An introduction to the anchoring problem

Abstract

Anchoring is the problem of connecting, inside an artificial system, symbols and sensor data that refer to the same physical objects in the external world. This problem needs to be solved in any robotic system that incorporates a symbolic component. However, it is only recently that the anchoring problem has started to be addressed as a problem per se, and a few general solutions have begun to appear in the literature. This paper introduces the special issue on perceptual anchoring of the Robotics and Autonomous Systems journal. Our goal is to provide a general overview of the anchoring problem, and highlight some of its subtle points.

Semantic world modeling using probabilistic multiple hypothesis anchoring

“…Therefore, the world modeling algorithm has to link measured attributes to semantically rich objects in the world model. Creating and maintaining this link between instances of grounded object classes and measurements is called anchoring [6]. 2.…”

“…In the field of artificial intelligence, the first ''systematic study of anchoring as a problem per se'' was presented in [8] and refined in [6,[9][10][11]. Next to meeting Requirements 1 and 4, these anchoring strategies all allow for the incorporation of tracking (Requirement 3).…”

“…As a possible solution, [12] extended the work of [6] by adding a strategy that collects additional sensory information with the aim of resolving the ambiguous situation, e.g., let the robot drive around the object and detect visual differences that were invisible from the robot's original position. However, this solution is considered to be incomplete, since this often is impossible or undesirable.…”

“…An extension of [6] focusing on cooperative anchoring in multirobot systems is [13]. Their approach can be used for single robots as well and data association is included.…”

“…Even though the term anchoring is not used, their system can be interpreted as an anchoring algorithm. The main advantage of this method over [6,10,12,14] is the incorporation of a Bayesian framework and the usage of a data association algorithm. However, their data association algorithm is rather simple and, therefore, unable to deal with ambiguous situations (Requirement 2).…”