This is the second part of Francisco Corredera’s article. First part can be found here.
How a LSS works? LSS internal hardware, basically, is shooting a laser beam and measure the time the beam takes to return to the hardware. In this hardware there are at least optic electronic emission, angular positioning, processing data and laser beam reception systems. Hardware knows very well the speed of the laser beam, its characteristics, environmental situation, time to get it back and many others. With all these, it calculates the distance of the object that cause laser beam rebound.
This, along with the rest of the core hardware controlling relative geometry of rays, vertical and horizontal angles is able to allocate XYZ coordinates to recorded points, whose are 3D points in the nearby area around LSS.
Actually, LSS doesn’t shoot only a wave pulse, but hundreds of thousands per second, enough to obtain resolution required. Topic is much more complex than this. There are measure time scanners and others called “of triangulation” depending on how they survey the distances, and those first at the same time could be divided on flight time scanners or phasing comparison, everyone of them with advantages and disadvantages and focused on different requirements and targets. Everything known as LSS technologies. Obviously, an explanation would be a little of topic for this article.
Not every LSS works at the same resolutions and qualities. Neither have the same range on distances. Every LSS allows to select many different qualities and resolutions ranges depending on the case we need it. Resolution and quality are the most important parameters in LSS. Resolution means distance between two points, and depends also of the distance of the target related to the LSS position. For instance, in one survey, nearby elements will be represented by a cloud of points more dense than the farther objects. Quality means number of times LSS surveys every point. Bear in mind that there is no only one survey per dot, there are many of them trying to get a reliable average for the points. At the same time, LSS can be also classified by range. There are short range equipments, to survey caves or inner spaces, a medium range for distances up to 120 m and long range hardware to scan objects up to 350 m. There are even longer ranges, but we’ll not focused on them on this article.
Most of the modern LSS hardware are able to survey up to 1000000 points per second. Its an impressive number. And that’s not all, because they are able to get coordinates of all these points within precision <1cm. More than than, hardware is able to assign a “reflectance” value to every point so you can “see” different textures. Obtained cloud could be easily interpreted by a viewer. Some of the last LSS hardware also add cameras to assign a RGB value to every point and we are able to obtain a picture of that space.
After an scan process with a LSS, we should simply obtain a cloud of points used to identify surveyed elements. With that information, not a surface nor an element yet, only points in the space, we need to get to work to materialize it and finally build a 3D model.
In the following pictures, A is showing a single colour points cloud with a low resolution. It is clearly visible spaces between scans and distances between points.
In B, we have a sample of an LSS take where the points contain information about reflectance, so seems easy to distinguish different textures: it is far from a photographic image yet, but it’s far better than A image.
The next sample, C, LSS is equipped with a photographic internal camera which assigns an RGB colour to every point. Therefore the sample seems and feels like a picture. This greatly helps in the next process: modelling and delineation.
Besides LSS has or hasn’t that internal camera, this kind of work use to be done with a complementary photographic study of all surveyed space. Taken pictures are finally overlaid over the 3D structure, which allow to obtain a photo realistic model. You can imagine a sheet over the original structure to capture forms and pictures impressed on this sheet. This would work like this:
- Obtain a point cloud with LSS
- Build surface and structures based on surveyed points. a 3D modelling.
- We use taken pictures to “wallpaper” the space.
LSS portable and static. Methodology on iRacing videos
Let to talk about the field methodology used with LSS. They can be portable or static.
iRacing staff shown static devices on its videos. Those static LSS are on tripod and from those base survey and scan a 360º degrees horizontal space and from 270º to 300º degrees on the vertical. So, along the scanned track, LSS are being moved from an static base to another until to complete a whole survey of the space. A usual methodology (which is also what iRacing staff does) use to be allocate a series of targets, place strategically to be surveyed by the LSS and work as a geometric and georeference points. This helps to order them to shape a whole block of the scanned track.
These cloud points needs to be ordered in 3D so, it is important use targets to understand how to place all of them on a tridimensional space on the computer and get a comprehensible model. There are some software suites able to fit all them together automatically using GPS positioning and other parameters. But tracks are a little more difficult because of their open space nature and targets are the way to go. Therefore, with some kilometers long and a few meters on slope, work has to be done with all the necessary thoroughness. Otherwise we could find and impressive result with no likelihood with a real counterpart.
Following picture from iRacing development shows a whole track with all the cloud points of every survey already georeferenced. Expanded picture is used to understand those black circles. These are different stations where equipment was placed. Remember a LSS scan vertically only 270º to 300º degrees and it is impossible to scan its own position. That fact leaves a black lagoon on the ground without being scanned. Also you can see cloud is more dense near to these “lakes” as we already explained. As we push away from the station points this density decrease and elements are reflected with less resolution. This is how iRacing scanned tracks.
To build a model using this method in a 2 km length track is necessary place LSS at least 35 times taking 15-30 millions points in every station. With coordinates, RGB and reflectance values of every point is easily to build an understandable and measurable model with 600 to 1000 million points.
iRacing, in our opinion, has been using an excellent working and equipment method. Hardware, software, methodology and costs for this kind of projects. Consequently they have obtained a high quality model using a more traditional method with stations and targets with a constant resolutions and controlled precisions.
To be continued…