Robot active touch exploration: constraints and strategies

The problem of using active touch (haptic) exploration to recognize a 3-D object taken from a known set of models is examined. Two approaches are combined: (1) the use of geometric constraints between components to eliminate interpretations and of interpretation-tree methods for choosing the best active sensing move; and (2) exploratory moves made by tracing continually along the surface of the object (and not through free space). The study is restricted to polyhedra, and a set of geometric constraints tailored for matching components acquired from haptic exploration against components in the models is given. A new constraint using pairs of line segments is presented. A set of active sensing moves, each with an associated cost measure, and strategies for choosing the next move are given