Human-Guided Learning and Benchmarking of Robotic Heap Sorting
In HEAP we focus on advancing the state-of-the-art for sorting a heap of unknown, irregular objects and provide appropriate benchmarks. In our scenarios, we deal with unknown, broken or deformed object instances such as concrete, metal pipes and other plastic/metal parts of possibly complex shape. A major goal of this project is to make challenging manipulation tasks easily accessible and reproducible, and to allow for a comparative evaluation of different approaches using a standardized robotic platform and an open source simulation framework. Our intention in providing such a benchmark framework is to (i) evaluate state-of-the-art grasping and manipulation algorithms in these complex heap sorting setups and (ii) to define new challenges in terms of object recognition and manipulation that need to be solved by the community.
FUNDING:
The project is funded by FWF – Austrian Science Foundation & CHIST-ERA.