Search for: All records

New generative AI tools are successfully responding to symbolic streams that represent vocalization, e.g., English sentences and paragraphs, but these tools struggle to engage with bodily action, e.g., a slouch that accompanies a punch. Developing robots that are responsive to such “body language” requires a symbolic system corresponding to the postural and gestural bodily shifts that accompany vocalization. This pictorial paper presents initial work developing a system of annotation, rooted in movement theory and dance notation, for short video clips of gross bodily movement. The paper analyzes annotations from two annotators across two examples of movement in three distinct notation formats. A similarity score is developed to highlight trends in inter-annotator agreement across these examples in order to take steps toward a regularized, efficient system that can be employed at scale. This work supports the development of large repositories of annotated human data for robotics and AI and demonstrates a key opportunity where new advances in movement analysis can directly impact the design and development of technology. 

Genetic screens provide a mechanism to identify genes involved with different cellular and organismal processes. Using a Flp/FRT screen in the Drosophila eye we identified mutations that result in alterations and de-regulation of cell growth and division. From this screen a group of undergraduate researchers part of the Fly-CURE consortium mapped and characterized a new allele of the gene Hippo, HpoN.1.2.