AOSE

1. AOSE Advanced Software Engineering University College Dublin December 2007 Dr Rem Collier

2. Coordination ”The process by which an agent reasons about its local actions and the (anticipated) actions of others to try and ensure that the community acts in a coherent manner.” Nick Jennings,1996

3. Why Coordinate? • Preventing anarchy or chaos • Dependencies between agents’ actions. • Need to meet global constraints • No individual has sufficient competence, resources or information to solve the entire problem. • Efficiency

4. A1 A2 A3 How to Coordinate? • There are two basic ways of achieving this: • Task sharing: • When a problem isdecomposed intosubproblems andallocated to differentagents. • Result sharing: • When agents shareinformation relevantto their subproblems. Task 1 Task 1.1 Task 1.2 Task 1.3

5. Coordination Example • Consider an interaction between two robots, A and B, operating in a warehouse. • Each robot has been designed a different manufacturer. • They both have the capability to stack and unstack boxes that contain goods that have been stored in the building. • Because both robots work concurrently, they need to coordinate their actions to share the work load and to avoid knocking into each other and dropping the boxes.

6. Coordination Example • Task Sharing: • Our warehouse agents agree a distribution of work and then go about that work. • Robot A works on the east side of the building. • Robot B works on the west side. • Result Sharing: • Agents keep each other informed of their current activities and decide what to do independently. • Robot A decides to move box number A10456. • Robot A tells Robot B. • Using this additional information, Robot B independently decides to move another box A20987.

7. Task Sharing • The agent decomposes the task into a set of sub-tasks that are assigned dynamically to the agents at run-time. • Key Steps: • Problem decomposition for distribution • Synthesis of sub-problem results to obtain the solution • Optimisation of the problem-solving activities of the agents • Techniques for the coordination of the agents’ activities • Example Technique: Contract Net Protocol

8. Task Sharing Process 1. Problem decomposition 2. Subproblem solution 3. Answer synthesis Ref: Smith & Davis, 1980

9. The Contract Net Protocol I have a problem! manager Potential contrators announcement (b) Task Announcement (a) Recognising the problem manager manager Potential contrator Award task bids (c) Bidding (d) Award Contract

10. The Contract Net Protocol

11. A1 A2 A3 Result Sharing • Problem solving proceeds by agents cooperatively exchanging information as the solution is developed. • The set of tasks are pre-assigned at design time • Results may be shared: • proactively - one agent sends another agent some information because it believes that the other will be interested in it. • reactively – an agent sends information to another in response to a request.

12. Question • Traffic Management: • I have a community of agents that are monitoring the level of traffic on a road network. • Individual agents are responsible for pre-defined area of the road network. • When an agent detects a build up of traffic in its area, it contacts a pre-assigned traffic management agent. • Mobile Computing: • A personal agent (located on a users PDA) requests a map centred around the users coordinates that shows nearby hotels. • The map request is sent to a pre-assigned map-broker agent who then contacts an appropriate “map agent” to get the basic map and an appropriate “restaurant agent” to get a list of nearby restaurants and their coordinates. • Are these examples of coordination? If so, what type?

13. Coordination Techniques

14. Coordination Techniques • Organisational Structures • Meta-level Information Exchange • e.g. Partial Global Planning (PGP), (Durfee) • Multi-agent Planning • Norms and social laws • Coordination Models based on human teamwork: • Joint commitments (Jennings) • Mutual Modelling

15. Organizational Structures • A pattern of information and control relationships between individuals. • Responsible for shaping the types of interactions among the agents. • Aids coordination by specifying which actions an agent will undertake. • Organisational structures may be: • Functional • Spatial

16. Organizational Structure Models • A pattern for decision-making and communication among a set of agents who perform tasks in order to achieve goals. • Example: Automobile industry • Has a set of goals: To produce different lines of cars • Has a set of agents to perform the tasks: designers, engineers, salesmen Reference: Malone 1987

17. Product Manager 2 Product Manager I Salesman Designer Engineer Salesman Designer Engineer Product Hierarchy

18. Design Manager Sales Manager Engineering Manager Designers Salesmen Engineers Functional Hierarchy Product Manager (several products)

19. Design Manager Sales Manager Engineering Manager Designers Salesmen Engineers Centralised Market Product Manager 3 Product Manager 1 Product Manager 2 Functional Managers

20. Decentralised Market Product Manager 3 Product Manager 1 Product Manager 2 Designers Salesmen Engineers

21. Comparison of Organization Structures

22. Organizational Structures - Critique • Useful when there are master/slave relationships in the MAS. • Control over the slaves actions – mitigates against benefits of DAI such as reliability, concurrency. • Presumes that at least one agent has global overview – an unrealistic assumption in MAS. • Focused on closed systems.

23. From Structures to Patterns • (Dignum & Dignum, 2001) introduce the idea of an organisational pattern as a mechanism for defining organisational structures. • Identify key roles & responsibilities • Shift from closed systems to open systems • They introduce 3 basic patterns: • Hierarchy • Market • Network • Expand the notion of an organisational structures to

24. The Hierarchy Pattern • Goal: Coordinate the flow of resources or information by controlling and directing it via some (management) central point. • Each agent controls a statically defines sub-hierarchy (possibly empty). • Agents at lower levels in the hierarchy are dependent on agents at higher levels in the hierarchy. • Interactions are determined at design time, and are hardcoded into the system implementation. • Manager agents are responsible for controling, coordinating, and optimising sub-system activities. • Well suited to closed systems, this pattern has been used in information agents and management of communication networks.

25. The Hierarchy Pattern • Key Infrastructure Agent Roles: • Controller. • Monitor and orient the overall performance of the (sub-) system. • Interface Agents. • Responsible for communication between the system and the ‘outside world’.

26. The Market Pattern • Goal: Facilitate exchange between self-interested agents. • Agents are heterogenous and represent a diverse set of services and/or competencies. • Agents compete to perform tasks leading to their satisfaction of their own objectives. • Interaction occurs through communication and negotiation. • Negotiation rules are normally fixed (e.g. a particular auction protocol must be used). • Payoff is immediate. • Well suited to open-systems, this pattern has been used to model e-commerce scenarios and virtual enterprises.

27. The Market Pattern • Key Infrastructure Agent Roles: • Identification. • registration of society members. • Matchmaker. • keeps track of needs and services of agents in system. • mediates in matching of demand and supply of goods or services. • Banking. • define ways to exchange goods and determine profit and fairness of exchanges.

28. The Network Pattern • Goal: Facilitate collaboration between self-interested agents that have a mutual goal. • Agents are willing to trade freedom in exchange for guarantees regarding security and trustworthiness. • Services may be traded for “soft rewards” such as increase in prestige. • Relationships are dependent on clear communication patterns and social norms. • Coordination is achieves through mutual interest, possibly via trusted third parties. • Coordination is governed by well-defined rules and sanctions. • Well suited to open systems in which security and trust are essential.

29. The Network Pattern • Key Infrastructure Agent Roles: • Matchmaker • keeps track of needs and services of agents in system. • Mediates in matching of demand and supply of goods or services. • Gatekeeper • responsible for accepting and introducing new agents to the market. • informs new agents of the possibilities and capabilities of the market. • Notary • registers collaboration contracts between agents. • Monitoring Agents • Trusted third parties that keep track of the execution of collaboration contracts.

30. Coordination Techniques • Organisational Structures • Meta-level Information Exchange • e.g. Partial Global Planning (PGP), (Durfee) • Multi-agent Planning • Norms and social laws • Coordination Models based on human teamwork: • Joint commitments (Jennings) • Mutual Modelling

31. Meta-level Information Exchange • Exchange control level information about current priorities and focus. • Control level information • May change • Influence the decisions of agents • Does not specify which goals an agent will or will not consider. • Imprecise • Medium term – can only commit to goals for a limited amount of time.

32. Agenti Overlapping area Vehicle track Agentj Partial Global Planning (PGP) 1 • A DAI testbed – Distributed Vehicle Monitoring Testbed (DVMT) – used to successfully track a number of vehicles that pass within the range of a set of distributed sensors (agents). • Each agent monitors a dedicated area • There could be overlapping areas • Information is shared througha blackboard

33. Partial Global Planning (PGP) 2 • Main principle: cooperating agents exchange information in order to reach common conclusions about the problem solving process. • Why is planning partial? • The system does not generate a plan for the entire problem. • Why is planning global? • Agents form non-local plans by exchanging local plans and cooperating to achieve a non-local view of problem solving.

34. Partial Global Planning (PGP) 3 • Starts with the premise that tasks are inherently decomposed. • Assumes that an agent with a task to plan for might be unaware as to what tasks other agents might be planning for and how those tasks are related to its own. • No individual agent might be aware of the global tasks or states. • Purpose of coordination is to develop sufficient awareness.

35. Partial Global Planning (PGP) 4 • PGP involves 3 iterated stages: • Each agent decides what its own goals are and generates short-term plans in order to achieve them. • Agents exchange information to determine where plans and goals interact. • Agents alter local plans in order to better coordinate their own activities.

36. Partial Global Planning (PGP) 5 • Partial Global Plan: a cooperatively generated data structure containing the actions and interactions of a group of agents. • Contains: • Objective: the larger goal of the system. • Activitymap: what agents are actually doing and the results generated by the activities. • Solution construction graph: a representation of how the agents ought to interact in order to successfully generate a solution.

37. Coordination Techniques • Organisational Structures • Meta-level Information Exchange • e.g. Partial Global Planning (PGP), (Durfee) • Multi-agent Planning • Norms and social laws • Coordination Models based on human teamwork: • Joint commitments (Jennings) • Mutual Modelling

38. Multi-agent Planning 1 • Agents generate, exchange and synchronise explicit plans of actions to coordinate their joint activity. • They arrange apriori precisely which tasks each agent will take on. • Plans specify a sequence of actions for each agent. • It is a trade-off between specificity and reactive.

39. Multi-agent Planning 2 • Two basic approaches: • Centralised: plans of individual agents analysed by a central coordinator to identify interactions. • Distributed: a group of agents cooperate to form a: • Centralised plan • Distributed plan

40. Multi-agent Planning 3 • Distributed Planning for centralised plans: • Example: Air traffic control domain (Cammarata) • Aim: Enable each aircraft to maintain a flight plan that will maintain a safe distance with all aircrafts in its vicinity. • Each aircraft sends a central coordinator information about its intended actions. The coordinator builds a plan which specifies all of the agents’ actions including the ones that they should take to avoid collision.

41. Multi-agent Planning 5 • Critique: • Agents share and process a huge amount of information. • Requires more computing and communication resources. • Difference between multi-agent planning and PGP: • PGP does not require agents to reach mutual agreements before they start acting. • Sometimes Plans can also become obsolete very quickly. • i.e. Short life-span.

42. AOSE Negotiation

43. Negotiation • ”The process of several agents searching for an agreement”e.g. about price. • Reaching consensus ”Rules of Encouter” by Rosenchein and Zlotskin, 1994

44. auctioneer bidders Price bidder auctioneer Auctions • An Auction takes place between an auctioneer and a collection of bidders. • Goal is for the auctioneer to allocate the goods to one of the bidders. • In most settings, the auctioneer desires to maximise the price; bidders desire to minimise the price.

45. Auction Parameters

46. Price Bidder x Bidder 1 auctioneer English Auctions • English auctions are: • First price • Open cry • Ascending • Dominant strategy: successively bid a small amount more than the highest current bid until it reaches the valuation, then withdraw. • Susceptible to Winners curse • Winner is the one who overvalues the goods on offer and may end up paying more than its worth.

47. English Auctions

48. Price Bidder auctioneer auctioneer Dutch Auctions • Dutch auctions are: • Open cry • Descending • Auctioneer starts at an artificially high price. • Then continually lowers the offer price until an agent makes a bid which is equal to the current offer price. • Dominant strategy: None • Susceptible to Winners curse

49. Dutch Auctions

50. Bidders auctioneer First-price Sealed-bid Auctions • One shot auction • Single round, where bidders submit a sealed-bid for the good. • Good is awarded to agent that made the highest bid. • Winner pays price of highest bid. • Best strategy: bid less than true value.