Computer Science meets Anthropology: A novel approach for reconstructing locomotion from fossil human footprints

两足运动是一种独特而基本的行为，它驱动和塑造了人类物种的进化。然而，人类两足动物的进化史仍然知之甚少。这个项目是第一个应用跨学科技术来理解两足动物进化的项目。PI还通过与史密森学会人类起源计划的持续合作，让更广泛的公众参与到这项研究中来。这项研究的结果及其对理解人类物种进化的意义，通过定期讲座和参与人类起源博物馆大厅的教育项目来展示。PI和联合PI从乔治华盛顿大学的Trachtenberg学者项目中招募和培训研究助理。该项目支持哥伦比亚特区公立学校系统中有前途的学生的教育和培训，这些学生通常来自STEM学科中历史上代表性不足的群体。在生物人类学、进化生物学和计算机科学领域，也不断努力将研究结果传播给不同范围的专业受众。近年来，几处保存人类足迹化石的遗址的发现为了解两足动物的进化提供了一种令人兴奋的新数据形式。然而，需要新颖的技术来分析这些不同数据源中包含的丰富信息。该项目将足迹形态与产生它们的运动模式联系起来的实验方法具有变革性。实验生物力学将与机器人视觉技术相结合，以了解如何通过3d足迹形态记录运动细节。这些数据被用于开发强大的机器学习算法，应用计算机科学和统计学交叉领域快速发展的技术，从化石人类足迹中建立详细的运动重建。通过这样做，该项目为人类物种的进化史提供了新的亮点。

Bipedal locomotion is a unique and fundamental behavior that has driven and shaped the evolution of the human species. Yet the evolutionary history of human bipedalism remains poorly understood. This project is the first to apply interdisciplinary techniques to understand the evolution of bipedalism. The PI also engages the broader public in this research through ongoing collaborations with the Smithsonian Institution's Human Origins Program. Results of this research, and its implications for understanding the evolution of our species, are presented through regular lectures and participation in educational programs at the museum's Hall of Human Origins. The PI and co-PIs recruit and train research assistants from The George Washington University's Trachtenberg Scholars Program. This program supports the education and training of promising students from the District of Columbia public school system, who typically come from groups historically underrepresented in STEM disciplines. Consistent efforts are also made to disseminate results to a diverse range of professional audiences in biological anthropology, evolutionary biology, and computer science. In recent years, discoveries of several sites preserving fossil human footprints have offered an exciting new form of data from which to understand the evolution of bipedalism. However, novel techniques are required to analyze the wealth of information contained within these diverse sources of data. This project is transformative in its experimental approach for linking footprint morphologies to the patterns of locomotion that produced them. Experimental biomechanics will be bridged with techniques for robotic vision, to understand how details of locomotion are recorded by 3-D footprint morphology. Those data are used to develop powerful machine learning algorithms, applying technologies from a rapidly-growing field at the intersection of computer science and statistics, to build detailed reconstructions of locomotion from fossil human footprints. By doing so, this project sheds new light on the evolutionary history of the human species.