Handheld Mobile LiDAR Scanning for Urban Power Line Inspection

An Initial Investigation

We recently released an article, Power Line Corridor Mapping – Methodologies to Acquisition and Data Processing, which presented our well-established and field-tested solutions for power line inspection using UAV LiDAR systems and LiDAR360 post-processing software. The article attracted huge interest and brought in many inquiries. Excited as we are, the reaction was not a surprise. The industry, like any other, is always on the lookout for new technologies and new solutions to improve productivity and efficiency. LiDAR, thanks to its superior capabilities including speed, accuracy, processing automation, and ability to easily integrate into existing engineering and GIS workflows, has become an important addition to the arsenal of many ESCOs (Energy Service Companies).

The previous article showcased GVI solutions using UAV LiDAR for high-voltage transmission line survey. We now turn to another question, what about using LiDAR to inspect power distribution lines in urban neighborhoods? Due to regulations, flying aerial systems over urban neighborhood is either strictly prohibited or inefficient because of the VLOS (Visual Line Of Sight) rules. What about conducting the inspection from the ground using terrestrial lidar, specifically, handheld mobile lidar systems? To answer this question, we performed a quick test. This article details what we did and what we found out.

It must be emphasized, rather than presenting a solution, this article presents the findings from a test which investigated the potential of utilizing mobile LiDAR for power distribution line inspection in urban neighborhood. We will perform follow-on tests in the coming weeks and we welcome any questions, comments, and suggestions.

Test Site

We chose for this test a 2-block area between the intersections of Allston Way & Spaulding Ave (SW), and Addison St & Jefferson Ave (NE). These are typical city neighborhood settings. This area was selected for the simple reason of being close to our downtown Berkeley office.

Figure 1. Test Site Overview in Google Maps
Figure 1. Test Site Overview in Google Maps
Figure 2. Test Site Ground View Photo
Figure 2. Test Site Ground View Photo

LiDAR System

A 2nd generation single-laser LiBackpack 50 unit was used. The 50 is our current best-selling SLAM-based mobile laser scanning product. It’s unique design also allows the system to be used in handheld mode which offers extra flexibility. The LiBackpack 50 is powered by a VLP-16 laser sensor from Velodyne.

The sensor has a maximum range of 100 meter, with +/- 3 cm accuracy and 2.0° angular resolution. In single return mode, the VLP-16 can generate up to 300,000 pulses per second. A major advantage of our SLAM-based handheld mobile scanners is its ability to continuously generate real time 3D point cloud with trajectory, loop and timestamp information at hand. Processing is also in real-time, resulting in no registration, SLAM processing, etc.

backpack21
Figure 3. GVI LiBackpack 50

Data Acquisition

Following a double-loop pattern, the scan was completed in approximately 10 minutes, with the LiBackpack in handheld mode. The operator walked at a relatively slow pace.

Upon completion, the LiBackpack system generated 3 output files: scanning route (.xyz), raw data (.bag), and point cloud (.ply). All 3 files are immediately available for post-processing. Total output data size was about 2 GB.

Figure 4. Test Scan Double-Loop Route Pattern
Figure 4. Test Scan Double-Loop Route Pattern

Data Processing and Results

Because the LiBackpack system can generate standard point cloud immediately after the scan, no complicated raw conversion is required. The output point cloud is ready to be used. Note the unit used for this test does not have the GPS module, thus no absolute coordinate system is available. However, if needed, georeferencing can be easily done using ground control points in LiDAR360.

LiDAR360 is a comprehensive software solution that provides tools for efficiently visualizing, generating & manipulating LiDAR data. A free 30-Day trial is available to all first-time users here.

Figure 5. Point Cloud from the Test Scan Visualized in LiDAR360
Figure 5. Point Cloud from the Test Scan Visualized in LiDAR360
Figure 6. Point Cloud from the Test Scan Visualized in LiDAR360
Figure 6. Point Cloud from the Test Scan Visualized in LiDAR360

We performed a basic feature classification using Machine Learning Classification in LiDAR360, but only attempted to classify power lines and poles. For this quick test, we did not separate power lines from cable and phone lines.

Figure 7. Classified Point Cloud vs. Google Earth View
Figure 8. Classified Point Cloud
Figure 8. Classified Point Cloud

Findings and Discussions

Again, this was a quick test to investigate the potential of using mobile handheld LiDAR scanners for power distribution line inspection in urban neighborhood. It is possible to capture good power line data with a mobile handheld LiDAR solution like the GVI LiBackpack. Even with the VLP-16, as shown by this test, the results were surprisingly satisfactory.

With the advanced Machine Learning Classifier in LiDAR360, it is possible to automatically extract key features from the point cloud. Note that inspection and manual editing, like with most other automated classifications, are always needed to achieve the best results.

Figure 9. Point Cloud vs. Site Photo
Figure 9. Point Cloud vs. Site Photo
Figure 10. Classified Point Cloud vs. Site Photo
Figure 10. Classified Point Cloud vs. Site Photo
Figure 11. Classified Point Cloud vs. Google Earth View
Figure 11. Classified Point Cloud vs. Google Earth View

Once the key features are classified, the results can be easily incorporated into standard engineering or GIS workflow. With tools offered by the Power Line Module in LiDAR360, danger/hazardous points detection and reporting can also be done efficiently as we illustrated in the last article.

The Velodyne VLP-16 does have its limits. To capture good data, the sensor needs to be aimed directly at the distribution lines (direction of the laser pulses). And its effective range for capturing power lines is approximately < 30~40 meters.

With the understanding of these hardware limits, route planning becomes critical. Besides basic loop-closure, the operator also needs to make sure the power lines are scanned by direct-targeting. The handheld mode available with the LiBackpack 50, therefore, is essential. And the performance of the system’s SLAM algorithm must be solid and reliable. With the recent release of the LiBackpack D50, scanning directly under the power lines for long periods of time will be possible due to the unique angle of two VLP-16 sensors. The range of the D50 has also been increased to 70 meters, offering a good solution to higher lines and poles.

Figure 12. Example Showing Insufficient Data Points for Power Lines in the Test Scan Data
Figure 12. Example Showing Insufficient Data Points for Power Lines in the Test Scan Data
dual head