Could robots become archaeological assistants, shuffling or trudging across sandy terrain like R2D2 and C3P0 in 1977’s original Star Wars?
Artificial Intelligence (AI) and machine-learning algorithms, along with geospatial data, are being used to uncover mysterious and ancient geoglyphs, courtesy of a collaboration between IBM power systems and Yamagata University. (A geoglyph is characterized by the formation of clastic rocks or likewise durable landscape elements, like stones, stone fragments, live trees, gravel, or earth into a large design or motif on the ground.)
And, using the new AI, scientists discovered a new formation of very large geoglyphs in the soil on the Nazca Lines in southern Peru— the first to be found using AI. While straight lines dominate the Nazca desert landscape, figurative designs of animals and plants have evolved.
The development, research, and recent discoveries by Japan’s Yamagata University and IBM Research’s IBM Power Systems, which was launched in February 2019.
Scientists say they are expanding the scope of the discovery and will deploy its proprietary tech of PAIRS geoscope and IBM’s AI-enabled cloud platform to quickly layer and analyze massive volumes of disparate geospatial temporal data. This will sense and examine the surface of the Earth, drone images, satellite visuals, and geographical survey information, according to the scientists.
SEE: The ethical challenges of AI: A leader’s guide (free PDF) (TechRepublic)
Before this unprecedented AI, formations were assessed through a “process that researchers typically did manually by studying and analyzing different individual photos, a very time-consuming and resource-intensive process,” said Hendrik Hamann, chief scientist for Geoinformatic Solutions, PAIRS, and distinguished research staff member, IBM research.
Two critical acronyms in the process are PAIRS (Physical Analytics Integrated Data Repository &amp; Services) and LiDAR (Light Detection and Ranging).
Uncovering new Nazca Lines has historically been difficult due to the amount of “white noise” surrounding them, i.e. irrigation lines, roads, flood trails and geographical discrepancies and changes, reported IBM research in a release. Using AI, PAIRS will sift through this volume of data to pinpoint relevant clues, with the goal of accelerating how quickly Nazca Lines can be uncovered and understood.
As a first step to gauge AI’s feasibility in archeological discovery, the team trained an AI model with Watson Machine Learning Accelerator on IBM Power Systems with known photos of geoglyphs.
“The Watson Machine Learning Accelerator (WMLA) is an AI platform that helps clients take advantage of deep learning frameworks, AI development tools, and machine learning built on IBM Power Systems servers,” explained Sumit Gupta, VP, AI, machine learning and HPC, IBM systems.
“By using AI on a highly secure and powerful server, WMLA accelerates time to insights by helping Nazca Line researchers to identify new glyphs more quickly and accurately in a matter of months instead of years, automating the traditional archaeology process,” Gupta added.
It took eight people four years to develop, Hamann said, and added, “Unlike conventional systems, PAIRS uses under-the-cover distributed storage and file system, which is much more ‘scalable’ (works with very big data sets) because in a distributed system workloads and AI can be executed in parallel among many more server nodes.”
“From a data perspective,” Hamann said, “this is pretty massive [AI] given the size of data sets to be analyzed; 80% of the AI exists, but it has never deployed to such a diverse and complex set of data.”
Hamann further explained: “From what we know, PAIRS is the only scalable data analytics platform for both raster/imagery and vector, like LiDAR, data. We believe that the next frontier of geospatial-temporal intelligence is exactly that, how to integrate and leverage both massive LiDAR and raster/imagery information to build more reliable/trusted ML and AI models. Developers can benefit from PAIRS as it presents a playground of data and analytical capabilities to develop novel applications.
“Organizations and businesses have already begun to use PAIRS to improve how multiple sources of data can be integrated to benefit large-scale operations,” Hamann said.
For example, PAIRS is being used by agricultural companies for better crop identification and to design irrigation management strategies. Additionally, utility companies have deployed PAIRS to monitor vegetation growth around assets such as power lines to reduce the risk of disruptions.