Pose Estimation for Objects with Rotational Symmetry


Pose estimation is a widely explored problem, enabling many robotic tasks such as grasping and manipulation. In this paper, we tackle the problem of pose estimation for objects that exhibit rotational symmetry, which are common in man-made and industrial environments. In particular, our aim is to infer poses for objects not seen at training time, but for which their 3D CAD models are available at test time. Previous work has tackled this problem by learning to compare captured views of real objects with the rendered views of their 3D CAD models, by embedding them in a joint latent space using neural networks. We show that sidestepping the issue of symmetry in this scenario during training leads to poor performance at test time. We propose a model that reasons about rotational symmetry during training by having access to only a small set of symmetry-labeled objects, whereby exploiting a large collection of unlabeled CAD models. We demonstrate that our approach significantly outperforms a naively trained neural network on a new pose dataset containing images of tools and hardware.


The dataset was obtained using 6669 CAD models in STEP format containing a large variety of objects. These were simulated in Maya to obtain images of scenes that contain objects in achievable static poses. We also used many different textures for the plane (table) and objects. The objects used represent ordinary objects and shapes that we may find in our environments. EPSON dataset containing real images to be released later.

The code contains Maya scripts to automatically generate scenes using the 3D models, and the model described on our paper, based on Pytorch.

Please email Enric Corona if you have any questions or comments.


Many industrial objects such as various tools exhibit rotational symmetries. This occurs when an object is equivalent under certain 3D rotations. Such objects are problematic for pose estimation tasks, where the loss function rarely takes this geometric aspects into account. We show in this work this has a huge impact on performance.


If you find this dataset useful, please cite the following publication:

Pose Estimation for Objects with Rotational Symmetry
E. Corona, K. Kundu and S. Fidler
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2018

This work was supported by Epson. We thank NVIDIA for donating GPUs, and Relu Patrascu for infrastructure support.