Congratulations to MANTIS and GSEN master student Sabin J. Pandey for successfully defending his dissertation, titled “Evaluating the Performance of Apple Pocket Lidar: Case Studies in Natural and Built Environments”.
As a GSEN student, Sabin worked under the advisement of Dr. Michael Starek. His research has focused on evaluating the capabilities and limitations of Apple pocket lidar (PL) sensor through case studies in both natural and built environments. Before his journey at MANTIS, Sabin received a B.E. in Geomatics Engineering from Purbanchal University in Nepal and worked as an academic for a brief period. Following graduation, he plans to relocate to Tampa, Florida, to join Element Engineering LLC and continue his career in the field of Geomatics. We wish him well in his future endeavors.
PRESENTATION ABSTRACT
Developments in light detection and ranging (lidar) systems have enabled integration of lidar sensors into consumer-grade Apple devices, opening new possibilities for low-cost, rapid, and accessible 3D data acquisition. This study evaluates the performance and practicality of the Apple pocket lidar (PL) sensor, using the iPad Pro with Pix4Dcatch and 3D Scanner apps, in both natural and built environments on the Texas A&M University–Corpus Christi (TAMUCC) campus. High-accuracy geomatics techniques, including real-time kinematic-global navigation satellite system (RTK-GNSS), terrestrial laser scanner (TLS), and total station (TS), were employed to acquire reference datasets. Uncrewed aircraft system (UAS)-based structure from motion (SfM) photogrammetry was utilized to explore data integration with PL-derived SfM/multi-view stereo (SfM-MVS) data. Lidar-only data from the 3D Scanner app produced smoother but less detailed surfaces, while photogrammetry-only scans offered richer texture but struggled in low-texture or homogenous areas, suggesting that a hybrid approach yields more complete data. Multiple scanning strategies (straight-line, loop, zigzag) were tested to examine data quality and algorithm behavior. Loop scans achieved the highest slope distance accuracy (2.7 cm root mean square error (RMSE)), with progressively decreasing errors in straight-line and loop scans, indicating the use of a simultaneous localization and mapping (SLAM)-like algorithm. Zigzag scans exhibited greater drift due to frequent directional changes.
In coastal applications, PL-derived digital elevation models (DEMs) of University Beach, a restored beach on the TAMUCC campus, achieved a vertical RMSE of approximately 10 cm relative to TLS reference DEMs, though targeting larger areas led to surface inconsistencies. For utility mapping, PL captured dense, high-resolution point clouds, with cloud-to-cloud (C2C) differences within 1.4 cm of TLS data. Data from Pix4Dcatch, outperformed raw PL scans from the 3D Scanner app in slope and width measurements. The integration of PL and UAS-SfM data within Pix4Dmatic provided improved geometric alignment compared to manual co-registration methods. Overall, the results demonstrate that PL is a practical tool for high-resolution, small to medium-scale localized 3D mapping, with strong potential in data fusion workflows and digital twin initiatives.
