LIDAR MAPPING
INTRODUCTION
Light Detection and Ranging (LiDAR) is a measurement technique that uses light emitted from a sensor to measure the range to a target object. In very simple terms the sensor emits a light pulse and then measures the time taken to receive the reflected pulse in order to estimate the range of the target object, given the constant of the speed of light. Modern LiDAR sensors have multiple lasers or channels, 8 to 128, that are able to produce up to 2.2 million points per second. The LiDAR unit scans from side to side, with some having a full 360 degree horizontal Field of View (FOV), which creates a very dense point cloud that represents the surrounding area.
LiDAR sensors are able to achieve range accuracy of 0.5 to 10mm relative to the sensor and a mapping accuracy of up to 1cm horizontal (x, y) and 2cm vertical (z). This makes them particularly useful as a remote sensing tool for mobile mapping. Additionally, LiDAR sensors are able to collect multiple returns from a single light pulse. This is because as the light pulses travel from the sensor they may encounter several objects that will reflect the pulse such as leaves and branches of a tree canopy. LiDAR sensors are able to record this information to provide a detailed understanding of not only the tree canopy but also the underlying structure. Using these multiple returns LiDAR mapping is able to produce both a:
- Digital Elevation Model (DEM), which is a topology model of the earth’s surface, with natural (trees and other vegetation) and built (buildings, bridges, communication and power lines) objects removed, and a
- Digital Surface Model (DSM), which includes all natural and built features, e.g. communications lines, bridges, vegetation etc.
Figure 1: DSM Point cloud
Georeferencing LiDAR data
Georeferencing is the process of applying a coordinate system to the point cloud so that it can be accurately located on a map. In order to georeference the point cloud the LiDAR sensor’s orientation and position, or Exterior Orientation Parameters (EOP), need to be known in order to create a planimetrically correct scan that can be used for Geographic Information System (GIS) analysis.
Unlike photogrammetry where Ground Control Points (GCPs) can be used to georeference mapped data using Aerial Triangulation. Ensuring that LiDAR data is accurately georeferenced requires the use of Direct Georeferencing, using an accurate GNSS receiver and an inertial measurement unit (IMU) measure the pose (orientation and positing) of the LiDAR sensor.
To accurately georeference LiDAR data, operators need to take particular care when configuring the mapping system. Any misalignments or offsets incorrectly configured will directly impact the accuracy of the point cloud. There is little that post processing can do to resolve errors and will most likely result in having to remap the environment/scene. When configuring the mapping solutions it is important to ensure that the following are correctly configured:
- Boresighting - this is the process of aligning the LiDAR sensor axes with the IMU axes.
- GNSS Antenna Offset - because the GNSS Receiver calculates the geographic coordinates at the antenna phase center we need to map this to the IMU origin. The GNSS lever arm needs to be configured in the IMU reference frame for x, y, z offset
- Reference Frame Rotation - The IMU coordinate frame may not align with the vehicle frame. It is therefore important that the IMU frame be mapped to the vehicle frame in order to ensure that the georeference data is correctly applied
Finally we need to implement synchronization to ensure that each range measurement is correlated with the correct pose data. LiDAR sensors typically have facility to accept a time synchronisation pulse, such as a Pulse Per Second (1PPS), from the GNSS receiver and NMEA timing data.
Figure 2: MMS Reference Frames and Offset calibration
LIDAR MAPPING APPLICATION NOTE
To learn more about the advantages of LiDAR over other surveying methods, the practical implementation of using LiDAR for surveying and the impact of GNSS/INS precision on LiDAR mapping accuracy please download our application note.
Sensor Selection for Direct Georeferencing LiDAR Data
Instead of using a stand alone GNSS and IMU, a more suitable solution for estimating EOPs of the LiDAR sensor is to use a GNSS/INS. The advantages of using a GNSS/INS include a more accurate attitude solution, ability to carry position and velocity updates through short GNSS outages and several other advantages. To learn more about how a GNSS/INS operates please read Section 1.7: GNSS-AIDED INERTIAL NAVIGATION SYSTEM (GNSS/INS) in the VectorNav Library.
Using a GNSS/INS allows for all points in the point cloud to be geo-referenced to a fixed global reference frame. The following equation can be used to solve the absolute position of a point on the ground:
𝒑g = 𝒑a + [C]⍴
Where,
- 𝒑a is the estimated position vector of the LiDAR sensor,
- [C] is the estimated direction cosine matrix mapping from the LiDAR to the inertial coordinate frame and,
- ⍴ is the LiDAR point offset vector.
The estimated LiDAR position vector (𝒑a) comes directly from the GNSS/INS system and is subject to both GNSS position errors and timing errors between the systems. The coordinate frame transform between the LiDAR reference frame and an inertial coordinate frame (eg. NED) comes from the attitude measured by the GNSS/INS. Errors in [C] come primarily from the sensor attitude misalignments, GNSS/INS attitude uncertainty, and timing errors between sensors.
Attitude Errors
Any angular uncertainty projects to the positional error based on the distance from the point to the LiDAR sensor. Therefore to achieve the required mapping accuracy, the attitude (Pitch/Roll) accuracy required is a function of the mapping height. The higher the altitude the greater level of attitude accuracy required.
Angular Accuracy = Arctan (Accuracy/Range)
Figure 3: Attitude accuracy requirement vs range for 4cm point cloud accuracy
Positioning Errors
Positioning errors come from the uncertainty in position of the GNSS/INS sensor. This position uncertainty is represented in the North, East, Down (NED) coordinate frame. The position uncertainty is mostly determined by the accuracy of the GNSS solution alone.
In order to achieve survey-grade results it is necessary to use GNSS Receivers capable of either Real-Time Kinematic (RTK) or Post-Processed Kinematic (PPK) corrections to determine image coordinates to less than 1cm. As there is little need for real-time mapping, most LiDAR mapping operators opt for PPK which has advantages of simplified operation and provides more accurate positioning data compared to RTK by utilizing precise satellite ephemeris data and forward/backward smoothing techniques.
LIDAR APPLICATION NOTE
To learn more about sensor selection, implementation and determining your error budget for LIDAR applications take a look at our in-depth application note.
Conclusion
As the design projects increase their reliance on modeling and verification, the demand for LiDAR scanning technology continues to grow. LiDAR mapping solutions rely on accurate range measurements coupled with accurate pose information. For LiDAR mapping to produce accurate georeferenced point cloud data it is vitally important that the GNSS/INS solution not only have suitably accurate position and attitude data, but that the sensor be aligned and boresited correctly.
A suitably configured Mobile Mapping System is able to produce data that rivals traditional techniques with additional advantages of speed and safety.
VECTORNAV SUPPORT LIBRARY
Everything you need to know about inertial navigation.
