The Need for Speed: Enabling Real-Time Manipulation through Optimization, Simulation and Deep Learning

Robots have the potential to increase productivity and reduce costs and injuries in a variety of industries, including logistics, manufacturing, agriculture, surgical procedures and healthcare. To do so, robots are required to operate robustly, reliably and increasingly more efficiently in unstructured or semi-structured environments under a high degree of uncertainty in perception and control. In this presentation, I will describe my work on improving and enabling efficient robot manipulation by addressing challenges in sensing, planning and control through optimization, simulation and deep learning.