SOAR TECHNOLOGY, LLC — National Aeronautics and Space Administration SBIR Phase I: H6

Description

Due to delays in communication during deep space missions, astronauts will need to solve more problems without connection to ground teams. To reduce overburdening, the astronauts will leverage new decision aids including procedure generation tools. New virtual assistants are needed that can explain decision aid outputs, reducing ambiguity, increasing task performance, and helping to calibrate astronaut trust. Virtual assistants implemented using cognitive agents can model and reason about problems and interact like humans do, enabling them to act as an onboard teammate and deliver relevant information in ways that crew members can easily understand. However, to serve the needs of astronauts, these agents must be 1) adept at providing coherent, interactive explanations, 2) able to tailor these explanations for different crew roles and 3) able to leverage different multi-modal output channels, as the situation demands. To address these requirements, we propose Virtual Agent for Interactive Explanation (AVA-Ex), a research effort to design and assess the feasibility of building a cognitive-architecture-based interactive explanation agent. AVA-Ex will support human deep-space exploration by demonstrating that such an agent can improve end user task performance and trust when conducting system diagnostics procedures with dynamically generated procedures. We will leverage Soar Techrsquo;s 25 years of experience developing and applying cognitive architectures, the Principal Investigatorrsquo;s 15 years of experience developing cognitive architecture based virtual agents, and SoarTechrsquo;s recent work developing cognitive architecture-based explanation agents. We will also leverage our SME Dr. Greg Chamitoffrsquo;s deep understanding of NASA space mission operations. Phase 1 will result in a design and feasibility analysis with a simple demonstration; Phase 2 will result in a working prototype, followed by a demonstration in a representative environment. nbsp;