Understanding and classifying lithic use wear: a systematic study using controlled tribological experiments and computer vision

点击翻译按钮获取中文摘要

The history of technology is crucial to understanding what makes us human. Microscopic wear traces present on the earliest (stone) tools excavated by archaeologists, offer key insights into how ancient people worked a variety of materials, like wood, skins, ivory, etc. Despite the enormous interpretive potential, archaeologists have experienced problems related to the difficulty and efficiency of securely identifying worked materials from wear traces, as well as reproducing each other’s results. This project will harness the power of tribology and artificial intelligence (AI) to advance understanding how wear patterns form and, at the same time, facilitate and democratize the identification process. Beside scientific publications, the investigators will produce an open-source tool for helping other researchers use their algorithms in their own work. The produced results will thus be useful to both the bio-engineering community as well as to archaeologists. To address pipeline issues in archaeological science, the project will increase diversity and the representation of minority groups by engaging high school students from New York City in the lab experiments.The project seeks to solve a 70-year standing problem in archaeology: how does wear form on the surface of stone tools as a result of working different materials? Answering this question will generate predictions that will enable large-scale studies of archaeological wear to be carried out in the future. To tackle this problem, an interdisciplinary team of archaeologists, together with mechanical and bio-materials engineers will first catalogue the mechanical properties of natural worked materials, such as wood, ivory, skins, etc. The researchers will then build surrogate biomaterials isolating the role of these properties (such as hardness, elasticity, fracture toughness, etc.), so that their effect on wear formation can be tested individually in experiments. Each surrogate material will be rubbed against experimentally produced stone bits using a tribometer. The result will be a large collection of recorded wear patterns that can be causally correlated with worked material properties. In a second experiment, stone bits will be used in more realistic tasks using a force-controlled robot and organic target materials to generate a large collection of wear patterns that approximate archaeological wear. Information from both experiments will be used to train computers to classify archaeologically occurring wear patterns according to the materials they most likely came into contact with. Moreover, classification algorithms with and without expert input for identifying worked materials will be compared, which addresses the application of artificial intelligence in archaeology, a topic which is currently of great interest to archaeologists in general.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.