Adaptive Constraint-Based Agents in Artificial Environments
[ Please note: The project has been discontinued as of May 31, 2005 and is superseded by the projects of the ii Labs. There won't be further updates to these pages. ]
A State Resource Constraint with a Symbolic State Domain
This section describes a simple version of a State Resource Constraint
with a symbolic state domain. More specifically, the constraint has the following
- The SRC has a finite number of possible states S.
- The TemporalReference variables of connected Precondition
Tasks have a domain of integers from 0 to a planning horizon
h, and the Precondition Tasks' State variables have
a domain of S.
- The TemporalReference variables of connected State Tasks
have a domain of integers from 1 to h.
- The SRC has a finite number of possible events that can cause a
- The SRC's synergy mapping is a partial function which maps a set
of State Tasks' Contributions to an event.
- The SRC's state transition is a partial function which maps a
state and an event to a successor state.
- The state is projected on a timeline of discrete time points from 0
to h according to the connected State Tasks and the initial
value of the Current State. If a Sensor Constraint is
connected to the SRC's Object, the state at the Current
Time c is set to the Current State, regardless of whether
there is a valid state transition or not.
- For the state projection, the present state is preserved if it is
not possible to derive an event from the set of all State Task
Contributions at that time point by the synergy mapping, or if no
state transition is possible from the SRC's present state for the
- If the state required by a connected Precondition Task is
not equal to the SRC's actual state at this time point, the
SRC's costs are increased by the minimal number of state transitions
required to transform the actual state to the required state.
The constraint's internal structures and heuristics are detailed in the following
For questions, comments or suggestions, please contact us.
May 20, 2001 by Alexander Nareyek