Your company name Multi-Agent System (MAS) for Manufacturing Students: Anand Prajapati (2011ME20764) Chilwant Nikhil (2011ME20769) B. Tech Project (MED421) Final presentation Supervisor: Prof. M. S. Kulkarni
Your company name
Multi-Agent System (MAS) for Manufacturing
Students:Anand Prajapati (2011ME20764)Chilwant Nikhil (2011ME20769)
B. Tech Project (MED421)Final presentation
Supervisor:Prof. M. S. Kulkarni
Your company name
Introduction
Machine
Local Scheduling
Agent
Local Maintenanc
e Agent
Global Scheduling
Agent
CustomerAgent
“An agent is a computer system that is situated in some environment, and that is capable of autonomous action in this environment in order to meet its design objectives. “ Source: Wooldridge, Michael. An introduction to multi-agent systems. John Wiley & Sons, 2009.
Your company name Your Logo
Objectives & Deliverable• Objectives:
– Provide BDI architecture– Provide blackboard architecture. – Improve scheduling algorithm – Devise an rescheduling architecture
• Deliverable:– A generic library for MAS with robust architecture &
advanced algorithms
Your company name Your Logo
Overview• Literature review
– Identifying research gap– BDI architecture– Blackboard
• Implementation– BDI architecture– Blackboard architecture
• Scheduling algorithm• Rescheduling architecture
Your company name Your Logo
Literature Review
BDI architecture
WHY BDI ? Robust in dealing with
complex and dynamic environments i.e. goal oriented agents
Easy to configure
Overview of BDI architecturePan, Yen-Tsung, and Men-Shen Tsai. "Development a BDI-based intelligent agent architecture for distribution systems restoration planning." Intelligent System Applications to Power Systems, 2009. ISAP'09. 15th International Conference on. IEEE, 2009.
PLAN LIBRARY
INTERPRETER( )REASONER
Goals)Desires(
TasksIntentions)(
MONITOR COMMAND GENERATOR
ENVIRONMENT
DATABASE(BELIEFS)
Belief : knowledge about environment
Desire : objective to be achieved
Intention : Action needs to be takenPlan : way of doing action
Your company name Your Logo
Work done
Implementation of BDI architecture
Earlier agent would poll continuously Now,Actions are triggered by events
Customer
Global Scheduling Agent (GSA)
Job received
add ‘raise bid’ goal
Raise bid plan
‘Send job to GSA’ Goal
Completed conversion of maintenance agent
Your company name Your Logo
Literature Review
Blackboard architecture
Acts as a • Communication centre• Central database
Your company name Your Logo
Blackboard architecture
Your company name Your Logo
• Developed a blackboard agent– Registry of agents– Re-routing of messages
• In every agent:– Subscribe/unsubscribe from blackboard registry– Register for parameter updates
• Now, any no. of machines can connect to MAS
Work done
Implementation of Blackboard architecture
Your company name Your Logo
Work done
Working of MAS
Blackboard
GlobalScheduling
Agent
LocalScheduling
Agent
LocalMaintenance
Agent
MachineAgent
Customersubscribe
subscribe subscribe
subs
crib
e
subscribe
All agents subscribe to blackboard
Your company name Your Logo
Blackboard
GlobalScheduling
Agent
LocalScheduling
Agent
LocalMaintenance
Agent
MachineAgent
Customer
Order
Work done
Working of MAS
Blackboard as message centre
Your company name Your Logo
• Don’t delay jobs beyond a certain regret value
Low Regret• High Regret• No more delay
Time
Regret ThresholdDue Date
Work done
Branch & Bound
Your company name Your Logo
Future Plan• Implementation of the devised rescheduling
architecture– Architecture has been formulated.
• Learning mechanism to adapt to dynamic environments.