I am motivated by synthesizing task-driven perception and control that is robust to task-irrelevant distractors and performs well in uncertain or unseen environments. In particular, I am excited about questions such as:

• How does a robot capture and represent task-relevant information?

• What (in the environment) should a robot pay attention to for its task?

• How does a robot switch what it is paying attention to when something (e.g., task or dynamics) changes in the environment?