[Simulator] If the topography origin is moved, pedestrian trajectories deviate more than expected
Summary
If the origin is moved (i.e. to 500.000, 500.000) at some point the trajectories deviate in the range of 10^5 up to 10^0.
What is the current bug behavior?
Trajectories differ if the same simulation is run with an offset (0, 0) or (500.000, 500.000)
What is the expected correct behavior?
Differences should be smaller than 10^-9
Steps to reproduce
Compare the simulation output.
Starting point
See Issue #151 (closed)
Measures taken
-
New PedestrianOffsetPositionProcessor introduced to simulate the translation of the origin. This processor will take the offset and adds this to the positions of each pedestrian while the date is written.
-
The user must remove the offset prior to the simulation and add the offset to the processor attributes.
{
"attributesType" : "org.vadere.state.attributes.processor.AttributesOffsetPositionProcessor",
"attributes" : {
"offset" : {
"x" : 564280.0,
"y" : 5933391.0
}
}
}
output off trajectory.postvis if the PedestrianOffsetPositionProcessor writes to the same file
timeStep | pedestrianId | x | y | targetId | x_offset | y_offset |
---|---|---|---|---|---|---|
1 | 1 | 30.3375000436387 | 46.9230370317711 | 1 | 564310.337500044 | 5933437.92303703 |
1 | 2 | 28.7881592659345 | 47.346968671193 | 1 | 564308.788159266 | 5933438.34696867 |
1 | 3 | 27.8494240180758 | 47.7560581074064 | 1 | 564307.849424018 | 5933438.75605811 |
Possible Solutions
-
Allow the user to transform the topography but run the simulation with origin (0,0). This would need changes in the gui and for all processors which use word coordiantes
-
find the bug :)