LSTM-based Cognitive Assistant System for Emergency Response; Evaluating Waste Sorting Methods for Recyclable Plastics

This STS thesis examines how AI technologies can be developed that address the traditional bottleneck in waste sorting in order to expand domestic recycling infrastructure. These technologies can be adopted from existing systems developed in China as long as they conform to local values. The degree to which they do may determine their success in a domestic context. The development of these technologies can be used to mitigate environmental risk by increasing the capacity of domestic infrastructure and decrease the reliance on foreign export markets. How do privacy and security concerns restrict the implementation of these waste processing technologies locally? How can these concerns be assuaged? This paper will examine components of specific technologies as well as the attitudes of American and Chinese stakeholders. This paper utilizes Value Sensitive Design as a lens through which to view plastic recycling and waste sorting technologies.
The technical thesis details the development of a subcomponent of a larger EMS cognitive assistant system that analyzes relevant speech data at an incident scene in order to infer incident context and suggest appropriate medical responses based on standard EMS protocols. The larger cognitive assistant system aims to dynamically recommend situation-aware response actions. Relevant speech data includes verbalized observation as well as conversations between first responders. This system builds off of past research and incorporates deep learning architectures into its design. The main task of the subcomponent is to serve as a modular component that is able to identify EMS protocols given a set of observed concepts.
The technical and STS theses are not related

School of Engineering and Applied Science
Bachelor of Science in Computer Science
Technical Advisor: Homa Alemzadeh
STS Advisor: Sharon Ku
Technical Team Members: