In this abstract we give a brief overview of the PRACTIONIST framework, which supports programmers in developing BDI agents and is built on top of JADE [1], a widespread platform that implements the FIPA1 specifications. Therefore, our agents are deployed within JADE containers and their main cycle is implemented by means of a JADE cyclic behaviour (figure 2).
A PRACTIONIST agent is a software component endowed with the following elements (figure 2): • a set of perceptions and the corresponding perceptors that listen to some relevant external stimuli; • a set of beliefs representing the information the agent has got about both its internal state and the external environment; • a set of goals the agent wishes or wants to pursue. They represent some states of affairs to bring about or activities to perform and will be related to either its desires or intentions (see below); • a set of goal relations the agent uses during the deliberation process and means-ends reasoning; • a set of plans that are the means to achieve its intentions; • a set of actions the agent can perform to act over its environment; and • a set of effectors that actually execute the actions.
As shown in figure 2, PRACTIONIST agents are structured in two main layers: the framework defines the execution logic and provides the built-in components according to such a logic, while the top layer includes the specific agent components to be implemented, in order to satisfy system requirements.
Fig. 2. Components of PRACTIONIST agents.
Therefore, a developer who wants to design an agent has to
develop (i) the Goals the agent could pursue, (ii) the means
(a set of plans, i.e. the Plan Library) to pursue such goals or
to react to the stimuli coming from the environment, (iii) the
Perceptors to receive such stimuli, (iv) the Actions the agent
could perform and the corresponding Effectors, and (v) the set
of beliefs and rules (Belief Base) to model the information
about both its internal state and the external world (details on
beliefs are given in [
In the following section we give an overview of how to
program some of agent components, with reference to the paper
”Reasoning about Goals in BDI Agents: the PRACTIONIST
Framework”, presented at the WOA 2006 [
The concepts and the examples given in this section refer to the tileworld demonstrator, which is a multi agent system with two types of agents, i.e. an agent that manages the environment and player agents.
Several simulation parameters can be altered at run time, such as the appear rate and the life cycle of holes, tiles and obstacles. These information was represented by beliefs 2 about the state of the environment represented through the following predicates: • gridSize(width: X, height: Y) represents the size of the grid, in terms of width and height, • holeBirth(rate: X) and holeLifecycle(rate: X) represent the frequency of holes’ birth and their mean life cycle, • tileBirth(rate: X) and tileLifecycle(rate: X) represent the frequency of tiles’ birth and their mean life cycle, • obstacleBirth(rate: X) and obstacleLifecycle(rate: X) represent the frequency of obstacles’ birth and their mean life cycle, • agent(name: X) represents other active player agents.
The framework provides the support to let agent make meta-level reasoning. In other words, each player agent, by reasoning on above information, will be able to select the optimal strategy to increase its score. For example, the plan FindTileInAmplitudePlan implements a depth search behavior, while the plan FindTileRandomicallyPlan implements a random search strategy. Thus these plans are used by the player to find a tile in several circumstances.
Analogously, agent beliefs about its state refer to the following predicates: • position(xPos: X, yPos:Y) represents the position of the player agent, • score(value: X) represents the current score of the player, • hold(obj: tile) states that the player agent holds a tile.
On the base of such beliefs, some goals are defined as well. As an example, the HoldTile is a state goal that succeeds when hold(obj: tile) is believed true by the agent for the same tile. Thus, in the AchieveTilePlan, the player agent has to identify a tile within the grid to satisfy the HoldTile goal and then hold such a tile by executing the action of picking it up.
The player agent is endowed with the Taker effector, which triggers and executes the pick up action and updates the environment status and its internal state. The agent is also provided with other effectors (e.g. Mover, Releaser, etc.) to be able to perform other actions, such as moving itself in the grid and releasing holding tiles.
Finally, the cognitive system of the agent includes a set of perceptors that receive stimuli from the environment. As an example, the player agent is equipped with the perceptors TileLifeCyclePerceptor, HoleBirthPerceptor, etc. to be able to perceive changes from the environment about tiles’ birth rate, the obstacles’ life cycle, and so forth.
III. PRACTIONIST AGENT INTROSPECTION TOOL (PAIT)
The framework also provides developers with the PRACTIONIST Agent Introspection Tool (PAIT), a visual integrated 2In PRACTIONIST beliefs can be about either predicates or other beliefs (expressed by the operator Bel). Moreover, predicates can be expressed by specifying the role of their arguments, i.e. predicate(role1 : element1, role2 : element2, ..., roleN : elementN ).
Fig. 3. The PRACTIONIST Agent Introspection Tool (PAIT). monitoring and debugging tool, which supports the analysis of the agent’s state during its execution. In particular, the PAIT can be suitable to display, test and debug the agents’ relevant entities and execution flow. Each of these components can be observed at run-time through a set of specific tabs (see figure 3); the content of each tab can be also displayed in an independent window.
All the information showed at run-time could be saved
in a file, providing the programmer with the opportunity of
performing an off-line analysis. Moreover, the PAIT provides
an area for log messages inserted in the agent source code,
according to the Log4j approach. The usage of this console
and the advantages it provides are described in more details
in [