A homotopy method for motion planning

Li, Yuying

Permalink

https://hdl.handle.net/2142/99999 Copy

Description

Title

A homotopy method for motion planning

Author(s)

Li, Yuying

Contributor(s)

Belabbas, Mohamed-Ali

Issue Date

2018-05

Keyword(s)

• motion planning
• non-holonomic system

Abstract

The field of robotics has developed at a promising speed these years, and its potential usage is tremendous in both traditional industry and frontier research. Motion planning, as an important branch of robotics, is the process of completing tasks and avoiding obstacles while achieving optimization at the same time. Even though many ad hoc algorithms have been proposed to solve such problems, most of them will deal constraints and optimization separately. As a result, there is an urgent need for an algorithm to handle them together. With the aim to solve this problem, this paper introduces a new method with the help of homotopy, and such a method and its application for motion planning is examined comprehensively. This method will assemble all holonomic, non-holonomic, and obstacle constraints into a metric. Furthermore, applying partial differential techniques on this metric allows continuously deforming an arbitrary path into another one in the configuration space. A shortest path will be found that satisfies all given constraints such as obstacles and implementation requirements. The corresponding mathematical and control background will be reviewed. Then this thesis will present the details of the homotopy method and how to implement it in reality. Finally, two models with different degrees of freedom will be examined to look at the feasibility of such a method.

Type of Resource

text

Language

en

Permalink

http://hdl.handle.net/2142/99999

Owning Collections