One of the drone world’s most exciting new technologies is LiDAR, where light is used to provide users with accurate geospatial data in a wide variety of new environments. In this article, we delve into how LiDAR works, explore its plus points over photogrammetry, and discuss the features and benefits of the Atmos LiDAR in particular.

‍

LiDAR fundamentals

Difference with traditional RGB photogrammetry

Photogrammetry uses off-the-shelf photographic cameras to capture thousands of images of an area. These cameras work under the following principle; The sun illuminates an object, which partially absorbs or transmits the light, and reflects the rest. The reflected light is captured by the sensor of the camera. This sensor can see colour & brightness, but it can not sense depth.

Because the camera can not sense depth, reconstruction from hundreds, if not thousands of overlapping photos are needed to obtain 3D coordinates. In a high-quality model, each feature is usually included in around 8 overlapping images.

Once these images are attained, photogrammetry in software packages like Pix4D or Metashape, etc is utilised. In this, bundle adjustment is used to obtain the accurate orientation of the images in six degrees of freedom. Images are then compared to detect and match features between images and their position is surmised using triangulation. It’s important to note that this is a statistical calculation of positions rather than a direct measurement.

‍

How LiDAR Works

A LiDAR sensor works a bit differently, as it uses active sensing to directly measure the distance to objects using laser pulses, rather than the statistical method used by photogrammetry.

LiDAR itself acts as the light source (instead of the sun), which reflects off the object in question and is received by the LiDAR sensor. The LiDAR’s laser beam is also slightly divergent. This allows the beam to ‘shine through’ objects like tree branches. The detector can detect multiple returns allowing multiple surfaces to be surveyed.

‍

The LiDAR sensor measures the range of the laser pulse, and the scan angle of the rotating head is accurately known. Because a LiDAR drone is flying through the air at a high speed, the main qualitative specs revolve around how accurately the position and orientation of the LiDAR is known.

The part of the LiDAR that measures the position and orientation is known as the INS. The INS consists of an IMU which measures orientation, and a GNSS which measures position to a high degree of accuracy. This information can be used to directly ascertain the 3D location of a point on earth.

The output product from the LiDAR is a point cloud, the most common use for these being to view elevation, but the intensity of the received pulse is also measured. Visualising the point cloud with intensity on a pseudocolour/false-colour scale helps to distinguish different objects on the ground which have the same elevation.

These outputs can then be viewed in Pix4D, or other products like Displaz, ArcGIS or Fugro, and then used for vectorization, DTM/DSM development, biomass estimation, volumes, profiles, progress monitoring, and many more!

‍

Advantages over Photogrammetry

Photogrammetry has a lot of strengths, namely in cost, colorization, and accuracy. However can LiDAR provide a tremendous advantage over photogrammetry when it comes to;

^{1. Thin Structures}
Photogrammetry struggles with powerlines and complex on-site machinery, lacking enough reference points. LiDAR’s active sensing method has no issue with these.

^{2. Foliage-heavy Spots}
‍Farms and forests with thick trees don't interfere with LiDAR. It captures data from both the canopy and terrain beneath, unlike photogrammetry which only captures a DSM of the forest canopy.

^{3. Uniform Surfaces}
‍On smooth surfaces like roads or beaches, photogrammetry falls short due to a lack of unique points. LiDAR, with its active measurements, doesn't have any issues with this.

^{4. Unique Light Conditions}
‍LiDAR surveys do not require external light conditions to generate data, so it is possible to survey in low-light or even nighttime conditions.

‍

Atmos LiDAR Features and Benefits 

As mentioned above, because a LiDAR drone is flying through the air, the accuracy of the final model depends heavily on how accurately the position or orientation of the LiDAR is known, along with FOV, range accuracy, etc. Therefore the main question that each customer should ask is, how accurate is the LiDARs INS?

The Atmos LiDAR

Atmos’ LiDAR features a combined Hesai LiDAR scanner and Inertial Labs / Novatel INS. This includes two GNSS antennas for positioning and heading estimation.

This combination of a high-end INS and a double heading antenna is really important. A high end INS results in highly impressive accuracy figures, namely a +/- 1cm [0.4in] ranging accuracy, 0.5cm [0.2in] precision, a pitch/roll accuracy of <0.01°, and max altitude of 150m [492ft] AGL. Double heading antennae result in a yaw accuracy of 0.05°, and contribute to the above positional accuracy as well.  

Atmos LiDAR Benefits

What does this mean for users? 

Firstly the data has an exceptionally low noise level, meaning that small features such as power lines are easily detectable without the need for extreme filtering and altering of the source data. 

Secondly this means that no strip alignment is required, which often causes inaccuracies, and adds at least 1 hour to the workflow each time the user performs a flight. This also benefits the user by resulting in a higher point cloud accuracy, and a lower required overlap meaning users can cover more area per flight.

‍

A major factor in choosing a LiDAR system is also the trade-off between point density & area coverage. Having an effective trade-off between point density and area coverage will determine ROI of the system as a whole. The Atmos LiDAR is able to achieve 52 pts/m² [4.9pts/ft²] at 480Ha [1190 Acres].

Conclusion

LiDAR’s advantages over photogrammetry, particularly in handling complex & vegetated environments, coupled with impressive accuracy figures, present a compelling case for its adoption. This technology not only overcomes the limitations of traditional methods but also opens up new possibilities for applications in Forestry, Mining, Construction, and more. Within this world, the Atmos LiDAR, with its remarkable accuracy figures and efficient workflow, emerges as a powerful tool for professionals seeking detailed and noise-free point clouds.

‍

‍

FAQs

_{What are Returns?}

A laser beam is not a singular item, it contains pieces of light, some of which reflects off the ground, and others reflecting off other objects in between like grass, leaves, and tree branches. When the LiDAR sensor detects that some pieces of light have been reflected, it is recorded as a return. If you only have one return, you only get the first reflection which has come from each laser beam, meaning that you are only receiving data on the closest object to the drone, much like photogrammetry. Therefore, capturing more than one return allows you to get more information on sub-surface objects, like the ground beneath a forest canopy. The Atmos LiDAR is capable of 3 returns, there are others which offer up to 5 returns, but these additional 4^th and 5^th returns are often found to not offer much insight and instead contribute to increased noise which needs to be filtered out in post-processing, adding more requirements to the workflow.

_{Is Atmos’ LiDAR colorized?}  

Each point in LiDAR point cloud consists of a position in x, y, & z, and an intensity level. Some users like to also assign a colour to each point in the point cloud, as you can then distinguish between certain elements which aren’t distinguishable by their intensity plot. Whilst the Marlyn doesn’t carry a camera embedded into the LiDAR sensor, it is possible with the Marlyn to capture first the photogrammetry data, then swap out the payload for the LiDAR and capture the LiDAR. Following this you would use Cloud Compare or ArcGIS to colorize it. It is also possible to colorize the point cloud using commercially available satellite data.  

_{Is it required to use GCP’s?}

It is indeed possible to inform your data using control points using two possible methods - intensity or elevation. For intensity, natural features with a high intensity contrast (@905nm wavelength!) can be used. For elevation, objects with sharp edges and some elevation (>10cm) can be used, commercially available check points are available for this.

‍