Title: Hardware Configuration Diagram
1Evaluation of Software Agent Performance while
controlling the temperature of a test
chamber by
David Hooley for CIS447, Artificial
Intelligence, Fall 2004 Kutztown
University Professor Mark Jones December 7,
2004
2Goal Test the Proposition More sophisticated
agents perform better than simpler agents
- Assemble hardware and software to test the
proposition - Develop several agents
- Simple Reflex Agent
- (Model or Utility) Agent
- Others (future work an expert system?)
3Hardware Configuration Diagram
Notebook Computer Red Hat V9 Linux OneWire Driver
Software Agent Performance Application
OneWire Hardware Interface Sensor Switch
Interface Power Drivers
Temperature Chamber Enclosure Fan for
Cooling Light Bulb for Heating Temperature Sensor
Temp Sensor
Temp Sensor Interface
Temp Sensor
OneWire Interface
Switch Interface
Cooling Fan
Heating Lamp
Power Controller
4Software Architecture
- Modular
- Sequence executed 40 times
- Data Acquisition
- Decisions by Agent
- Control
- Performance Analysis Report
- Communicate via files
5Software Configuration Diagram
Main Program
Internet
Run Initialization
Java OneWire Server
Java Client Read Temp
Data Acquisition
Agent
OneWire Interface
Java Client Control Switch
Actuator Control
Done? yes
no
Performance Analysis
Evaluation Report
6Agents
- Simple Reflex Agent (1)
- Compare Temperature to set point
- If cooler, turn heater on
- If warmer, turn fan on
- (Model or Utility) Agent (2)
- Add memory, calculate changes
- Try to avoid overshoot close to set point
7Performance Analysis Examples
- Simple Reflex Agent (Agent 1)
-
- - pow err sum
- clog-1204-2057nf 451 400 851 1 7w no int
fan clog-1204-2046wf 465 525 990 1 7w w/int
fan -
- (Model or Utility) Agent (Agent 2)
-
- clog-1204-2120nf 406 470 876 2 7w no int
fan - clog-1204-2034wf 466 580 946 2 7w w/int fan
8Project Results
- Proposition was NOT demonstrated
- Little difference in agent performance noted
- Perhaps overshoot would be seen if more power
9Future Work
- Higher power heating and cooling
- Graphical data displays
- Automate data runs
- Implement more/better agents
- Different models
- Perhaps an expert system?
10 Run Data Summary
Evaluation of Software Agent Performance date
time name pow err agent comments 041204
0931 clog-1204-0931wf 481 580 1 7w w/int
fan rm temp 25 041204 0944 clog-1204-0944bad 55
6 775 1 7w no int fan, bad run 041204
0959 clog-1204-0959nf 450 460 1 7w no int
fan 041204 1033 clog-1204-1033nf 436 440 1
7w no int fan 041204 1058 clog-1204-1058nf 451
460 1 7w no int fan 041204
1112 clog-1204-1112nf 466 485 1 7w no int
fan 041204 1126 clog-1204-1126wf 466 565 1
7w w/int fan hi rm temp 041204
1148 clog-1204-1148wf 466 555 1 7w w/int
fan hi rm temp 041204 1246 clog-1204-1246wf 541
655 1 7w w/int fan 041204 1257 clog-1204-1257
bad 511 715 1 7w w/int fan battery ran
down 041204 1916 clog-1204-1916nf 345 385 2
7w no int fan rmtemp 225 041204
1926 clog-1204-1926nf 465 425 1 7w no int
fan rmtemp 225 041204 1956 clog-1204-1956wf 437
565 2 7w w/int fan rmtemp 230 041204 2009
clog-1204-2009wf 421 490 1 7w w/int fan
rmtemp 235 041204 2021 clog-1204-2021wf 451
605 2 7w w/int fan rmtemp 230 041204 2034
clog-1204-2034wf 466 580 2 7w w/int
fan rmtemp 230 041204 2046
clog-1204-2046wf 465 525 1 7w w/int fan
rmtemp 230 041204 2057 clog-1204-2057nf 451
400 1 7w no int fan rmtemp 220 041204
2120 clog-1204-2120nf 406 470 2 7w no
int fan rmtemp 225
11(No Transcript)