Path planning for mobile robots with manipulators
Abstract
This thesis focuses on the problem of path planning for mobile robots with manipulators in a common workspace. Several collision-free path planning algorithms for mobile robots whose shape can be changed during motion are described. The first part studies the path planning problem for two tightly cooperating robot manipulators. The two robot arms, the carried object and the straight line connecting the two robot bases together are modeled as a 6-link closed chain. The problem of path planning for the 6-link closed chain is solved by two major algorithms: the collision-free feasible configuration finding algorithm and the collision-free path finding algorithm. The second part considers the path planning problem for a mobile robot with manipulators in a two-dimensional environment. A mobile robot is modeled as a serially connected chain of rectangles, and the mobile robot and the carried object together are defined to be a reconfigurable moving object. A collision-free path for a reconfigurable moving object in a 2-D environment can be obtained as well. In order to plan a path for a mobile robot whose shape can be changed in a 3-D environment, a 3-D path planner is used. Different sizes of cuboid objects are used to model different states and operations of a mobile robot. A channel representation is employed to represent the free space in the environment. Two levels of planning, are used. The global path planner generates a best candidate path for the robot. The local path planner employs an expert system to build a connection graph which contains all the possible ways for a robot to travel along the candidate path generated by the global path planner. The best path for the robot is obtained by searching the connection graph. When multiple robots work concurrently in a common workspace, it is important to plan a set of collision-free paths such that the motion time for the robots can be minimized. The problem is formalized as a minimax time collision-free path search problem. The complexity of the problem is first reduced by applying the collision-free constraint. Subsequently, a solution is found by the golden section search method. (Abstract shortened with permission of author.)
Degree
Ph.D.
Advisors
Maciejewski, Purdue University.
Subject Area
Electrical engineering|Artificial intelligence
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.