Abstract This papers presents a software architecture for robot position estimator. The architecture is targeted at applications characterized by multiple, low resolution and erroneous sensory inputs and variability in possible sensor configurations. This may be the case, e.g., with a house robot equipped, for cost reasons, with several inexpensive sensors. The architecture is based on two blackboards. Specialist knowledge sources, which embody signal interpretation and other expertise, cooperatively construct position estimation hypotheses on the domain blackboard. Scheduling knowledge sources opportunistically control the estimating activity based on the data recorded on the control blackboard. In their operation, knowledge sources refer to a database containing customized descriptions of the premises in which the robot works. A prototype position estimator has been implemented on a testbed consisting of a mobile robot with several low cost sensors (Health Hero-1) and a high performance personal computer (IBM PC XT). The prototype position estimator software was written in GCLISP, a subset of Common Lisp. Initial position estimation results are presented and discussed.
[1]
Barbara Hayes-Roth,et al.
A Blackboard Architecture for Control
,
1985,
Artif. Intell..
[2]
Eswaran Subrahmanian,et al.
A rule-based blackboard kernel system: some principles in design
,
1984
.
[3]
James L. Crowley,et al.
Navigation for an intelligent mobile robot
,
1985,
IEEE J. Robotics Autom..
[4]
Donald Lee Mccracken,et al.
A production system version of the Hearsay-II speech understanding system
,
1981
.
[5]
Edward A. Feigenbaum,et al.
RULE-BASED UNDERSTANDING OF SIGNALS1
,
1978
.
[6]
Edward A. Feigenbaum,et al.
Rule-based understanding of signals
,
1977,
SGAR.
[7]
Robert Balzer,et al.
HEARSAY-II: A Domain-Independent Framework for Expert Systems
,
1980,
AAAI.
[8]
Victor R. Lesser,et al.
The Hearsay-II Speech-Understanding System: Integrating Knowledge to Resolve Uncertainty
,
1980,
CSUR.