Y value fix:
Previously we discovered that the Y value of CARLA objects are wrong.
We incorrectly assumed that the hitpoints report absolute values. hitpoints from sensor is probably correct as they are in their own frame (small values, smaller the distance).
However, the object's locations and sensor's locations all report in negative Y value. Therefore, it is the best to do all the calculations as is and only change the Y value before reporting the detected objects to the caller of the sensor.
PR Details
Description
Y value fix: Previously we discovered that the Y value of CARLA objects are wrong. We incorrectly assumed that the hitpoints report absolute values. hitpoints from sensor is probably correct as they are in their own frame (small values, smaller the distance). However, the object's locations and sensor's locations all report in negative Y value. Therefore, it is the best to do all the calculations as is and only change the Y value before reporting the detected objects to the caller of the sensor.
Also for complete occlusion-fix, this is accompanied by this PR: https://github.com/usdot-fhwa-stol/carma-carla-integration/pull/54/files
More Data: Data about actors in the scene where CAR is actually lower Y value in correct map, but the data returned from CARLA is inverted:
Sensor data returned from the sensor has also inverted Y:
I am not sure how the internal CARLA logic works, but the Y in lidar hitpoint is also inverted as shown above which should report positive Y values.
Related GitHub Issue
Y value fix: Fix for PR: https://github.com/usdot-fhwa-stol/carma-utils/pull/191 Also fixes occlusion: https://github.com/usdot-fhwa-stol/carma-utils/issues/194
Related Jira Key
Closes: https://github.com/usdot-fhwa-stol/carma-utils/issues/190
Closes CDAR-534
Motivation and Context
How Has This Been Tested?
CDASIM integration tested, platform reports y values in positive
Types of changes
Checklist: