Computationally modeling device repair for augmented repair guidance

Electronic waste or e-waste is the fastest growing consumer waste stream in the world. Oftentimes, e-waste is generated when electronic devices break and the average consumer determines that it’s easier to throw away a device and replace it anew rather than to find a way to repair it. In fact, many devices are intentionally designed to make repair difficult, and relevant repair guides can be limited. Furthermore, it is extremely challenging, time-consuming, and at times even impossible to acquire repair guides for the exact make and model for a given device. Moreover, it can be even more challenging for novices to extrapolate which fixes from a similar device might be worth attempting. These barriers to repair are highly unsustainable as devices are increasingly built for planned obsolescence, assuming they will become e-waste in a few years’ time. This project aims to reduce the barriers to repair for the average consumer by using computational modelling to synthesize data from diverse repair data sources (videos, manuals, guides) and organize the most relevant information in an interactive repair guide tool. We aim to build our tool so that it can be used across diverse device types (kitchen appliances, IOT devices, laptops, etc.) and to support diverse repair strategies (electrical, mechanical, etc.).

“A significant driver of e-waste is the fact that when a device breaks down, it’s much easier to toss electronic devices out and buy them anew than to try to repair them. This is in part because knowledge and infrastructure for device repair have become increasingly scarce. To address this, our work focuses on consolidating diverse knowledge bases (repair guides, device manuals, and how-to videos) through computational modeling. Using our models, we aim to generate inferences on which fixes would be most appropriate for a given device and issue. Based on our models, we will generate an interactive flow chart tool, allowing users to explore the reasoning behind repair instructions to facilitate users with building an intuition for repair knowledge. This tool offers a data-driven intervention approach to supplant limited and inaccessible repair knowledge. Through this work, we aim to support people without repair experience to feel empowered in repairing their devices.”

Jasmine Lu, Ph.D. Student and NSF Graduate Fellow, Computer Science, University of Chicago