Robot manipulation planning for complex structures

Zhang, Zebang (2022). Robot manipulation planning for complex structures. University of Birmingham. Ph.D.

[img] Zhang2022PhD.pdf
Restricted to Repository staff only until 31 December 2023.
Available under License All rights reserved.

Download (3MB) | Request a copy


Robots performing pick-and-place tasks need to be given a specific end-effector pose to grasp the object. This pose is usually generated by a grasp planner without considering the future motion of the robot. This is acceptable for manipulating small objects with simple geometry inside a large free space. However, for complex structures being manipulated in a cluttered environment, some grasp poses may lead to redundant robot motion or even failure of transferring the object to the desired goal location. This thesis addresses this limitation by designing methods and algorithms that can solve the integrated problem of grasp selection and motion planning in three different scenarios. In the first scenario, feasible grasps can be sampled directly given the geometry of a complex 3D pipe structure. The robot needs to manipulate the pipe following a given object path while optimising several objectives. In the second scenario, a set of precomputed feasible grasps of an arbitrarily complex object is assumed and only the start and goal pose of the object are given. An optimised path needs to be generated offline and the robot will execute this path repeatedly. Finally, the third scenario assumes the same planning problem as the second scenario but requires the robot to be extremely responsive. The robot needs to generate a feasible motion online as fast as possible to manipulate the object to the desired pose. In these scenarios, a desirable grasp can be chosen, and the subsequent motion of the robot is planned simultaneously by using the proposed methods. The developed methods are generally applicable. However, they will be most useful in cases involving the manipulation of large and complex objects that have many available grasps in a relatively narrow environment. A specific application scenario of the developed algorithms is the Factory-In-A-Box (FIAB) project. The project aims at manufacturing complex pipe structures fully automatically inside a compact container and an industrial robot is employed to transfer pipe structures between different manufacturing processes. Experiments show the advantages of the developed algorithms in terms of both path costs and planning time compared to the traditional manipulation planning method where only a single grasp is generated for motion planning.

Type of Work: Thesis (Doctorates > Ph.D.)
Award Type: Doctorates > Ph.D.
Licence: All rights reserved
College/Faculty: Colleges (2008 onwards) > College of Engineering & Physical Sciences
School or Department: School of Engineering, Department of Mechanical Engineering
Funders: Other
Other Funders: Innovate UK
Subjects: T Technology > TJ Mechanical engineering and machinery
T Technology > TS Manufactures


Request a Correction Request a Correction
View Item View Item


Downloads per month over past year