Course Overview

How did you move your sofa into your new apartment? How did you put together your new table from IKEA? How did you find your way from home to class this morning? How did you avoid getting run over by a bus crossing the street into the engineering quad? In each case, you were motion planning---constructing a sequence of low-level actions to achieve some high-level goal in the physical world.

Over the past 30 years, an enormous amount of work has been done to automate this process of reasoning, with a variety of applications: mobile robotics (museum tour guides and planetary explorers), industrial manufacturing (robotic assembly, welding, and painting), bioengineering (robotic surgery and drug design), and civil engineering (checking access for the disabled, safe egress in an emergency, and robotic construction).

In this course, we are interested in motion planning for aerospace vehicles, defined loosely as cars and planetary rovers, underwater vehicles and spacecraft, unmanned aerial vehicles, and air traffic control systems. These vehicles are both safety-critical and challenging to operate.

But what is special about motion planning for aerospace vehicles? Simply, we require algorithms that are (1) practical, (2) easy to implement, and (3) have performance guarantees. This is so that, as engineers, we can make good design decisions.

One goal of this course is to allow engineers to understand and implement some of the powerful existing motion planning algorithms for aerospace vehicles. However, motion planning is still a young field. Most algorithms do not satisfy all of our requirements. So this course will also be focused on evaluating the state-of-the-art and identifying problems for future research.

In the first few classes, I will present the basic mathematics and algorithmic techniques of classical motion planning. Then, we will study more advanced topics together. In general, we will read two papers from the literature in each class. I will take 20 minutes to introduce the material and put it in context, and a student will take 35 minutes to present each paper.