Feat/pointcloud clustering

[yyajima21]

Description

This project contains a software package that segments lidar point cloud into clusters and creates a bounding box over each cluster in real time. This package adds a feature to localize objects surrounded by the robot such as barrels and people. Here is readme file for this project.

This PR does the following:

• Ground plane segmentation
• Point cloud clustering that categorizes lidar points into several clusters using two well known method (region growing and euclidean clustering methods)
• Visualization tool that creates a bounding box over each clusters

Fixes #777 and #651

Testing steps (If relevant)

Test Case 1

1. Download a bag file (e.g. JESSII_RUN_SUN_2_2019-06-10-15-30-13.bag) from the robojacket cloud
2. Modify launch file and add your file path
3. Run the following command to visualize the point cloud segmentation.
   roslaunch igvc_perception pointcloud_filter.launch
roslaunch igvc_perception ptseg.launch

Expectation: You should see the following result in your rviz screen. Each cluster has its own bounding box.

Note: here is readme file for this project.

Self Checklist

• I have formatted my code using make format
• I have tested that the new behavior works

[andrewyarovoi]

Overall, this code looks good. The main thing is that there are a lot of hard coded parameters. These should be added to the pointcloud_filter.yaml file and then should be passed in to the util functions. That way, they can be tuned without requiring rebuilding the project and they can all be in one place (with comments and units).

[yyajima21]

@andrewyarovoi I removed hard coded values based on your feedback.

[andrewyarovoi]

Thanks for addressing my comments. Everything looks good now. Great job!