AGV Simulation

1
AGV Simulation

• Tutorial
• Nelis Boucké
• Alexander Helleboogh

2
Overview

• Introduction
• Architecture
• Installation
• Use
• Extensions

3
Simulation is crucial for MAS applications

• AGVs (Automated Guided Vehicles)
• Transport loads in warehouse
• Complex network of road segments and crossroads
• Advantage of using MAS
• Decentralized system behavior
• Robust flexible
• In dynamic unpredictable environments
• Test before deployment!
• Decentralized systems
• No risk of damaging AGVs
• Challenge for simulation
• Test robustness flexibility
• Dynamic test scenarios needed!

4
How do we test?
Agent software
Deploy
Test
Simulated environment
Warehouse environment
5
Overview

• Introduction
• Architecture
• Installation
• Use
• Extensions

6
Architecture
XML
Java VM
7
AGV batch simulator
Deliberation Time Model
Interface
Interface
Interface
Simulated Environment
8
Programming AGV Behavior
Interface

• Class
• AGVRandomWalkBehavior
• Specify your behavior here!
• Package agentwise.agvsimulation.behaviordemo
• Classes you can use
• AGVBehavior
• The complete set of instructions to steer the AGV
  robot
• package agentwise.agvsimulation.agv

9
Scenario decisions of agents
A
B
Influences
10
Scenario decisons result in activities
A
B
11
Activities interfere!
A
B
Intended outcome
Actual outcome
12
What does this imply?

• Specification of AGV programming interface
• No post conditions in terms of what will happen
• All kinds of catastrophes can happen
• Not indicated by exceptions
• Only specifies what an agent tries to do
• Thinking versus acting
• Agents can think and act in parallel
• Agent have to wait explicitly while their AGV is
  driving
• waitsecond()
• Agents can send messages while their AGV is
  driving
• Agents are not notified of anything
• have to rely on polling their sensors