Closed tlb9551 closed 4 years ago
Hello~ I found a small bug in random_forest_sensing.cpp https://github.com/HKUST-Aerial-Robotics/Btraj/blob/5686dededa1cae7e46d0e3969dd9c6f60837c641/src/random_forest_sensing.cpp#L67 Here x,y is added by 0.5 * resolution, so x,y should not add 0.5 * resolution again in line 77 again. https://github.com/HKUST-Aerial-Robotics/Btraj/blob/5686dededa1cae7e46d0e3969dd9c6f60837c641/src/random_forest_sensing.cpp#L77 This will cause the coordinates of the cloud points to be a multiple of resolution, due to imprecision of double data type, there will be some gaps in the collision map built by the point cloud. I found this bug when I set the margin coefficient to 0, some paths pass through the middle of the obstacle.
@tlb9551 Thanks for pointing out this bug.
Hello~ I found a small bug in random_forest_sensing.cpp
https://github.com/HKUST-Aerial-Robotics/Btraj/blob/5686dededa1cae7e46d0e3969dd9c6f60837c641/src/random_forest_sensing.cpp#L67 Here x,y is added by 0.5 * resolution, so x,y should not add 0.5 * resolution again in line 77 again. https://github.com/HKUST-Aerial-Robotics/Btraj/blob/5686dededa1cae7e46d0e3969dd9c6f60837c641/src/random_forest_sensing.cpp#L77 This will cause the coordinates of the cloud points to be a multiple of resolution, due to imprecision of double data type, there will be some gaps in the collision map built by the point cloud. I found this bug when I set the margin coefficient to 0, some paths pass through the middle of the obstacle.