Situation analysis exploiting feasibility constraints in uncertain environments

Abstract

Deals with the problem of uncertainties in the nominal world model of a robot. In particular, the problem of analyzing unexpected situations in a 2D polygonal world is attacked. Such a situation analysis can be used to acquire information on a contact situation encountered during the execution of a motion. The analysis is based on the concept of so-called test motions, i.e. motions, the feasibility of which depends on the actual contact situation present. Using a probabilistic uncertainty model, the focus is on the computation of motion constraints due to contact and the resulting selection of an optimal test motion.<>