Robotics: Probabilistic Methods for State Estimation and Control
Full Day Tutorial
September 19, 2013
Informatik 2013
Koblenz
Abstract
Probabilistic robotics is a new and growing area in robotics,
concerned with perception and control in the face of
uncertainty. Building on the field of mathematical statistics,
probabilistic robotics endows robots with a new level of robustness in
real-world situations. Probabilistic approaches have been discovered
as one of the most powerful ways to address highly relevant problems
in mobile robotics, including robot state estimation and
localization. Major challenges in the context of probabilistic
algorithms for mobile robot navigation lie in the questions of how to
deal with highly complex state estimation problems and how to control
the robot so that it efficiently carries out its task. Robots are
inherently uncertain about the state of their
environments. Uncertainty arises from sensor limitations, noise, and
the fact that most interesting environments are - to a certain degree
- unpredictable. When "guessing" a quantity from sensor data, the
probabilistic approach computes a probability distribution over what
might be the case in the world, instead of generating a single "best
guess" only. As a result, a robot using probabilistic methods can
gracefully recover from errors, handle ambiguities, and integrate
sensor data in a consistent way. And, moreover, it knows about its own
ignorance. Autonomous robots must act in the face of uncertainty - a
direct consequence of their inability to know what is the case. When
making decisions, probabilistic approaches take the robot's
uncertainty into account. Some consider only the robot's current
uncertainty, others anticipate future uncertainty. Instead of
considering the most likely situations only (current or projected),
many probabilistic approaches strive to compute a decision-theoretic
optimum, in which decisions are based on all possible
contingencies. This tutorial will discuss both aspects and present
recently developed techniques based on particle filters for estimating
the position of a mobile robot and for efficiently learning a map of
an unknown environment with a mobile robot. We will furthermore
describe how graph-based representations can be utilized to find
maximum likelihood solutions. We will finally describe how the
complexity of this state estimation problem can be reduced by actively
controlling the vehicle. For all algorithms, experimental results will
be presented that have been obtained with mobile robots in real-world
environments.