Lukas Fraser, Program Manager for Unmanned Systems at NV5, talks about the rise of drone-based LiDAR. He explains best use cases, how data accuracy is guaranteed when capturing drone-based LiDAR, why LiDAR and high-resolution imagery are complementary technologies and new market opportunities for topo bathymetric LiDAR.
LiDAR Technology and Topographic Mapping-based LiDAR
LiDAR stands for light detection and ranging and is a remote sensing method that is more and more being used in combination with drones. LiDAR from drones uses scanners while flying, shooting out pulses of light, travelling to the ground, penetrating through vegetation, getting down to the ground and returning to the scanner.
Using the speed of light, measuring the two-way travel time from the beam of light from the scanner, one can measure the distance to the surface of the earth and build a 3D point cloud. Based on the intensity of the return signal, you can tell which surface type the return came from: this does roughly correspond with what one would see in a picture, so that it’s possible to distinguish real-life objects in LiDAR data.
Collecting LiDAR data of an area for topographic mapping results in 200-300 points per square meter. But because a lot of geospatial software is not able to manage with so many points, there are processes to thin that data down so that clients can interact with it, explains Fraser: “First, the entire data set is classified in an automated way, after which different algorithms can be applied to isolate one point every square meter to reduce the data quantity, for example by smoothing the data based on elevation differences.”
Guaranteeing Data Accuracy
Data accuracy can be broken down into absolute accuracy, relative accuracy, and density, says Fraser.
LiDAR utilizes direct georeferencing through a GNSS and Inertial Measurement Unit (IMU), as opposed to photogrammetry where multiple control points are used to guarantee data accuracy. The GNSS is used to calculate the three degrees of freedom, and along with the longitude, latitude, and altitude, an IMU (inertial measurement unit) calculates the role, pitch, and yaw of the sensor.
While drone-based LiDAR theoretically only needs a single ground control point as all points are positioned off the coordinates of the base station, this does not guarantee absolute accuracy. To solve this problem, ground surveyors use truthing shots (or check shots) on hard, unvegetated surfaces that are independent from the control points. Generally, the offset of all check shots to the elevation of data is taken and used for an accuracy calculation.
This term refers to UAV systems with a lower grade IMU where it can be difficult to tie the data together if it’s not calibrated properly and experiencing IMU drift. “We take every flight line and compare it to the flight lines directly beside it, so we have a difference between every lidar flight line, and then we generate a raster with all of those differences so we can see them and then build some statistics off of that”, explains Fraser.
Point density refers to the measurements per area at which the surface of the earth is sampled.
Why collect LiDAR from drones?
According to Fraser, the main reasons for collecting lidar from drones are time, money, safety, and the level of detail that is required. In a lot of situations, collecting LiDAR from drones is cheaper than mobilizing a plane or helicopter and get those to the location to collect the data.
“A drone can be driven by car to the location, which is more efficient than using a plane or helicopter. With regards to the level of detail, drones enable to capture features that you can’t see because planes and helicopters are flying too high. Also, drones are a lot safer than having someone on the ground collecting data in a busy street in the middle of the road. At the end of the day, traditional methods for collecting data on the ground are slower than using drones if you’re covering a large area”, says Fraser.
Drone lidar technology has been around for the last ten years and really exploded in the last five or six years. Regarding price, Fraser notes a big range in the current market. “Starting with $20,000 and $30,000, you will get a laser scanner, IMU, GNSS system and the drone itself. High-end systems go up to a couple of hundred thousand dollars. To be able to do LiDAR from drones, you need a modular system with different components, even though they come with their own components”, says Fraser.
“While most drones have GNSS systems and IMUs on them for their internal navigation, those aren’t good enough to georeferenced our lidar. NV5 has a Riegl scanner that is used with an Applanix IMU. They work together and have a really good integration, but it’s a high-grade IMU that’s getting 200 measurements per second to be able to pick u pall of the fine movements of the lidar scanner”.
Discussing use cases for drone-based lidar, Fraser distinguishes two main markets: utility and engineering. “Utility companies are always replacing poles, building new ones, and looking at their existing infrastructure. We use drone-based LiDAR and weather data to calculate how different power lines and poles are going to react under a variety of weather conditions and when a new pole is added to the line.” Another use case is fire prevention, where LiDAR data is analyzed to determine which trees are at risk on falling on power lines and where vegetation needs to be cleared away.
For the engineering space, the main use case is topographic mapping. “With the drones, we’re able to get a very high level of detail. With regards to the LiDAR data, we are getting close to what would be traditionally collected with mobile LiDAR from a vehicle. We use that data to get 3D brake lines on curbs, retaining walls and inside people’s backyards, which is not possible with collecting LiDAR on the ground.” Drones enable to capture large swaths, which is used to do detailed mapping for road improvements, another NV5 use case.
Fraser sees the LiDAR and aerial imaging technologies as complimentary instead of competitive. “I think it’s a nice combination to be able to do both independently and compare each other to make sure we’re not having any issues. Our imagery is used for all 2D portions of our engineering drawings, the lidar for the surface model, and then we bring them together at the end.” While he admits that LiDAR is competitive with traditional stereophotogrammetry as it was used for break line generation (creating 3D polylines to represent features such as curbs, retaining walls and bridges), NV5 has just switched over to doing everything 3D related with LiDAR because being able to see under the vegetation and to generate so many points on the ground has been a huge benefit for the company, says Fraser.
The Democratization of LiDAR
While LiDAR is currently being democratized by Apple and enables anyone to with an iPhone to do a quick scan, that doesn’t mean that anyone can produce a survey-grade deliverable, says Fraser: “Especially when you’re working for utility companies, where you’re dealing with power lines and forest fire management, you better know what you’re doing because there could be real world consequences in case you make mistakes or overpromise.”
Another aspect of the democratization of lidar is the production of public lidar datasets. One example is the national 3D elevation program in the US where the goal is collect LiDAR data for the whole country and recollect at an interval. With over 600 business cases, it’s clear that there’s a lot of value in such a data set. It also makes people more careful about what they’re putting out into the world in terms of the accuracy metadata, which states when it was collected, published, its accuracy and the number of control points used, adds Fraser.
The future of topo bathymetric LiDAR
In the next five to ten years, Fraser expects drone-based LiDAR systems to become smaller, lighter, and cheaper. He singles out topo bathymetric lidar as an area that will drive joint innovation. “While there are a couple of companies that have made these sensors, we’re not seeing a widespread use of them. In the next five to ten years, we’ll get some systems that will perform well on UAVs that can get both topographic and sub-water surface data”.
The application for topo bathymetric lidar is the nearshore, which will fill in the gap between topographic data and sonar from vessels. Right now, governments are the biggest purchasers of shoreline information, but Fraser sees potential for other applications as well, such as flood mapping.
“Because it can be very difficult and time-consuming for sonar vessels to collect data really close to the shore, it’s more efficient to use LiDAR because of its large ranges. Depending on the water clarity, you can get anywhere from 10 cm to 30-40m. The difference between topographic lidar and bathymetric data is that for topographic lidar, you only need to watch for cloud coverage, while for bathymetric lidar, water turbidity is an important factor to be dealt with, for example by lowering a circular disk into the water and see how far down it goes, to using a live turbidity monitoring buoy that spits out real-time measurements to your cell phone”.