Prof. Daniele Fontanelli presents Dealing with Uncertainties in Robotics

Measurement uncertainty represent the doubt and the extent of the doubt for any
empirical data collection gathered from real applications. The main consequence
of the presence of the unavoidable measurement uncertainties is largely not
explicitly considered in robotic applications, despite the possibly large
negative impact they may have. This phenomenon is exacerbated when nonlinear
dynamics come into play.

In this talk, we will analyze the effect of measurement uncertainties for
mobile robots applications, considering the analysis of the structural
properties of the systems, such as observability and constructibility, which
are preconditions to build an effective robotic solution. We will then focus on
the localization and control problems and we will show how uncertainties can be
dominated by a wise analysis of their effects for the specific problem at hand.
We will also show that, the uncertainty analysis can guide the design of
planning algorithms. We finally introduce the distributed aspects of such
problems and what are the robotics team applications that can be solved.