Paper Title: Exploiting Locally Consistent Patches in Probabilistic Sensor Models

• Authors: Michael Krainin, Kurt Konolige, and Dieter Fox

Documentation

All code from this paper is available in the pose_corrector ROS package: pose_corrector

Additionally, Ecto Python bindings are available in the ecto_corrector project: https://github.com/mkrainin/ecto_corrector

Data

We collected a total of 46 frames of test data for pose estimation in scenes with clutter and occlusion. This data can be downloaded here: test_data.tar.gz

The mesh models for the objects can be found here (in addition to other objects originally used for the Solutions In Perception Challenge: object_meshes.tar.gz