Lessons from the Amazon Picking Challenge: Four Aspects of Building Robotic Systems

By Clemens Eppner, Sebastian Höfer, Rico Jonschkowski, Roberto Mart'in-Mart'in, Arne Sieverling, Vincent Wall, Oliver Brock. 2016

202 citations recorded by the source corpus.

We describe the winning entry to the Amazon Picking Challenge. From the experience of building this system and competing in the Amazon Picking Challenge, we derive several conclusions: 1) We suggest to characterize robotic system building along four key aspects, each of them spanning a spectrum of solutions—modularity vs. integration, generality vs. assumptions, computation vs. embodiment, and planning vs. feedback. 2) To understand which region of each spectrum most adequately addresses which robotic problem, we must explore the full spectrum of possible approaches. To achieve this, our community should agree on key aspects that characterize the solution space of robotic systems. 3) For manipulation problems in unstructured environments, certain regions of each spectrum match the problem most adequately, and should be exploited further. This is supported by the fact that our solution deviated from the majority of the other challenge entries along each of the spectra.

Subfields: Artificial Intelligence, Control and Systems Engineering, Mechanical Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Control and Systems Engineering

In Mechanical Engineering

Citations in the corpus

RLBench: The Robot Learning Benchmark & Learning Environment (2019, methodology): “The ACRV Picking Benchmark [26] aimed to solve this by creating a similar, but reproducible setup to the ARC.”
Interactive Visual Grounding of Referring Expressions for Human-Robot Interaction (2018, background): “In robotics, object recognition is often treated as a classiﬁcation task, with a predeﬁned set of object category labels [5, 28].”
Dex-Net 3.0: Computing Robust Robot Suction Grasp Targets in Point Clouds using a New Analytic Model and Deep Learning (2017, methodology): “In 2015, Team RBO [7] won by pushing objects from the top or side until suction was achieved using a vacuum with high air flow, and Team MIT [36] came in second place using a strategy that positioned the suction cup directly above the centroid of objects with flat surfaces.”
Multi-view self-supervised deep learning for 6D pose estimation in the Amazon Picking Challenge (2016, methodology): “Vision algorithms for robotic manipulation typically output 2D bounding boxes, pixel-level segmentation [4, 5], or 6D poses [6, 7] of the objects.”
Learning Visual Affordances for Robotic Manipulation (2019, methodology): “Vision algorithms for robotic manipulation typically output 2D bounding boxes, pixellevel segmentation [55, 26], or 6D poses [127, 121] of the objects.”
Analysis and Observations From the First Amazon Picking Challenge (2016, methodology): “Object recognition pipeline employed by the RBO team [6], [7].”
The ACRV picking benchmark: A robotic shelf picking benchmark to foster reproducible research (2016, background): “2D images and point clouds were created [16]1, and raised questions about how to trade off generality versus engineered solutions in the research community [17].”
Dex-Net 3.0: Computing Robust Vacuum Suction Grasp Targets in Point Clouds Using a New Analytic Model and Deep Learning (2018, background): “This defies our intuition since empirical evidence suggests that a single point of suction can robustly transport objects [6], [10].”