Object Detection and 3D Pose Estimation with Point Pair Features

Freitag, 24. Juni 2022

The „Point Pair Features“ method uses information extracted from 3D models to detect an object and estimate its pose in a point cloud scene.

As part of my bachelor thesis, I am currently working on a program that uses this method and returns the pose of the searched object.
To retrieve the pose of a 3D model can be very useful for object interaction and tracking, due to the fact that we know the shape of the object and its orientation.
For example, imagine a robot trying to grab a detected mug with a handle on the side. Using the pose and shape information that the program returns, the robot could grab the cup from the handle.