Design of Multi-Agent Systems

Design of Multi-Agent Systems • Teacher • Bart Verheij • Student assistants • Albert Hankel • Elske van der Vaart • Web site • http://www.ai.rug.nl/~verheij/teaching/dmas/ • (Nestor contains a link)

Overview • Deductive reasoning agents • Planning • Agent-oriented programming • Concurrent MetateM • Practical reasoning agents • Practical reasoning & intentions • Implementation: deliberation • Implementation: commitment strategies • Implementation: intention reconsideration

Deductive Reasoning Agents • Decide what to do on the basis of a theory stating the best action to perform in any given situation • ,  |– Do(a) with a  Ac • where •  is such a theory (typically a set of rules) •  is a logical database that describes the current state of the world • Ac is the set of actions the agent can perform

Deductive Reasoning Agents • But: • Theorem proving is in general neither fast nor efficient • Calculative rationality (rationality with respect to the moment calculation started) requires a static environment • Encoding of perception & environment into logical symbols isn’t straightforward So: • Use a weaker logic • Use a symbolic, not logic-based representation

Planning: STRIPS • Only atoms and their negation • Only represent changes Blocks world (blocks + a robot arm) • Stack(x,y) • Pre {Clear(y), Holding(x)} • Del {Clear(y), Holding(x)} • Add {ArmEmpty(y), On(x,y)

Problems with planning Frame problem Describe what does not change by an action Qualification problem Describe all preconditions of an action Ramification problem Describe all consequences of an action Prediction problem Describe the duration that something remains true

Agent-oriented programming • Agent0 (Shoham) • Key idea: directly programming agents in terms of intentional notions like belief, commitment, and intention • In other words, the intentional stance is used as an abstraction tool for programming!

Agent-oriented programming • Shoham suggested that a complete AOP system will have 3 components: • a logic for specifying agents and describing their mental states • an interpreted programming language for programming agents (example: Agent0) • an ‘agentification’ process, for converting ‘neutral applications’ (e.g., databases) into agents

Agent-oriented programming • Agents in Agent0 have four components: • a set of capabilities (things the agent can do) • a set of initial beliefs • a set of initial commitments (things the agent will do) • a set of commitment rules

Agent-oriented programming • Each commitment rule contains • a message condition • a mental condition • an action • On each ‘agent cycle’… • The message condition is matched against the messages the agent has received • The mental condition is matched against the beliefs of the agent • If the rule fires, then the agent becomes committed to the action (the action gets added to the agent’s commitment set)

A commitment rule in Agent0 • COMMIT( • ( agent, REQUEST, DO(time, action) • ), ;;; msg condition • ( B, • [now, Friend agent] AND • CAN(self, action) AND • NOT [time, CMT(self, anyaction)] • ), ;;; mental condition • self, • DO(time, action)

A commitment rule in Agent0 • Meaning: • If I receive a message from agent which requests me to do action at time, and I believe that: • agent is currently a friend • I can do the action • At time, I am not committed to doing any other action then I commit to doing action at time

Concurrent METATEM • Concurrent METATEM is a multi-agent language in which each agent is programmed by giving it a temporal logic specification of the behavior it should exhibit • These specifications are executed directly in order to generate the behavior of the agent • Temporal logic is classical logic augmented by modal operators for describing how the truth of propositions changes over time

Concurrent METATEM • important(agents)it is now, and will always be true that agents are important • important(ConcurrentMetateM)sometime in the future, ConcurrentMetateM will be important • important(Prolog)sometime in the past it was true that Prolog was important • (friends(us)) U apologize(you)we are not friends until you apologize • apologize(you)tomorrow (in the next state), you apologize

Concurrent METATEM • MetateM is a framework for directly executing temporal logic specifications • The root of the MetateM concept is Gabbay’s separation theorem:Any arbitrary temporal logic formula can be rewritten in a logically equivalent past  future form. • This past  future form can be used as execution rules

Concurrent METATEM • A MetateM program is a set of such rules • Execution proceeds by a process of continually matching rules against a “history”, and firing those rules whose antecedents are satisfied • The instantiated future-time consequents become commitments which must subsequently be satisfied

Concurrent METATEM • Execution is thus a process of iteratively generating a model for the formula made up of the program rules • The future-time parts of instantiated rules represent constraints on this model All ‘asks’ at some time in the past are followed by a ‘give’ at some time in the future

Concurrent METATEM • Execution is thus a process of iteratively generating a model for the formula made up of the program rules • The future-time parts of instantiated rules represent constraints on this model • ConcurrentMetateM provides an operational framework through which societies of MetateM processes can operate and communicate

Practical reasoning • Practical reasoning is reasoning directed towards actions — the process of figuring out what to do: “Practical reasoning is a matter of weighing conflicting considerations for and against competing options, where the relevant considerations are provided by what the agent desires/values/cares about and what the agent believes.” (Bratman) • Practical reasoning is distinguished from theoretical reasoning – theoretical reasoning is directed towards beliefs

Practical reasoning • Human practical reasoning consists of two activities: • deliberationdeciding whatstate of affairs we want to achieve • means-ends reasoningdeciding howto achieve these states of affairs • The outputs of deliberation are intentions

Intentions in practical reasoning • Intentions pose problems for agents, who need to determine ways of achieving them.If I have an intention to , you would expect me to devote resources to deciding how to bring about . • Intentions provide a “filter” for adopting other intentions, which must not conflict.If I have an intention to , you would not expect me to adopt an intention  such that  and  are mutually exclusive. • Agents track the success of their intentions, and are inclined to try again if their attempts fail.If an agent’s first attempt to achieve  fails, then all other things being equal, it will try an alternative plan to achieve .

Intentions in practical reasoning • Agents believe their intentions are possible.That is, they believe there is at least some way that the intentions could be brought about. Otherwise: intention-belief inconsistency • Agents do not believe they will not bring about their intentions.It would not be rational of me to adopt an intention to  if I believed  was not possible. Otherwise: intention-belief incompleteness • Under certain circumstances, agents believe they will bring about their intentions.It would not normally be rational of me to believe that I would bring my intentions about; intentions can fail. Moreover, it does not make sense that if I believe  is inevitable that I would adopt it as an intention.

Intentions in practical reasoning • Agents need not intend all the expected side effects of their intentions.If I believe  and I intend that , I do not necessarily intend  also. (Intentions are not closed under implication.)This last problem is known as the side effect or package deal problem.

Intentions in practical reasoning • Intentions are stronger than mere desires: • “My desire to play basketball this afternoon is merely a potential influencer of my conduct this afternoon. It must vie with my other relevant desires [. . . ] before it is settled what I will do. In contrast, once I intend to play basketball this afternoon, the matter is settled: I normally need not continue to weigh the pros and cons. When the afternoon arrives, I will normally just proceed to execute my intentions.” (Bratman, 1990)

Practical reasoning (abstract) • Current beliefs and perception determine next beliefs: • Current beliefs and intentions determine next desires: • Current beliefs, desires and intentions determine next intentions: • Current beliefs, desires and available actions determine a plan:

Implementing practical reasoning agents • B := B_initial; • I := I_initial; • loop • p := see; • B := brf(B,p); //Update world model • I := deliberate(B) •  := plan(B,I); //Use means-end reasoning • execute(); • end;

Interaction between deliberation and planning • Both deliberation and planning take time, perhaps too much time. • Even if deliberation is optimal (maximizes expected utility), the resulting intention may no longer be optimal when deliberation has finished. • (Calculative rationality)

Deliberation • How does an agent deliberate? • Option generationin which the agent generates a set of possible alternatives • Filteringin which the agent chooses between competing alternatives, and commits to achieving them.

Implementing practical reasoning agents • B := B_initial; • I := I_initial; • loop • p := see; • B := brf(B,p); • D := option(B,I); // Deliberate (1) • I := filter(B,D,I) // Deliberate (2) •  := plan(B,I); • execute(); • end;

Commitment Strategies • The following commitment strategies are commonly discussed in the literature of rational agents: • Blind commitmentA blindly committed agent will continue to maintain an intention until it believes the intention has actually been achieved. Blind commitment is also sometimes referred to as fanatical commitment. • Single-minded commitmentA single-minded agent will continue to maintain an intention until it believes that either the intention has been achieved, or else that it is no longer possible to achieve the intention. • Open-minded commitmentAn open-minded agent will maintain an intention as long as it is still believed possible.

Commitment Strategies • An agent has commitment both to ends(i.e.,the wishes to bring about), and means(i.e., the mechanism via which the agent wishes to achieve the state of affairs) • Currently, our agent control loop is overcommitted, both to means and ends • Modification: replanif ever a plan goes wrong

B := B_initial; • I := I_initial; • loop • p := see; • B := brf(B,p); • D := option(B,I); • I := filter(B,D,I) •  := plan(B,I); • while not empty() do • a := head(); • execute(a); //Start plan execution •  := tail(); • p := see; • B := brf(B,p); //Update world plan • if not sound(,B,I) then •  := plan(B,I); //Replan if necessary • end; • end; • end;

Commitment Strategies • Still overcommitted to intentions: Never stops to consider whether or not its intentions are appropriate • Modification: stop to determine whether intentions have succeeded or whether they are impossible:(Single-minded commitment)

B := B_initial; • I := I_initial; • loop • p := see; • B := brf(B,p); • D := option(B,I); • I := filter(B,D,I) •  := plan(B,I); • while not (empty() or succeeded(B,I) or impossible (B,I) do • a := head(); //Check whether intentions succeeded • execute(a); //and are still possible •  := tail(); • p := see; • B := brf(B,p); • if not sound(,B,I) then •  := plan(B,I); • end; • end; • end;

Intention Reconsideration • Our agent gets to reconsider its intentions once every time around the outer control loop, i.e., when: • it has completely executed a plan to achieve its current intentions; or • it believes it has achieved its current intentions; or • it believes its current intentions are no longer possible. • This is limited in the way that it permits an agent to reconsider its intentions • Modification: Reconsider intentions after executing every action

B := B_initial; • I := I_initial; • loop • p := see; • B := brf(B,p); • D := option(B,I); • I := filter(B,D,I) •  := plan(B,I); • while not (empty() or succeeded(B,I) or impossible (B,I) do • a := head(); • execute(a); •  := tail(); • p := see; • B := brf(B,p); • D := option(B,I); //Reconsider (1) • I := filter(B,D,I); //Reconsider (2) • if not sound(,B,I) then •  := plan(B,I); • end; • end; • end;

Intention Reconsideration • But intention reconsideration is costly!A dilemma: • an agent that does not stop to reconsider its intentions sufficiently often will continue attempting to achieve its intentions even after it is clear that they cannot be achieved, or that there is no longer any reason for achieving them • an agent that constantly reconsiders its attentions may spend insufficient time actually working to achieve them, and hence runs the risk of never actually achieving them • Solution: incorporate an explicit meta-level control component, that decides whether or not to reconsider

B := B_initial; • I := I_initial; • loop • p := see; • B := brf(B,p); • D := option(B,I); • I := filter(B,D,I) •  := plan(B,I); • while not (empty() or succeeded(B,I) or impossible (B,I) do • a := head(); • execute(a); •  := tail(); • p := see; • B := brf(B,p); • if reconsider(B,I) then //Decide whether to reconsider or not • D := option(B,I); • I := filter(B,D,I); • end; • if not sound(,B,I) then •  := plan(B,I); • end; • end; • end;

Design of Multi-Agent Systems