AD ASTRA: Automated Detection of Attitudes and States through Transaction Recordings Analysis
NASA's Human Research Program needs methods to identify the most likely and serious threats to task performance, teamwork, and psycho-social performance during long duration flight. Factors such as lack of team coherence, workload, social monotony, access to family and psychosocial support, and interpersonal and cultural differences are known to affect both crew welfare and task performance. While the techniques exist in the social sciences to perform this research and eliminate this knowledge gap, the real problem is to develop automated techniques and validate them using data from high fidelity simulations or actual spaceflight.
Since psychosocial states of interest to space operations are affected, even largely the product of, interpersonal communication, it is not surprising that interpersonal communications are our primary key to them. Recent research suggests that verbal and non-verbal communications can be automatically processed in a variety of ways to provide insight into team cohesion, affective and cognitive states and team performance. We leverage prior work of our own and of others in cultural and socio-linguistic theory to develop standardized, non-intrusive and automated methods for data collection and knowledge extraction about factors salient to crew psychosocial well-being, including mood, topic specific sentiment, team dynamics, temporal orientation, self-vs.-others focus, and meaningful work. We have developed assessment techniques for relevant team coherence and performance factors, and validated them using historical spaceflight data as well as a series of experiments in three separate ground based analogs, including UTMB bedrest, HERA, and HISEAS.
A fact sheet is available for download: [pdf]