Underwater laser scanners have started to become a preferred technology for close range, high precision underwater measurement work.
The ability of these sensors to quickly obtain a complete 3D model of the environment has proven to be useful for sub sea pipeline inspection, inland pipe and tunnel inspection, offshore jacket inspection and archaeological and scientific applications. Using the digital 3D model, created by the laser scanner, engineers can accurately assess the status of an asset with more certainty and make better decisions. The traditional underwater measurement technology, sonar, is great for longer range measurements but does not provide the level of detail the laser scanner can obtain at close ranges. However, determining the optimal range to switch from one technology to the next can be challenging. This paper describes why laser scanners are range limited compared to sonar technology and presents some results from a test using a laser scanner at range.
Sonar & Laser Comparison
Underwater sonar and laser measurement systems operate based on significantly different principles. Where sonar uses a vibrating sound pulse, laser uses projected and scattered light. It is because of this fundamental difference between the technologies that sonar is better at long range and laser is better at high precision. Sonar is better at longer range because laser light is blocked by particles in the water. The sound wave propagates by vibrating adjacent particles irrespective of the particle material. Thus when there is silt in the water, these silt particles vibrate principally like the water particles to transmit the sound pulse. Underwater laser scanners generally use blue/green laser light as this wavelength of light has the best transmissivity through the water.
2G Robotics ULS systems use an internal laser diode to emit the light signal, the light is reflected off the target surface and collected using an offset sensor. Based on the known offset between the laser and the sensor and the angle of the reflection measured from the offset sensor, the distance to the target is calculated.
Effect of Range
Using a triangulation technique is ideal for underwater measurements however, the downside of this technique is that the performance of the system degrades with increasing distance.
Like all sensors there is some noise and error to the measurement that is obtained. For the ULS systems, this error is applied to the measurement of the reflected beam angle. Noise in the measurement angle results in noise to the point cloud. As demonstrated in Figure 2, at close ranges, the measurement error window (MEW) in the measured angle has less impact on the calculated distance than at far ranges.
As a way to overcome the error to the measurement at further ranges, increasing the distance between the laser and the offset sensor can reduce the noise in the point cloud data. As demonstrated in Figure 3, for the same measurement error window (MEW), you can see there is a reduction to the size of calculated distance error (CDE) when the offset sensor is spaced further from the laser.
Range Test Description
To experimentally demonstrate the impact of range on the data quality using the ULS-500 system, a test target was constructed and submerged into a test tank. The test target was a trash can as this provided both large and small features to measure. The trash can was fixed with ropes to a stool and weighted such that it was held firmly on the floor of the tank. The trash can and ULS-500 underwater laser scanner were separated by 3m, 6m and 9m in successive scans. By adjusting scanner parameters, approximately equal point density in all three scans was obtained. Ideal scanning conditions of clear water and no ambient light were present. The point clouds from all three scans are compared to demonstrate how reducing the range from the target results in decreased noise in the point cloud data. These results were captured without use of 2G Robotics filtering settings to demonstrate the impact of range. By implementing filtering, which require longer scan times, point cloud data noise similar to that at 3m can be achieved at the 9m range.