Divya joined the ORI as a DPhil student in September 2021 under the CRG (cognitive robotics group), led by Dr. Lars Kunze. She completed her bachelors in electrical engineering at Princeton University in 2018, where she specialized in computer vision and completed a senior thesis on LiDAR segmentation for self-driving vehicles. She then worked at the self-driving car company, Cruise for two years where she worked on various ML projects involving video classification, audiovisual multimodal detection, pedestrian path prediction, LiDAR camera fusion, human pose estimation, semantic instance segmentation, and object recognition. Afterwards, she also worked at Amazon for five months on backend infrastructure.
Her research interests lie at the intersection between trajectory prediction, social interaction modeling, and long-tailed learning. She aims to explore techniques that help to overcome dataset imbalance due to the long-tailed characteristics of most datasets, specifically in regression tasks like trajectory prediction. Furthermore, she aims to improve trajectory prediction by incorporating spatial, temporal, and social information into the prediction, and wants to improve social interaction modeling by learning the more uncommon social cues.