Algorithmic Design of Origami Mechanisms and Robots

University Of Pennsylvania, Philadelphia PA

Investigators

Abstract

The goal of this project is to develop algorithms for automated design of folded robots with custom geometry and motion requirements. Folding is a commonly used and scalable manufacturing process for making 3D objects from sheets of metal or plastic. Currently, designing folded mechanisms and robots is a complex process that requires even skilled engineers to go through multiple design iterations. At the same time, however, many of the constraints that engineers must deal with are geometric in nature and can be addressed algorithmically. New algorithms for end-to-end design that convert design specifications directly into 3D physically realizable robots would simplify the design process, lower barriers to robotics, and open opportunities for custom robotics on demand when rapid deployment is required, for example in emergency relief or in search and exploration of unknown environments. In order to develop these algorithms, the researchers will address fundamental questions in origami-inspired engineering, robotics, and computational geometry. Although the work will focus on origami-inspired fabrication, parts of the resulting system will also be applicable to other fabrication techniques, including 3D printing or more traditional machining. The research will provide formal insights into algorithmic design that enables fully automated end-to-end design pipelines, resulting in valid, functional, robotic components within hours or days. The specific objective of this project is to formalize the design of kinematic chains and trees that have required workspaces or achievable trajectories. The main insight underlying this project is that geometric design problems of this nature can be mapped directly onto path planning problems that have been explored by the computational geometry and robotics communities. In particular, the researchers take a modular approach, in which a robot is constructed as a combination of link and joint components from a database, and each component is associated with an equivalent path type. By casting the problem in this way, the problem of kinematic design can be partially abstracted from the specific fabrication details. The researchers will develop algorithms for converting kinematic joint specifications into equivalent paths and robot designs, as well as formal guarantees for when such a process is or is not possible. They will fold the resulting designs out of plastic sheet in order to ensure that they are physically realizable. The results will be disseminated through a combination of open-source software with an associated graphical user interface for ease of use, workshops at engineering conferences, and arts and educational workshops in and around Philadelphia. The research will have broad societal impacts by making the resulting design software accessible to a diverse population, including K-12 students, teachers, hobbyists, and other non-engineering users. 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.