Constrained manipulator path planning using local geometry information국지적 기하학 정보를 이용한 구속 조건 하의 메니퓰레이터 동작 계획

Cited 0 time in webofscience Cited 0 time in scopus
  • Hit : 773
  • Download : 0
In this thesis, we introduce new algorithm for motion planning problem under constraints. So far, the research of motion planning has been progressed. As the degree-of-freedom of system is going higher, sampling-based motion planning method becomes main solution of nding path from initial to goal point. One of the most general sampling-based motion planning algorithm is Rapidly-exploring Random Tree(RRT) algorithm. RRT algorithm has good performance of searching unexplored region in any dimension of the searching space. Combining the greedy heuristics of bidirectional search with RRT algorithm, Bidirectional RRT algorithm is also introduced. It shows powerful performance of connecting two search trees, one is from initial point the other is from goal point. One of the specific research area is motion planning problem considering many kinds of constraints. Many research of this area has also been proceeded for many researchers. To satisfy constraints, especially for the equality constraints, projection method is typically used which is implemented based on the Newton-Raphson method. It has an advantage of not to discarding any random samples when sampling procedure of RRT-based algorithms. Instead, it projects these random samples onto the constraint manifold, subspace of the joint space which is created by the equality constraints. Also, there is an issue of selecting distance metric when extending search trees. Many of the researches approach this topic as if they just hope for Lp metric is going well. But selecting appropriate metric is as hard as solving motion planning problem. It causes the unexpected results when searching trees such as an unnecessary nodes, ability of searching unexplored area is decreased. Proposed algorithm focuses on these problems to help planner create searching tree as the user expects. It decreases the creation of useless nodes and alleviate the burden of projection method. As a result, it reduces total computational costs and i...
Advisors
Lee, Ju-Jangresearcher이주장
Description
한국과학기술원 : 로봇공학학제전공,
Publisher
한국과학기술원
Issue Date
2014
Identifier
568858/325007 / 020123415
Language
eng
Description

학위논문(석사) - 한국과학기술원 : 로봇공학학제전공, 2014.2, [ v, 40 p. ]

Keywords

Sampling-based; 곡률; 구속 조건 메니폴드; 쌍방향 검색; 샘플링 기반; Curvatures; Bidirectional Search; Constraint manifold

URI
http://hdl.handle.net/10203/197200
Link
http://library.kaist.ac.kr/search/detail/view.do?bibCtrlNo=568858&flag=dissertation
Appears in Collection
RE-Theses_Master(석사논문)
Files in This Item
There are no files associated with this item.

qr_code

  • mendeley

    citeulike


rss_1.0 rss_2.0 atom_1.0