Small Unmanned Aerial Systems (sUAS or drones) along with exciting new image processing algorithms are enabling a fundamental change in how stockpile volumetric computations are performed. It is now possible to create accurate 3D models of stockpile areas using inexpensive cameras carried by a low cost sUAS (drone). From these 3D models, volumetric computations are easily derived. These new technologies can lower the cost and improve the accuracy of data collection as well as providing a non-disruptive solution that keeps collectors safe.
What mission planning software to choose, how much control and what configuration is needed for the required final product? The answers to these and other important considerations must be known prior to planning and flying a mission. For example, is the data going to be used for simple volumes or topo mapping? If so, what does that mean for how the mission is planned? Data that is not acquired in the right way for the needed end result is a waste of time and resources. Often, a lot of time is used in trial and error to figure out what can easily be conveyed in a succinct training session again reducing both time, effort and frustration associated with any new learning curve. GeoCue can provide the training needed to plan, fly, process, and distribute these extremely valuable end products.
Mission Planning is the process of creating the overall plans for on-site data collection. This is, of course, dominated by the drone flights but it must also consider supplemental data collection such as ground-based surveys in vegetated areas (if the sensor is a camera rather than a LIDAR) and, of course, the execution of the control/check strategy defined in the Project Plan. GeoCue recommends and supports DJI’s Ground Station Pro (GS Pro) for mission planning when using DJI drones. Technical training for GS Pro, camera calibration and control/check strategies can also be provided.
Accuracy, both horizontal (“planimetric”) and vertical is a critical element of a mapping project. The required accuracy is driven by the desired products. For example, a one-time stockpile volume computation does not require a tie to a geodetic network (e.g. being referenced to a State Plane spatial reference system); good local accuracy will suffice. A differential volume computation (e.g. a borrow pit) will require very high network accuracy. Accuracy strategies can range from simple scale checks to the use of image identifiable checkpoints surveyed in with a real time kinematic (RTK) ground-based survey kit. Ultimate accuracy is achieved by using a direct geopositioning system on the drone itself. GeoCue’s Loki system for DJI and other drones provides on-board multi-frequency direct geopositioning for the ultimate in high accuracy mapping. GeoCue provides training on specific strategies that can be used for obtaining acceptable project accuracy for the products being generated. GeoCue’s LP360 point cloud software contains a number of tools for mapping out control and check points.
Under the control of a mission plan uploaded to the drone, the flight is fully automatic. The drone, such as a DJI drone, flies a pattern of lines, automatically triggering the camera at the appropriate locations. The drone then self-lands and shuts down.
Trips to the field can be costly. Two trips due to not getting it right the first time is doubly so. An easy way to avoid these time consuming and costly trips is to ensure your data is correctly collected before leaving the site. For example, blurry photos are one thing that can negatively impact the point cloud. Using tools such as Agisoft’s Metashape Pro makes it easy to identify and check potential problems in the newly collected data. Finding these problems on-site rather that back in the office will ensure the proper point cloud solution saving both time and resources.
Global Navigation Satellite System (GNSS) post-processing includes the steps necessary to ingest raw data from the Loki direct geopositioning system, process to refined image station coordinates and convert to a format that can be used by downstream point cloud generation software such as Metashape. This is a simple, wizard-driven process contained in the ASPSuite software. The general steps include (All are carried out within ASPSuite):
Creating Point Clouds and Orthos is accomplished with software tools such as Metashape in conjunction with GeoCue’s ASPSuite. The images and flight log are transferred from the drone to a computer by a USB cable. The images are then processed in an image processing application (Metashape) that generates both a 3D model (as a dense collection of very accurate points) as well as an image map (an orthophoto mosaic) of the site. In addition to the image processing application, ASPSuite is often needed to do some height adjustments on data acquired with DJI drones and is also used in processing data flown with Loki, GeoCue’s PPK solution.
Accuracy assessment of the 3D point cloud and 2D orthophoto mosaic that will be used for subsequent analysis is one of the most critical steps in the workflow. Catching problems with horizontal and vertical accuracy as early as possible in the flow will reduce the amount of rework time. Accuracy assessment is perform within GeoCue’s LP360 point cloud/imagery processing application. Even if you are outsourcing your point cloud generation to a cloud-hosted provider, we strongly recommend that you download the generated point cloud and orthomosaic to perform accuracy testing. The workflow in LP360 is very straightforward:
Results are presented as individual statistics as well as American Society for Photogrammetry and Remote Sensing (ASPRS) Map Accuracy Standard designators.
Before the final solution of stockpile volumetrics or site topography can be achieved the point cloud data will need to be “cleaned” to remove unwanted structures such as conveyors or vegetation. This is accomplished by tagging these points in the point cloud and putting them into another classification so that they are not used in the stockpile volume calculations or when creating contours for a topo map. Cleaning the point cloud data is easily accomplished using LP360 for sUAS. In addition to automatic classification routines, having extensive manual classification tools available in both the map and profile windows aided by visualization capabilities in the 3D window make LP360 for sUAS the point cloud editing tool of choice.
Before a stockpile volume can be calculated a toe must be defined. Defining stockpile toes may not always be as cut and dried as it would seem. Often, the toe of a “textbook” stockpile is easily determined, these are the isolated stockpiles on relatively flat ground. However, what about “non-textbook” stockpiles such as overlapping piles on sloped ground with lots of overhead structures? Either scenario can be handled with ease with LP360 for sUAS. From the automatic toe extractor point cloud task where a simple click of the mouse creates a toe polygon and removes overhead structures to the advanced editing tools for more complicated piles, the solution of choice again is LP360 for sUAS.
Measuring volumes is really the sweet spot for drone mapping. The sites tend to be small, the frequency of data collection high and the accuracy of the systems more than adequate to satisfy financial auditors. Basic volumetric analysis is included in most Structure from Motion (SfM) software such as Metashape and a myriad of cloud-hosted solutions. These tools allow you to hand sketch a stockpile toe and then compute volumes. Only GeoCue’s LP360 contains the more sophisticated tools that you will need to handle more complex (which are quite common in the real world) cases of toe creation and volumetric computation. Scenarios handled by GeoCue’s LP360 include (the Base is the bottom of the computation and the “Hull” is the top):
The LP360 tools include additional features such as graphic images of cut/fill areas, selectable units regardless of coordinate system units, pile labeling and so forth. These volumetric tools have been (and continue to be) developed based on input from users in a wide variety of industries. They will provide what you need to handle a mixture of volumetric problems.
Quality Checks include the point cloud as well as features injected into (e.g. a priori bottom data) or added to a model (model constraints). GeoCue’s LP360 (sUAS licensing level) contains a robust collection of tools for addressing both point cloud and feature QC and cleanup tasks.
All point cloud data (whether from LIDAR or Structure from Motion, SfM) will contain anomalies. These anomalies are generally referred to as noise. Noise manifests as low points, high points or points that are somewhat spread even on flat surfaces. If these noise points occur in areas of modeling, they will have to be suppressed. This is done in point cloud editing by setting their classification tag to Low or High Noise. LP360’s point cloud cleaning tools include:
LP360 also includes Feature Analyst, a collection of tools for analyzing and repairing model constraint (and other) features. These tools address problems such as:
In addition, Feature Analyst allows scrolling though features at both the feature and vertex level for detailed yet rapid inspections.
The final step in any mapping workflow is to deliver results to stakeholders. While this can be accomplished by physically mailing storage devices, emailing files or providing File Transfer Protocol (FTP) links, it is not optimal in that it puts the burden on end-users to sort out and find the information they desire. GeoCue’s Reckon provides a cloud-hosted, centralized hosting site that allows users to log in to a secure site, visualized projects organized by data layers and download just the information they need. Of course, in many circumstances, users may only be interested in visually inspecting some aspect of the project or performing measurements. These can be accomplished directly from the browser interface.
Posting to Reckon is performed via the Reckon Publisher desktop application (provided as part of the Reckon subscription). After supplying appropriate login credentials, the analyst simply drags data files produced in the various analytic workflow steps to the publishing interface. Data can be designated as unique to this project (“temporal”) such as volumes or common to all times for the site (for example, a project boundary). General reports associated with the site can also be posted. This allows a site manager to deliver reports on machinery condition and other information not directly related to a mapping mission to stakeholders. When all data have been pushed to the site, stakeholders receive an automatic email notification informing them of the new postings.