Georgia Institute of Technology
The goal of this research is to continue to support RT and CC projects in the Institute that are researching systems focused on the context of the kitchen and supporting kitchen tasks. Our goal is to help inform and then translate foundational AI research into services could improve quality of life for people with MCI and their support network by reducing the burden on care partners and allowing them more autonomy when it comes to meal planning, prep, cleaning, maintaining healthy diet, entertaining etc.
Over the past year, we have been collaborating closely with Brian Jones’ team, who are instrumenting the CEP and AwareHome kitchens, creating prototype interventions, and evaluating sensing efficacy as well as appropriateness for target users. We have also been working with the Y2 Arena 1 researchers to instrument the kitchen with additional sensor streams (body motion via wearables; speech; use of stove; opening of drawers and cabinet/fridge doors) to ground technical work in detecting activities and routines with real-world data. We are collecting Mbient data that is needed for human activity recognition for the target population. We have also been supporting Alan Ritter’s team, who are working on language models for chatbots that guide people through the steps in a recipe. We are able to capture the audio of conversations that are happening during our kitchen data collection cooking sessions. Lastly, we are currently preparing to run wizard of oz (WoZ) experiments based on Larry Heck and Alan Ritter’s work that explore conversational assistance for kitchen activities.
As the year progresses our activities will evolve more into user experience design and evaluation, to translate the foundational AI research into usable and useful interventions for our target community. We will collaborate early on with the RT/CC research teams to help define the requirements for the algorithms and subsequent interactive prototypes to ensure that they are designed with a user-centered approach and to provide HCI expertise where needed. We will also work to ensure that the prototypes have the features needed for future evaluation in our kitchen testbeds (e.g., designing a wizard of oz interface for specific protocols etc.). We will also assist with the design of the study protocol and can conduct the study or assist the team in running it in an ecologically valid context and with target users.