USING POINT CLOUDS FOR TRANSPORTATION PROJECTS Ron Grant, Director of Marketing, Roads, Bentley Ray Filipiak, Global Pre-Sales Technical Manager, Bentley Presented by Sponsored by
TABLE OF CONTENTS Introduction... 1 Reality modeling... 1 Industry drivers... 2 OTHER BENEFITS OF point clouds... 2 LiDAR... 3 Acute3D... 4 Typical point cloud workflow... 7 Preprocess and creating point cloud deliverables... 7 Processing data... 7 Clipping and Sectioning Point Clouds... 8 Creating terrain models and 3D objects... 9 RGB values... 12 Coordinate systems... 14 Tiling... 15 Creating Terrain Models... 16 No classifications?... 18 Other line and shape extraction... 23 Bringing it all together... 26 Conclusion... 28 Contributors... 29
INTRODUCTION Data acquisition is a requirement for any civil project. But it can be expensive, time consuming and even dangerous. The increasing prevalence of point clouds, coupled with the advent of drones for gathering them is reducing costs and improving safety. On top of that, there are now many tools available that make point clouds easier to use, and many methods for extracting relevant data from point clouds. There are also many innovations on how those point clouds are prepared for use on transportation projects. But how is this done and what are some real-world applications for this content? REALITY MODELING Point clouds provide us with the ability to model our world as it really exists. The term for this is reality modeling. But why are we focusing on point clouds, visualization tools, and reality modeling and their relationship to 3D modeling in the engineering workflow? Figure 1 1
INDUSTRY DRIVERS The Federal Highway Administration s Every Day Counts initiative focuses on bringing 3D modeling into the world of transportation engineering. Much of this involves using point clouds or LiDAR-based data to either create the models of existing structures along the roadway, or provide asset management. The Administration is behind digital asset management, but one key element of this initiative is to provide a safe way to collect data. LiDAR and other means of collecting point clouds offer this safe way of collecting data. This is welcome news to anyone working within this industry who knows just how dangerous these types of projects can be. OTHER BENEFITS OF POINT CLOUDS Beyond issues of safety, and other industry drivers, we are focusing on the use of point clouds because they bring many benefits. When carefully incorporated into a project, point clouds can: Increase productivity Shorten project timeline Decrease project cost Enable better collaboration Improve decision making Improve Information Mobility Increase safety 2
Figure 2 LIDAR Point clouds are most often created using LiDAR. Though it might seem like a modern invention, LiDAR has been around for some time. It was invented during World War II as Allied forces were looking for different ways to capture information. They looked at light intensity as a potential to capture data, but LiDAR quickly lost out to sonar and radar since laser focus was not advanced enough, and there wasn t enough computing power developed at that point to address this issue. It wasn t until the 60s and 70s that laser technology became more mature. During this time computers began to advance much faster, allowing for innovations in laser technology. With the vast computing power available today, we can accomplish great things. For instance, today we can stream data to any desktop in the form of a scalable DTM. A scalable DTM works much like 3
Google Earth where, from a distance, you can see a vast array of information, but as you zoom in closer and closer you can start to see streets, hotel locations, street names or other more micro information. In this way, the closeness of your view determines the amount of information you are presented with. A scalable DTM works along the same line. For example, if you wanted to know the contours of something that is 50,000 feet, you could break this up into smaller pieces and analyze the data at 20 or 30-foot intervals, seeing a more detailed view of the data. Combining technology like LiDAR and computing power unveils new possibilities. Today, point clouds are very powerful, they can be a tool that we use in preliminary design work, or throughout the asset lifecycle. They're becoming more and more a part of our everyday workflow. PROJECTWISE As streaming and other kinds of data portability solutions becoming a big part of the modern world, it becomes our responsibility to uphold quality of data. Today, it is easy to transfer large data sets onto a flash drive and hand that off to someone. But if this data changes, how will you ensure that person has the most relevant information? It is a responsibility we face today and one which requires strict attention. This is where project management tools like ProjectWise come in. This technology allows you to manage data, but it also allows you to limit the portion of the data that you may be using. This becomes relevant when you are dealing with large-scale projects that might span miles of road, for instance. If you only are worried about a portion of that road, ProjectWise will allow you to stream this data anywhere in the world and manage it effectively. ACUTE3D Rich 3D models can be created using Bentley s Acute3D. These models are created by shooting multiple photos of a site either with a cell phone or a drone. 4
Figure 3 This obviously has uses for many industries, but the transportation sector is a particularly effective arena for these applications. With as little as 40 pictures taken with a cell phone camera, 3D models can be generated in a short amount of time to reflect important building components such as conduits and pipes. Modeling as-built structures with these applications provides an accurate, effective representation of the current structural components. On a larger scale, flying over an interchange like this one in Japan can allow for the creation of effective 3D meshes that can aid in transportation-related projects. 5
Figure 4 Figure 5 6
TYPICAL POINT CLOUD WORKFLOW Though there are many ways to create a point cloud, the typical process involves having a service provider collect and register the point cloud data. From here, this service provider processes the information to create a deliverable. This might involve bringing that data into an environment to integrate into a design or producing a hybrid model of a 3D model along with the point cloud information. PREPROCESS AND CREATING POINT CLOUD DELIVERABLES In the initial stages of a project, Pointools can be used to prepare data. The application is intuitive, easy to learn and requires minimal training. It takes the pain out of working with large point cloud files and is useful because it delivers meaningful output and enhances point cloud workflows. For the processing portion of this workflow, tools like Descartes will allow you to extract features, integrate with design, manage large terrain models and produce hybrid deliverables. This 3D imagery advanced processing allows for better image processing, terrain modeling, point cloud processing, hybrid raster/vector and 3D modeling. PROCESSING DATA When it comes to actually processing the data once it s in the acquisition environment, any type of point cloud can be used in Bentley Pointools. Users can look at the point cloud in animation mode or rendering mode, and even perform tasks like clash detection. Essentially, Pointools is a tool that brings the data in and allows you to view and manipulate it so that your design team can create rich models. A typical point cloud might consist of billions of points. This is often the case when multiple point clouds are merged to create one massive dataset. High points, low points and bounds of information can all be seen within Pointools. You can turn information on and off within Pointools and decide how much of it you need within the view stream to be able to work with it effectively within that environment. Data can also be recolored to fix things like reflectivity that may not indicate the real information. To adjust these discolorations, a paintbrush or 7
eyedropper tool can pick the color you want to use from the data and recolor the information to make it be more reflective of what is really in the field. This ensures that you have the most accurate representation of the true data. Looking at a point cloud made up literally of billions of points can be a bit overwhelming. With Pointools, it s easy to zoom in smaller portions of data, and turn features or surrounds off and on to allow you to focus properly on what you intend to study. And with the ability to rotate and change your view, you can look at data from a variety of perspectives. Figure 6 CLIPPING AND SECTIONING POINT CLOUDS Filtering out data allows you to provide a more useful end product for your client. This allows you to ask yourself, what do they really want? Do they want to just see bare earth? Or do they want the whole 3D model that s created representing that ground? As data moves into the design environment, there are many tools available to manipulate the point cloud data. When working with a large point cloud 8
you can isolate a volume of interest with the Clip Box tools. The Section Viewer allows you to move a thin Clip Box through your data. Figure 7 Figure 8 CREATING TERRAIN MODELS AND 3D OBJECTS Applications like Bentley Descartes can be used to manipulate point cloud data then MicroStation or InRoads can be used for terrain creation and manipulation. So how does this work in the transportation industry? 9
The main goal of the following example is to look at some of the features that are delineated by the different reflectivity or RGB values from a point cloud to be able to extract break lines or other feature objects like copper rails. With this information, a 3D model can be created. There are many ways to look at data, so it is important to choose the right design presentation environment. There are also numerous ways to move around the point cloud data within the design environment. In the following example, the data was colorized based upon the classification and what was reflected back to the scanner when it picked up the data. Here, you can read intensity, elevation, or classification of all points. There are many different ways to view this data. First there is bare earth. Figure 9 There is elevation intensity. 10
Figure 10 And there are ways just to view the raw point with no reflectivity associated with it. Figure 11 11
Changing the views in this way gives you as the designer different ways to look at your information so you can decide the right backdrop to allow you to get the desired effect from that point cloud information. RGB VALUES Capturing RGB values with your point cloud data provides effective, photo realistic information. When you combine this with orthophoto, you ll have high quality point cloud information that you can use in a design environment. Figure 12 12
With programs like InRoads, you can set up cameras within your point cloud and walk around virtually. As you move around the point cloud, you can shift your perspective to look at a wider expanse of data, or to narrow in. Walking around a point cloud in this way gives you a good perspective of what needs to be done for your project. It also shows you what information is out there that can be used for your design. You can set the width of your camera, walk forward or backward, and look left or right as you make your way through the point cloud. Figure 13 With this ability to visualize your projects within the point cloud, you don t need to go into the field and visit projects as frequently. Having the option to see what s in the real world without actually going on site opens up many possibilities for reality modeling. 13
COORDINATE SYSTEMS With the right coordinate system, you can re-project on-the-fly. So if you didn t use the right coordinate system, you can easily transform your data into the right geodesy. Figure 14 It s also possible to merge different datum together as needed. Figure 15 14
TILING About 25 years ago, computer speeds struggled to keep up with the dense raster imagery picked up from photogrammetry. In order to improve this, tiling was created. Tiling essentially allows you to take large data sets and look at or use only information that falls within that block of tiles. Figure 16 These tiles can be whatever size you choose. The process itself is efficient and fast, so you don t need to take a break or wait for the information to load. This information can be streamed anywhere you chose. You can also choose only to stream the project information relevant to what you are working on. 15
CREATING TERRAIN MODELS Creating terrain models can be done directly from a point cloud using tools like GEOPAK, InRoads or MX. You can create terrain models from the whole point cloud, or just from one area. Then you can create contours or slope vectors. You can also colorize based upon elevation slopes or other features. In the following example, bare earth is captured within a terrain model. First, a boundary is set up. From here, a portion of the point cloud was extracted for use in civil design products. With this tool, the user was able to specify edge length, maximum triangle length, pitch boundaries, remove exterior slithers, and input points into a terrain model. Figure 17 Once the terrain model is created, display tools within products like InRoads can turn on triangles. The data within the triangle might be tighter or looser. 16
Figure 18 Figure 19 17
NO CLASSIFICATIONS? Typical data collection for a point cloud project might involve working with millions or even billions of points. But maybe the only thing you re interested in are break points. Or, maybe you want to pick up points every 25 feet on a grid within a model, instead of every centimeter. But what do you do when you don t have classifications? How do you snap to points, drape points, or pick up cross section data from your massive dataset? This example looks at a rail line. Using Descartes tools, top of rail points along this rail path are gathered. This can be done by grabbing the highest point in the model and setting up parameters within a cylinder within two or three meters. Figure 20 18
Once the highest point within the boundary is established, points can be connected as a top of rail. This is done by identifying and snapping to a point. The highest point within this cylinder can be connected to the top of rail. This data can then be used by civil designers. Once top of rail data is captured, the next thing to do might be to gather drape lines or other linear features parallel to the track. This done using MicroStation. Figure 21 In this case, the user paralleled the top of rail and created linear features. Because they are at a parallel point, they have no elevation assigned to them. These features are then draped along the point cloud. 19
Figure 22 From here, the user can specify that he wants to pick up vertices that cross a point in the point cloud, or only pick up every 25 or 100 feet. Next, the user wants to pick up random points outside of the road bed section of the track. To do this, he uses an array tool to specify that he wants to lay five points perpendicular to the outside linear feature. Figure 23 20
Once these points are laid down and evenly spaced, the array tool can copy those five points to create a 5 foot grid of random points along the track that are draped to the point cloud. Figure 24 With these steps an accurate but less dense model can be created. From here the user picks the points that he s going to array to and copies them along parallel to his track. Figure 25 21
Another tool allows you to look at lines and trace them based upon intensity. Here, if you turn down the intensity of surrounding points and assign higher intensity to something like a rail, the system can automatically trace that object. Figure 26 Figure 27 22
Figure 28 OTHER LINE AND SHAPE EXTRACTION This technology also works for 3D structures. In this instance, a template is created to tell the product just to follow along those points and a 3D model of this Jersey barrier. When doing this you can look at section views in the point cloud, see the barriers and create a template of that barrier and store it within the library. 23
Figure 29 Once the template has been created, you can compare that template to scan lines in your point cloud. When you do this, you might realize you need to make slight adjustments to the barrier. And because each these key points are interactive, you can move them to match the point cloud data. That way is you get to a point where the template changes drastically, you can insert or delete points and change the 3D perspective of the barrier. 24
Figure 30 The same thing can be done with objects like bridge decks. Here, a point cloud of the bridge deck has been scanned on top and underneath the bridge. A 3D template of the bridge deck can then be created to make the 3D model. Making adjustments in intervals allows you to create an accurate 3D model of the existing bridge deck. 25
Figure 31 BRINGING IT ALL TOGETHER Now that the 3D structures have been created, it s time to bring everything together. Using a 3D model and a point cloud, a 3D model has been created within Descartes, InRoads and MicroStation. The end result is a rich, 3D model of the project that is using all your point cloud data and gives you a solid model of what you re working with within that environment. 26
Figure 32 Figure 33 27
You can also create a model along the track and then include the 3D model, along with your point cloud data to have two different end products. In other words, part of your deliverable is a model of the actual rail you created, and the other part is your point cloud. Within MicroStation and InRoads you can even drive the rail track and see what your model looks like. Figure 34 CONCLUSION With 3D imaging and data capture, there are infinite possibilities to create rich, accurate end products that both improve efficiency and minimize safety hazards from putting workers onsite. Merging huge point clouds together and filtering to extract only specific structures or features allows users to access real-world information from a project site, without having to plan ahead to collect data only for one section. Nowhere is this more useful than in the transportation industry. From collecting and 28
documenting rail lines, to virtually driving through a project site, the various software tools available today allow for better collaboration possibilities. And with the ability to stream information anywhere with tools like ProjectWise, teams across the world can share in the building and development of rich, 3D models that offer insight into real-world data. CONTRIBUTORS Ron Gant is the director of marketing for Roads Bentley. He is a registered professional engineer in Texas who worked for 13 years in the electrical utility industry as a design construction engineer on major fossil and nuclear power projects. In 1990, Ron joined Intergraph Corporation. He moved to Bentley in 2000, when Bentley acquired Intergraph s civil and plotting products. For more than 25 years, Ron has managed technical marketing or sales organizations for Bentley civil products. Ray Filipiak is the global pre-sales technical manager at Bentley. He worked in the civil industry for 25 years at the Illinois Department of Transportation. Ray moved to the Atlanta, Georgia area in 1995 to work for C&G Survey Software. He joined Bentley Systems as an Applications Engineer in 1997. He has been involved with the Bentley civil and platform products for the past 25 years, either as a user or in technical sales & support. Ray has been with Bentley for the past 18 years. * Note: This whitepaper is the product of the transcript of a live webinar on www.sparpointgroup.com. While the speakers are cited here as contributors, this whitepaper was not written by the contributors or speakers who appeared in the presentation, nor is it endorsed by the contributors or speakers, or any company, organization or entity they represent. For more information on how this whitepaper was produced, send inquiries via email to info@sparpointgroup.com. 29