Date of Award

2014

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Graduate Group

Mechanical Engineering & Applied Mechanics

First Advisor

Vijay Kumar

Second Advisor

Nathan Michael

Abstract

Large teams of mobile robots have an unprecedented potential to assist humans in a number of roles ranging from humanitarian efforts to e-commerce order fulfillment. Utilizing a team of robots provides an inherent parallelism in computation and task completion while providing redundancy to isolated robot failures. Whether a mission requires all robots to stay close to each other in a formation, navigate to a preselected set of goal locations, or to actively try to spread out to gain as much information as possible, the team must be able to successfully navigate the robots to desired locations.

While there is a rich literature on motion planning for teams of robots, the problem is sufficiently challenging that in general all methods trade off one of the following properties: completeness, computational scalability, safety, or optimality. This dissertation proposes robot interchangeability as an additional trade-off consideration. Specifically, the work presented here leverages the total interchangeability of robots and develops a series of novel, complete, computationally tractable algorithms to control a team of robots and avoid collisions while retaining a notion of optimality.

This dissertation begins by presenting a robust decentralized formation control algorithm for control of robots operating in tight proximity to one another. Next, a series of complete, computationally tractable multiple robot planning algorithms are presented. These planners preserve optimality, completeness, and computationally tractability by leveraging robot interchangeability. Finally, a polynomial time approximation algorithm is proposed that routes teams of robots to visit a large number of specified locations while bounding the suboptimality of total mission completion time. Each algorithm is verified in simulation and when applicable, on a team of dynamic aerial robots.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS