Ph.D., Purdue University, May 1990
Co-Major Professors: Phillip C-Y. Sheu and Anthony A. Maciejewski
This thesis focuses on the problem of path planning for mobile robots with manipulators in a common work space. Most of the existing research of robot path planning has assumed that a mobile robot has a fixed shape, i.e., a circle or a polygon. In this thesis, several collision-free path planning algorithms for mobile robots whose shape can be changed during motion are described. The first part of the thesis studies the path planning problem for two tightly cooperating robot manipulators. In our approach, 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 collisron-free path finding algorithm. These algorithms can deal with cluttered environments and are guaranteed to find a solution if one exists.
The second part of the thesis considers the path planning problem for a mobile robot with manipulators in a two-dimensional environment. In this case, 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. Extending the results obtained in the first part, 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. In this path planner, 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. Based on these representations, two levels of planning, i.e., global path planning and local path planning, are used. The global path planner generates a best candidate path for the robot. Subsequently, 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. Finally, the best path for the robot in the 3-D environment 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. In order to acquire more insight to the problem, this thesis studies an off-line discrete-time collision-free path planning scheme for two circular robots working in a common space. 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.