MIT EECS - Draper Laboratory Undergraduate Research and Innovation Scholar
Data Structures for Robotic Motion Planning in Related Spaces
Online robotic motion planning is computationally expensive and can generate noisy and unpredictable paths. It also is limited to implicit queries for paths and reachability in a robots configuration space of safe poses. By representing these configuration spaces with hybrid data structures that have the properties of spatial partitioning trees and graphs, online motion planning may be faster. These data structures amortize motion planning costs through offline computation and online updates to respond to changes in the world. Additionally, explicit representations of the configuration space will allow for approximately optimal object placement. Finally, studying the relationships between several low-dimensional configuration spaces for different robot parts will give insights on how to couple data structures to represent the complete configuration space.
My first research experience on image processing research at the Naval Research Laboratory was published in Naval Engineers Journal. Since freshman year, I have worked with the Learning and Intelligent Systems Group on robotic motion planning projects as UROP. Most recently, I interned with the Revenue team at Twitter.