Information Technology Reference
In-Depth Information
[15]; decision-theoretic planning is initiated with
solve
(
p,h
)
,where
p
is a G
OLOG
pro-
gram and
h
is the MDP's solution horizon). Two important constructs used in this re-
gard are the non-deterministic choice of actions (
a
b
) and arguments (
pickBest
(
v,l,p
)
),
where
v
is a variable,
l
is a list of values to choose from, and
p
is a G
OLOG
program.
Then each occurrence of
v
is replaced with the value chosen. For details we refer to [9].
|
2.2
Related Work
We want to build on the theory of
speech acts
as introduced by Austin [1] and Searle [19].
Based on these works, Cohen and Levesque [5] already investigated a formal theory of
rational interaction. We restrict ourselves to command interpretation and do not aim for
a full-fledged dialogue system. Nevertheless, we follow their formal theory of interpre-
tation and we carry out our work in the context of the situation calculus.
The use of definite clause grammars for parsing and interpreting natural language
has already been shown in [2]. Despite being relatively ad hoc and the fact that the
small grammar only covered a constrained subset of English, their system provided a
wide spectrum of communication behaviours. However, in contrast to their approach we
want to account for incomplete and unclear utterances both by using a larger grammar
as well as adding interpretation mechanisms to the system.
[10] developed a system on a robot platform that manages dialogues between human
and robot. Similar to our approach, input to the system is processed by task planning.
However, queries are limited to questions that can either be answered with yes or no or
a decimal value. A more advanced system combining natural language processing and
flexible dialogue management is reported on in [4]. User utterances are interpreted as
communicative acts having a certain number of parameters. The approach is missing
a proper conceptual foundation of objects and actions, though. This makes it hard to
adapt it to different platforms or changing sets of robot capabilities.
[11], on the other hand, built a dialogue management system well-founded by mak-
ing use of a concept hierarchy formalised in Description Logics (DL). Both, the linguis-
tic knowledge as well as the dialogue management are formalised in DL. This is a very
generic method for linking lexical semantics with domain pragmatics. However, this
comes with the computational burden of integrating description logics and appropriate
reasoning mechanisms. We want to stay within our current representational framework,
that is, the situation calculus and Golog, and we opt to exploit the capabilities to reduce
computational complexity with combining programming and planning.
3
Method and Approach
As mentioned before, we cast the language processing of spoken commands on a do-
mestic service robot as an interpretation process. We decompose this process into the
following steps. First, the acoustic utterance of the user is being transformed into text
via a speech recognition component which is not part of this paper's contribution. The
transcribed utterance is then passed on for syntactic analysis by a grammar. After that,
the interpretation starts, possibly resolving ambiguities and generating intermediate re-
sponses. If the utterance could be interpreted successfully, it is executed, otherwise it is
being rejected. We will now present the individual steps in more detail.
