  |
Engine Load at 500uu/m scale[2006.07.20] |
|
|
This time I'll compare engine loads between 250 and 500 uu/m scales. This report continues the previous "Engine Load at Different Scales". This is the new Yellow map scaled at 500 uu/m: 
DM-USAR_yellow500.zip
 |
Start Location A: an attempt to test long range scanning.
Start Location B: short range scan.
Start Location C: mid range scan with complex geometries.
You can download all the reports for the 500uu/m scale (Excel format): LoadGraphs_500.zip
Scale 250uu/m |
Scale 500uu/m |
|
 |
 |
|
 |
 |
|
Map without robots (250uu/m) |
Map without robots (500uu/m) |
|
 |
 |
|
Map with P2DX (250uu/m) |
Map with P2DX (500uu/m) |
I controlled everything several times, I even tried to compile USARSim with different Karma settings, but the graph was always the same. P2DX in the 500 uu/m map, at locations A seems 60% ~ 70% faster... (is it really possible?)
Scale 250uu/m |
Scale 500uu/m |
|
 |
 |
|
 |
 |
|
Map without robots (250uu/m) |
Map without robots (500uu/m) |
|
 |
 |
|
Map with P2DX (250uu/m) |
Map with P2DX (500uu/m) |
Unreal engine is really optimized for big maps...
Scale 250uu/m |
Scale500uu/m |
|
 |
 |
|
 |
 |
|
Map without robots (250uu/m) |
Map without robots (500uu/m) |
|
 |
 |
|
Map with P2DX (250uu/m) |
Map with P2DX (500uu/m) |
It seems that bigger is better :-) Even if 3 experiments in the same map using only 1 robot are not a definitive proof, we now know that the 500 uu/m scale could be even more efficient than 250 uu/m. Considering that a 500 uu/m scale "restricts" the world to 1 km x 1 km I think that this is the maximum scale we can adopt.