Analysis and Observations From the First Amazon Picking Challenge

Abstract

Abstract-This paper presents a overview of the inaugural Amazon Picking Challenge along with a summary of a survey conducted among the 26 participating teams. The challenge goal was to design an autonomous robot to pick items from a warehouse shelf. This task is currently performed by human workers, and there is hope that robots can someday help increase efficiency and throughput while lowering cost. We report on a 28-question survey posed to the teams to learn about each team's background, mechanism design, perception apparatus, planning and control approach. We identify trends in this data, correlate it with each team's success in the competition, and discuss observations and lessons learned based on survey results and the authors' personal experiences during the challenge.Note to Practitioners: Abstract-Perception, motion planning, grasping, and robotic system engineering has reached a level of maturity that makes it possible to explore automating simple warehouse tasks in semi-structured environments that involve high-mix, low-volume picking applications. This survey summarizes lessons learned from the first Amazon Picking Challenge, highlighting mechanism design, perception, and motion planning algorithms, as well as software engineering practices that were most successful in solving a simplified order fulfillment task. While the choice of mechanism mostly affects execution speed, the competition demonstrated the systems challenges of robotics and illustrated the importance of combining reactive control with deliberative planning.

Introspective Evaluation of Perception Performance for Parameter Tuning without Ground Truth

“…We then estimate the 2D rigid displacement of the camera assuming all points are on a plane parallel to the camera and differentiate it to estimate the body velocity. Our software uses the OpenCV library's implementation of Lucas-Kanade with pyramids and rigid transformation estimation 1 . The six VO parameters we tune in our experiments are described in Tab.…”

“…Of the many core robot competencies, the one perhaps most challenged by this growing requirement of generality is perception [1]. In some ways this is natural, as perception systems typically provide task-specific interpretations of reality.…”

A novel gripper design for multi hand tools grasping under tight clearance constraints and external torque effect

“…These designs prove effectiveness for their target tasks like object picking [15], [16]. A two plates intelligent design is presented in [7] for picking objects with simple geometry.…”

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

“…However, what should come as a surprise is the sparsity with which the corresponding spectra have been explored by our community and how rarely these aspects are used explicitly to characterize robotic systems. Case in point: Our solution to the Amazon Picking Challenge explores different regions on these spectra than most other challenge entries [13]. We believe-and will support this belief in the remainder of this paper-that these differences were crucial for our success.…”

“…III-B). This positions our solution far to the feedback-side of the spectrum, standing in contrast to the majority of the other challenge entries (80% of the teams used motion planning, 44% used MoveIt [47], [13]). Feedback control is so successful in the Amazon Picking Challenge setting because the task only requires a limited range of motions, and the shelf provides plenty of contact surfaces to generate useful feedback.…”