search

praveen-palanisamy / multiple-object-tracking-lidar

C++ implementation to Detect, track and classify multiple objects using LIDAR scans or point cloud
MIT License
796 stars 229 forks source link

VLP 32C Data Interpertation #35

Closed SHamblin closed 4 years ago

SHamblin commented 4 years ago

Thanks for the work you've put into the project! I've got a question about using it with a VLP 32C. I'm attempting to detect a car in a parking lot and the clusters that are being picked up don't seem to match what I would expect.

I'm playing back the data from a bag file and remapping to the filtered_cloud topic. In the shots of Rviz bellow you can see the input data and then in white the detected clusters.

Is there a variable that I should adjust? <ec.setClusterTolerance (0.08); > Seems like a promising line to adjust but I'm unsure if I should increase or lower it.

2020-09-09 21_13_59-Window 2020-09-09 21_15_01-Window

Any help would be appreciated, Thank you!

praveen-palanisamy commented 4 years ago

Hi @SHamblin , In your use-case, is the car that you are intending to detect & track stationary or moving (w.r.t the VLP 32C)? For this tracker to work without modifications, the point cloud input is expected to be "filtered" before publishing onto the filtered_cloud topic. You could pre-process your LiDAR point clouds to remove the stationary points belonging to the background (concentric lines etc) and then publish to the filtered_cloud topic to feed this node.

praveen-palanisamy commented 4 years ago

Closing this as the question was answered. Please re-open if more information/clarification is needed.