CS 497: Algorithmic Motion Strategies


Fall 2001 TuTh 2:00-3:15 Room 149 English Bldg.
Registration: 1 unit, 08904 LECD SML

Instructor: Prof. Steve LaValle
2107 DCL, lavalle@cs.uiuc.edu

Consider these questions:
How do I move a piano out of a house without hitting anything?
How should a humanoid robot move to pick up a hammer?
Can the motions of virtual actors be automated in movie production?
How should thrusters should be fired to gently guide a satellite into the shuttle bay?
Does a proposed building design allow for easy wheelchair access?
How many reversals would it take to park a car in a tight spot?
Can I remove an automobile part without having to remove other parts?
Is a given drug molecule likely to dock with a given protein?
What foldings are possible for a given protein?
How can I visually monitor an animated character with a floating, movable camera?
How can I always win at the hide-and-seek game?

The course considers the design and implementation of algorithms that automate the motions of bodies in complicated environments. The material draws on recent, selected research in artificial intelligence, robotics, computer graphics, computational geometry, and computational biology. Each of these fields contains problems that involve constructing detailed geometric models, and the need to develop algorithms that compute motion strategies. A motion strategy must satisfy problem constraints, such as reaching a goal, avoiding collisions, and maintaining visibility of a target. The goal of this course is to provide a unified treatment of current mathematical and algorithmic techniques that address such problems across a wide variety of applications. In addition to the presentation of basic material, the course includes reading and discussion of recent research papers, and an implementation project.

Topics: Geometric and Kinematic Modeling. Transformation Spaces and Topological Concepts. Collision Detection Algorithms. Path Planning Techniques. Rapidly-exploring Random Trees (RRTs). Decision-Theoretic Models and Algorithms. Differentially-Constrained Motions and Trajectory Design. Sensor-Based Motion Strategies. Relevant Issues in Robotics, Computer Graphics, and Computational Biology.