IROS 2019

Object Pose Estimation Challenge for Bin-Picking

Short message /

Two-arm robot carries out bin picking
© Fraunhofer IPA/Photo: Rainer Bez
Two-arm robot carries out bin-picking

Fraunhofer IPA presents a challenge to all researches, universities and industrials taking part in 2019’s IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Everyone who is working or willing to work on estimating the pose of objects in bin-picking settings can compete in a challenge against industrial companies and research institutions from all over the world.

The task is to estimate the position and orientation of objects in scenes typical for bin-picking based on depth images or point clouds. These scenarios consist of multiple rigid objects of the same type being chaotically stored in a bin. As Object Pose Estimation is challenging due to cluttered scenes with multiple and heavy occlusions, the goal of this challenge is to compare different approaches and advance the development of novel methods. Various labeled large-scale datasets are provided to prepare for this challenge, so that machine learning methods, that need many data, can be developed, trained, and tested.

Test routines with training dataset

The registration is open until August 31, 2019. As of now, the attendees can download a training dataset, which is ideally suited for industrial applications, and scripts to facilitate its usage. This enables the participants to develop their algorithms, setup and test routines and methods to allow smooth performances on the competition day. The datasets used for evaluation will be provided on the competition day.

The performance of the different methods will be compared by using a metric by Romain Brégier et al. This evaluation methodology is suitable for rigid objects, scenes of many parts in bulk (bin-picking scenarios) and properly considers object symmetries. Only the pose of objects that are less than 50% occluded are relevant for the retrieval.

IROS 2019 is one of the largest and most impacting robotics research conferences worldwide. It brings together an international community of researchers, educators and practitioners to explore the frontier of science and technology in intelligent robots and systems, and to discuss the latest advancements in this fast growing and exciting field. Fraunhofer IPA is looking forward to get to know the contestants and their solutions. Interested parties can register for the challenge here: http://www.bin-picking.ai/en/Registration.html