IRES Track I: Computational Co-Design of Physical Systems with Embodied Intelligence by Integrating Data, Simulation, and User Interface

点击翻译按钮获取中文摘要

Designing embodied intelligence for physical systems is an emerging challenge in today's engineering and computing practices. This IRES project aims to tackle the challenge on both sides of software infrastructure creation to bridge scientific communities and research training curriculum development to educate STEM students. By collaborating with two user interface research groups at the University of Tokyo and Japan Advanced Institute of Science and Technology (JAIST), we bring together an interdisciplinary team with their complementary expertise in physics simulation, digital fabrication, data science, and human-computer interaction to devise novel computational tools to support the interactive design of intelligence-embodied physical systems. Fifteen U.S. students, both graduates and undergraduates, will visit Japan in the summer and jointly lead the research project by working with their collaborators at the University of Tokyo and JAIST. The three-month study and research experience will immerse the students in a unique research environment to develop their scientific understanding and programming practices to address the frontier research problems crossing artificial intelligence (AI), scientific computing, user interface, and computational design. The follow-up mentoring activities will guide the students' self-motivated exploration and future professional development in this interdisciplinary field. The overarching goal of this research project is to build an open-source software framework to lower the barriers for students, researchers, and engineers to conduct interactive design and visual programming of data-driven intelligent algorithms for various customizable physical systems, including examples of soft-bodied robots, drones, and materials. On the side of infrastructure development, we aim to democratize the creation of adaptable AI algorithms in specific physical and engineering contexts by providing an easy-to-access software framework to help designers conduct their data-integrated design tasks across engineering and computing communities. The software's architecture consists of three essential components --- frontend design interface, backend physics simulator, and data-driven physical-intelligent coordination --- which naturally transition to fifteen clustered sub-projects for the IRES students. On the side of research training development, we plan to redefine the standard way of problem-solving in physical intelligence by transitioning the workflow from the traditional data collection and model activity to a fully integrated design loop by interactively co-designing the data model, physical principles, and intelligent algorithms. We plan to train an inclusive group of students, with a particular focus on minorities and underrepresented students, by developing their new skillsets through developing this open-source infrastructure crossing physical and data sciences.This project is jointly funded by the International Research Experience for Students program and the Established Program to Stimulate Competitive Research (EPSCoR).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.