAR was used for this project.

Traditionally, a survey crew could have performed this job doing a limited or finite series of points to determine the topography and terrain. A job of this size would have taken a traditional crew, from the pre-site survey to performing the field work, to then finally processing and drawing, around three to four weeks, not taking into account possible weather delays. Heavy tree and brush cover at the site would have also added to the job time.

The next surveying method could have been mobile LiDAR, mounted on a car or truck, and is traditionally used on roadways and mobility corridors. The system allows for a certain amount of distance from the sensor and is extremely precise. Road mobile LiDAR scanners are becoming more common in freeway accident reconstruction, as the incident scene can be mapped out in minutes and not requiring the freeway to be completely closed. This particular site had the added challenge of embankments that had other features on their far sides that would not have been accounted for. The project required data 100 meters from the road, which was not achievable due to the terrain and trees. This project was not a good candidate for mobile lidar.

The next survey method that could have been used was drone photogrammetry. Photogrammetry works well in unobscured areas like the road or an open field, with highly accurate results. However, depending on the level of accuracy that the construction crew requires, grass and other vegetation become an issue because the photogrammetry measures the top of the grass and brush, as well as the tree canopy, not the top of the ground itself. In early March the trees, grass, and brush are dormant and will not be a true ground representation in August.  For photogrammetry to be used exclusively for a project of this scope, additional seasonal collections would be necessary for accurate comparisons.

Airborne LiDAR (fixed wing and helicopter) may have been an option, but on a project this small (20 acres) it would need to be rolled into several other projects for the pricing to be competitive. For a time-sensitive project, such as disaster relief, this would not work or it would have to be special ordered. Other factors also need to be considered, such as fuel costs, point density requirements, and restricted airspace. At only 20 acres, this project would not be affordable with fixed wing aircraft. The specified point density would require a helicopter, as fixed wing would not get the required data returns without a much more expensive sensor e.g. the mini-Vux.  Also, in this area of Virginia, there are many horse farms that have small runways for private planes, making many of the areas restricted airspace. The only viable, cost effective, and timely way this project could be performed was with a drone LiDAR solution.

As part of MODUS AI’s standard practice, we conducted project feasibility, risk mitigation, and safety studies. These studies were done both from a workstation in the office and on-site as we walked the physical terrain to get a better feel for challenges, restrictions, and potential issues that might arise. Some of the standard factors we account for are the topography, total area of the site, tree cover/height and vegetation, other potential hazards, airspace, etc. Other questions we must answer as part of our pre-flight checklist include compliance with regulations, so we had to ensure that the airspace was available.

Could we maintain constant visual line of sight to the drone? Would we need to fly several small flights or could we do one flight using spotters and hand-held radios?  How many people were in the vicinity? In the military we called this human car and pedestrian traffic activity “pattern of life”, and it is a vital consideration for safety.

Based on the customer’s accuracy and product requirements, we determined in the mission planning phase that the VLP 16 was both sufficient for data capture and within their budget considerations. The VLP-16 achieves a relative accuracy of 3cm (0.2ft), which is considered survey grade accuracy. Based on the reduced vegetation levels in March, the returns from the VLP-16 allowed for better ground measurement with virtually no post-processing cleanup. Using the VLP-16 point density chart to determine the best altitude to safely fly while getting the highest point-per-meter squared, we used 40 meters AGL at 5 meters per second. This provided a collection of 350 points-per-million squared on the returns for bare earth areas and 35-50 points-per-million squared in more vegetated regions.

Our number one safety concern was not to interfere with people or potentially cause traffic accidents. This is a region that has two major intersections converging together, but thankfully, little pedestrian traffic to contend with.  This meant that there would be no drone overflights of people or livestock, which can be agitated by the drone. The nearest human concern was the local community college, but it was well outside our flight path.  Several emergency landing points were selected if we encountered problems and routes were chosen that kept the drone away from the roads and traffic. The time of day is an important factor in mission planning to avoid the morning commute, when a distracted driver could cause an accident just through the sheer volume of traffic.  In rare situations, a nighttime collection might be the best option.

During the on-site walk-through, we realized that we were within one mile of a tall radio tower. Radio towers are different from cell phone towers, as the radio tower can transmit a wide variety of frequencies with various power outputs.  Our concern was of potential radio frequency interference with the command and control of the drone. We flew a small DJI Mavic around the tower to determine that there was no RF interference, but for risk mitigation purposes, we maintained a quarter mile buffer around the tower.

Overall, we determined that the area was accessible and that the M600 could cover it safely with minimal risk to people or equipment.

The actual on-site field work took a total of two hours. The team had to hunt for the initial base station survey points that had been previously buried by the customer, adding to the delay. The flight was predicted to take 15 minutes, but took 45 minutes due to a minor technical delay (battery issue.) We returned the drone safely back to the prearranged take off point, changed the batteries out, and re-flew the mission.

After leaving the field, we began the geodetic correction and data compiling phase. The processing rinex file was uploaded into the telemetry software, the survey control points were added, and then a PPK telemetry correction was completed. This process took around ten minutes. Then the uncorrected LiDAR data was combined with the corrected telemetry data, and a geo-corrected raw point cloud was completed within the hour. The 3-D point cloud was then used in the creation of the data analysis products.

Some of the key points to take away from this project:

Start the project with the end requirements in mind. The customers data requirements are the most critical aspect of the entire project. Having the right sensor for the job, along with the sensors data definition table for altitude/flight speed/resolution, is paramount for successful data collection and a satisfied customer. Too many companies out there are rather cavalier during the collection phase and hoping to correct any issues that arise during the post-processing phase.

Have a strong safety plan in place before takeoff. Safety management and risk management needs to be in place, with established SOPs and emergency procedures understood by the entire team.

A well trained and knowledgably team of professionals is essential for a successful mission outcome. From the customers original introduction phone call to the final deliverable, there are literally dozens of complex steps in the workflow that can go wrong.  This simple and relatively small survey of a road network required five different software packages to complete. Flight mission planning, drone operations, telemetry collection, data compiling, along with point cloud creation of data products (DSM/DEM/Contour Maps, to name a few), demands technical competence that takes years to acquire.  Playing pick-up soccer with cheaper companies to save upfront dollars will be far more costly in the long run. Remember, you get what you pay for.

One last item regarding the customer deliverables. On this job, the surveyors wanted contour maps of one-tenth of a foot (0.1ft), which we completed, along with one-foot (1 ft) contour maps. The processing time to create the 0.1ft maps was four hours alone, along with two additional hours to upload the data to the customer’s project folder. At $120/hour for data processing, ensure that your customer understands the value added for such an endeavor. As the two different contour maps demonstrate, unless heavily zoomed in, it is impossible to distinguish the various levels of data.



Learn Drone LiDAR Mapping


Check out our professional LiDAR and Mapping Courses.  


Drone LiDAR Accuracy White Paper

We just completed a 3 LiDAR sensor grant study for CalTrans and California State University Fresno to produce an accuracy comparison white paper and the California DOT Drone Aerial LiDAR Standards. 

GET ON THE LIST TODAY and get to be one of the 1st to receive this report when it is released.  

You have Successfully Subscribed!