Biography

I am a research fellow at MIT CSAIL working with Nicholas Roy. My research is focused on learning robust, visually-guided navigation and planning policies for heterogeneous robots operating in unstructured, unknown, real-world environments. I am interested in semi-supervised and unsupervised learning methods that enable robots to continually improve their understanding of the environment that will lead to both higher navigation performance and richer scientific data collection. Specifically, I am interested in:

• Visually guided navigation and open-world scene understanding
• Manipulation in unstructured environments
• Hybrid model-free and model-based learning
• Planning and control over long horizons from compact latent spaces
• Imitation learning from image data and low-dimension human labels
• Transfer learning from high-fidelity simulators to the real-world
• Multi-modal planning and control during sensor dropout and partial observability
• Improved scientific data collection
• Active perception for improved localization
• Robust performance in unstructured outdoor environments, such as forests or underwater