What are point clouds?
Point clouds refer to individual points, generated millions of times, that are used for measurement in 3D space. These points form the first layer of data obtained by a 3D scan, subsequently filtered and analyzed.
Each point is linked to its coordinates using all three axes (X, Y and Z axes). Vector data, which provide information about the orientation of the data points, are also used here.
In addition, each data point can also contain information about the color, which is stored in RGB format. Sometimes the luminance or intensity level is also stored in the data points. This level helps to uniquely determine the brightness of each data point.
Point clouds by laser scanning and photogrammetry
Point clouds are generated with the help of laser scanning or by using photogrammetry.
In laser scanning, the scanner emits light pulses. It then measures the time it takes for these pulses to return to their point of emission after hitting an object or a wall. Depending on the time required, the positions of the respective data points are calculated. Based on these calculations, a point cloud is finally created.
In photogrammetry, on the other hand, photos of the object must be taken from all angles and uploaded to the appropriate software. In order to generate a point cloud from the objects, points are plotted on the photographed objects in three-dimensional space.
One of the main disadvantages of photogrammetry is that it does not take into account scaling. This is because only images (i.e. digital models of the objects) are used and the computer cannot determine how big or small the object is in reality. This method therefore requires the use of software such as Netfabb.
Since 3D scanners are expensive, using a 3D scanner is more costly than using photogrammetry. Provided that the camera quality is adequate, a smartphone can also be used for photogrammetry.
f the camera quality is good enough and the photogrammetry is done properly using the smartphone, a deviation of only 0.2 mm can be achieved compared to a 3D scanner.
Software needed to create a 3D model from point cloud data
In order to generate point clouds, numerous software solutions are offered. Freely available programs in this context are Cloud compare, Meshroom (for converting the photos into a 3D model), MeshLab (for 3D scanning and cleaning as well as aligning different 3D scans), Netfabb (for scaling), GOM inspect (for comparing scans), etc.
What happens after point clouds are recorded?
Since point clouds represent the simplest and coarsest form of 3D models (in the sense of raw data), they are filtered, analyzed and subsequently converted into a polygon mesh using various algorithms. In simple terms, the individual data points are connected in the form of triangles, creating a kind of polygon mesh. This happens because most three-dimensional software programs work with polygon structures.
However, polygon meshes are used only to reconstruct the surfaces of the object and not its volume. The mesh includes a large number of polygons and elements such as faces, vertices and edges. Here, mathematical models such as NURBS are used to create curves and surfaces of the objects. In this context, software tools such as Autodesk can be used for simplification.
Steps to create a 3D model using point clouds
- Creation of point clouds of the object or building using laser technology or photogrammetry.
- Filtering of point clouds.
- Segmentation of individual points from the point clouds into different planar (plane) regions.
- Surface reconstruction for windows and doors.
During segmentation, the points from the point clouds are converted into planar regions using special algorithms. These regions are segmented according to their properties. Points with the same properties are placed in the same region. Planar fit and roughness of the surface are two general criteria that the algorithm looks for when segmenting points into regions.
After segmentation, four different classes are created. The first three classes give us the boundaries and outlines of the building. These classes consist of floors, walls and ceilings. The fourth and last class contains obstacles. This class can even contain columns, since they can appear as obstacles when planning navigation routes and are classified as such.
The next step is to reconstruct windows and doors. This is where the GHT method comes into play. This method uses binary images instead of color-related information from the point clouds. Once these planar images are converted to a grid, the points are projected onto the wall. A rectangular matrix is then created in which the points are coded with the values 0 and 1. These values are given to the points depending on whether they fall within the rectangular matrix or not.
Point clouds are not only used for 3D modeling. They can also be used for indoor navigation. But what exactly is indoor navigation?
What is an indoor navigation?
Do you remember? How many times have you missed your train or bus even after arriving at the station, just because you couldn’t find the right platform in time? Wouldn’t life have been so much easier if you had known the right direction at the time to reach your destination in the shortest possible time?
Indoor navigation is a technology that helps locate and navigate people or objects when they are indoors, where GPS is not sufficient. This significantly reduces the time and effort required. It can also determine which floor the user or object is on.
Have you ever noticed that GPS stops working as soon as you enter the interior of a building? The simple reason is that GPS works on the basis of the signals emitted by the satellites, which cannot reach the receivers due to the surrounding walls and ceilings. In contrast, indoor navigation technology uses different technologies depending on the specific requirement and application structure.
So, indoor navigation systems make it possible to find the right platform in a train station, the desired store in a shopping mall, the right hospital/shop space, or even any needed object in a warehouse. From positioning to tracking and navigation of people or objects, there are many potential use cases and real-life case studies.
Modeling and navigation in indoor spaces using point clouds
Normally, information for indoor navigation and positioning is obtained from 2D drawings and layouts. In this case, the actual state of doors, windows or obstacles is not taken into account. This is exactly where point cloud technology comes in. It helps to create rich 3D models of the room, also taking into account obstacles such as furniture and other objects when planning routes.
Editing the point cloud for navigation
In the case of navigation using point clouds, processing begins with the detection of obstacles, which is followed by route planning. This is done by creating a buffer area. To get a better idea of this, think of the buffer as a rectangular box moving from point A to point B. The algorithm moves the buffer box further and further towards B. Points of the fourth class, i.e., the class of obstacles, are observed to determine whether any obstacles enter the buffer space.
If it is determined that there are points in the buffer space, the next step is to check whether they can really be identified as obstacles. Size and accumulation are the two factors that are checked in such a case to determine if the points are indeed obstacles. Depending on the density and the number of points that enter the buffer space, it is decided whether they are obstacles or not….
If the buffer recognizes an obstacle as such, it changes the route and finds a new route to reach point B.
The next step is path planning or finding and route correction. Logical and navigable networks are used for path planning. Calculations of Voronoi cells are performed to find the path for the new point. This is done after it is inserted into the network. Also, new point nodes are created when new obstacle points are found that were not included in the path calculation.
If buffers of the obstacle nodes overlap, a new buffer area is created. This helps navigation and helps to find a new route without obstacles.
Advantages of point cloud technology
- Time savings in surveying. With good laser scanning technology, companies can save time and manpower.
- An advantage is also the accuracy achieved. Data obtained through the use of point clouds is very precise and detailed. A level of detail that cannot be guaranteed with conventional surveying methods.
- Design analysis for renewals can help companies create better quality models.
- The overall cost of the project is also reduced by point cloud technology.
Alternative technologies for indoor navigation
Now that we have learned about the four basic principles commonly used in indoor navigation systems, let’s find out how some of these technologies work.
Tags and Beacon
These reader-based tags continuously send their signals to the readers. The data is then forwarded to the main system where, after processing the data, the location can be generated.
Beacons function like lighthouses. They are Bluetooth devices with low energy consumption (Bluetooth Low Energy, or BLE for short). They emit radio signals that are received by smart mobile devices that have the required app. The number of beacons required depends on the size of the room, i.e. the larger the room, the more beacons are needed. Depending on the highest signal strength, the closest BLE indicates the user’s location.
Beacons work on a reference point based system with an accuracy of 1-3 meters. Intelligent location-based identifiers integrated into the system help to find their location by means of the position of the corresponding reference points.
In terms of price, beacons perform better than tags. This is because tags require a higher number of connected readers, which in turn increases the overall cost of the system. In contrast, beacons are less expensive due to the use of reference points, as they do not require as many readers as tags.
Tags act as WIFI transmitters that send Wi-Fi packets to different Wi-Fi access points. The duration of sending and/or receiving the signals and the strength of the data received are used to determine the location of the user or object to which the tag belongs. Compared to other options, these tags are quite expensive.
The accuracy of a Wi-Fi-based system ranges from 5 to 15 meters. Another advantage of this technology is that it can also be used to locate the exact floor where the object in question is located.
A disadvantage, on the other hand, is that the system is not available for iOS users.
Ultra Wideband Technology (UWB)
Ultra-wideband is a short-range wireless communication technology that transmits billions of short pulses (2 nanoseconds each) over a wide frequency range of the spectrum (>500MHz). The reader then converts the intercepted pulses into data by comparing already known pulse sequences by listening to the pulse noise. In this way, highly accurate data can be obtained in terms of both direction and position.
With UWB, real-time location is also possible. But in terms of price, it is the most expensive method because it requires at least three expensive readers to compensate for the short-range tags.
Acoustic systems work like ultra-wideband systems, with the main difference that they use sound waves instead of radio waves. The tags emit ultrasonic waves, which is inaudible to our ears. After the emitted waves are captured and processed by the receivers, the location is indicated.
Depending on the length of time it takes for the signal to return, the presence of a multiple propagation path (multipath) can be determined.
This system uses infrared light pulses emitted by the IR tags. After these tags emit the pulses, they are picked up and read by the IR receivers. Since light is used instead of radio waves, the possibility of an erroneous readout by another receiver receiving the signal through the wall is eliminated. This is because light, unlike radio waves, cannot penetrate walls. This consequently increases the accuracy of the system at the room level.
Both infrared and acoustic systems are less expensive to install in new buildings. However, when retrofitted, the costs increase.
Advantages of indoor positioning systems
Aside from basic positioning, tracking and navigation tasks, data from indoor navigation systems can be used in other ways to support and advance businesses.
Indoor navigation using point clouds could help blind people or wheelchair users navigate or get out of the building as quickly as possible during an emergency.
Customers benefit not only from easy navigation and the ability to more easily locate a specific store in a shopping center, the room in a hospital, or even the right platform at a station. Thanks to the positioning system, customers can also find products in supermarkets or other stores more quickly.
With the help of the installed app, users can receive notifications about current offers and sales while walking past stores, for example, that they would otherwise have missed.
However, there are also further advantages for the business sector. These are primarily in the tracking of customers. Both the number of visitors and the amount of time they spend in a store or a particular area of the store can be determined in this way. This can provide companies with valuable insights into customer behavior. The insights are valuable in that products can be targeted, i.e. optimized and developed based on the customer’s benefit, on the basis of the knowledge gained. This can help companies increase sales and generate higher revenues. A particular incentive for companies is also to provide their customers with personalized offers that motivate them to buy in the store.
Knowing various technologies that can be used in the interior positioning system is good, but unfortunately still not enough. While you have certainly acquired a certain amount of knowledge about a particular subject area, as is so often the case, this does not automatically give you its concrete benefits. In other words, knowledge is all well and good, but it must also be clear how and where these technologies can be used profitably to create value.
The software used to process this data is also crucial. Better software leads to more precise results, which at the same time can be achieved in a shorter time.
For example, infrared systems are a poor choice when it comes to using indoor positioning systems in open spaces such as warehouses and production areas. Here, the use of radio waves would be more suitable. This is because radio waves can avoid mismeasurements that would exist with infrared systems due to the difficulty of measuring signal strength. IR systems can best be used in enclosed spaces such as hospital rooms, etc.
In order to select the right indoor navigation system for a specific application, it is advisable to first be aware of factors such as required accuracy, budget, presence of enclosed/open spaces, complexity of the building, and so on. But even that is not all that needs to be considered here!
loehn.digital will be happy to help you and offers you various services, including the provision of an indoor navigation system with point clouds. If you are thinking of using or need such a system for your projects or buildings, do not hesitate to contact loehn.digital.