arl-affpose-dataset-utils

This repo contains tools used to prep the Advanced Robotics Lab (ARL) Affordance Pose (AffPose) Dataset.

Real and Synthetic images can be found here. Object Mesh files can also be downloaded by clicking the link.

The dataset was created using LabelFusion for Real Images or NDDS for Synthetic Images. Note that a Stereolab ZED camera at a resolution of 1280x720 was used or capture or simulate images.

I used the arl-affpose-dataset with the following repos.

1. LabelFusion for generating real images.
2. NDDS for generating synthetic images.
3. pytorch-simple-affnet for predicting an object affordance labels.
4. densefusion for predicting an object 6-DoF pose.
5. arl-affpose-ros-node: for deploying our network for 6-DoF pose estimation with our ZED camera.
6. barrett-wam-arm for robotic grasping experiments. Specifically barrett_tf_publisher and barrett_trac_ik.

Here are several examples!