- Faster RCNN object detection
- Point cloud segmentation with Point Cloud Library
- Point cloud clustering (k-mean?)
- ApproxMVBB get bounding box of point cloud
- Create obstacle for motion planning
- 2D pose estimate to get initial pose(human input)
- Navigate to the door way to the target room
- Hand control over to depth camera to navigate through door way
- Object detection
- Tracker based navigation towards object
- Depth camera based grasping