Back to search

Cornell University

REPRODUCING ENCODED TRAJECTORIES FOR NOVEL LOCATIONS

Abstract

dc:description.abstract

Trajectory learning from demonstrations has the ability to interactively teach robots newskills, eliminating the need for manual programming of a desired behavior. In this thesis, I have implemented Trajectory Learning from Generalized Cylinders (TLGC), a trajectory-based learning approach through demonstrations. Generalized Cylinder (GC) is a geometric representation composed of an arbitrary space curve defined as the spine, along with a varying cross section. GCs are able to extract implicit characteristics of a skill by encoding the demonstration space. They are able to reproduce multiple trajectories maintaining the characteristics of the demonstrated trajectory. This constructed model can be used to generalize a skill for unforeseen circumstances through Laplacian trajectory editing. This is achieved by adding fixed points that must be passed through, while keeping the local shape similar. This work has been validated through a real-world demonstration with a Hello Robot – Stretch RE1. The strengths and weaknesses of this approach are discussed in this thesis.

Degree

thesis:*
Name thesis:degree_name
M.S., Mechanical Engineering
Level thesis:degree_level
Master of Science
Discipline thesis:degree_discipline
Mechanical Engineering
Grantor
Cornell University
Year dc:date.issued
2023

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Shiva, Shashank
Chair dc:contributor.chair
  • Kress Gazit, Hadas
Committee member dc:contributor.committeemember
  • Bhattacharjee, Tapomayukh

Subjects

dc:subject × 5

Rights

Language dc:language.iso
en

Identifiers

dc:identifier.*
Dc Identifier Bibid
16219335
Dc Identifier Other
ProQuest Submission ID: 11863
ProQuest Publication ID: 30572112

Chain of custody

source
Harvested from
Cornell University
Base URL
ecommons.cornell.edu/server/oai/request
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Shiva, Shashank. REPRODUCING ENCODED TRAJECTORIES FOR NOVEL LOCATIONS. Master of Science thesis, Cornell University, 2023. https://hdl.handle.net/1813/114470