WRAPP-up: A Dual-Arm Robot for Intralogistics

Manolo Garabini¹, Danilo Caporale², Vinicio Tincani³, Alessandro Palleschi¹, Chiara Gabellieri⁴, Marco Gugliotta⁵, Alessandro Settimi⁶, Manuel Catalano³, Giorgio Grioli³, Lucia Pallottino¹

¹Research Center "Enrico Piaggio", University of Pisa, ²Autonomous Robotics Research Center, Technology Innovation Institute (TII), Abu Dhabi, ³Soft Robotics for Human Cooperation and Rehabilitation, Italian Institute of Technology, Genoa ⁴Robotics and Mechatronics Lab, EEMCS Faculty, University of Twente, ⁵XStar Motion Srl, Pisa, ⁶Proxima Robotics srl, Pisa
IEEE Robotics and Automation Magazine, 2020
Presented at the 2021 IEEE International Conference on Robotics and Automation

Paper Cite

Abstract

The diffusion of the e-commerce has produced larger and larger volumes of different items to be handled in warehouses, with the effect to increase the need for picking automation. Conventionally, automation can be achieved through a custom plant in case of large scale productions where the items have well-known characteristics that are expected to change slowly and little over time. However, today the challenge is to realize a solution that is flexible enough to handle goods with different shapes, sizes, physical properties, and grasping modes. To solve this problem we first analyzed how humans perform picking and then synthesized their behavior in four main tactics. These have been used as guidelines for the design, the planning and the control of WRAPP-up: a dual arm robot composed of two anthropomorphic manipulators, a Pisa/IIT SoftHand and a Velvet Tray. The system has been validated and evaluated through extensive experimental tests.

Block scheme of the integrated system. The pose of the object estimated by the perception system is the input of the planning module that generates a suitable picking strategy, accordingly to pre-defined manipulation primitives taking into consideration the dimensions, the shape, and the pose of the target object. The reactive planner algorithm is used to adapt and re-plan online the Cartesian trajectory based on contact information from force torque sensors at the end-effectors

The output of the perception algorithm is shown overlaid on the point cloud of the scene. In the top row results for cuboid objects are displayed. From left to right: single green box, single brown box, a green box and a brown box in a loosely-packed configuration, a brown box and a white box in a loosely-packed configuration. In the bottom row results for cylinders are displayed. From left to right: single green bucket, single yellow bucket, two yellow buckets in a loosely-packed configuration, green bucket stacked over a yellow bucket in a tightly-packed configuration.

By observing expert operators, we identified four main maneuvers they commonly adopt to pick boxes and cylinder in warehouses. These maneuvers have been encoded into four parametric motion primitives used to plan the motion of the robot during the task execution.

The figure shows the schematic approach used to implement the reactive motion planner (a), and a practical implementation for a picking task (b). The conceptual scheme is used to represent the state machine approach, where the picking strategy is decomposed in a set of states each planned online. An example of this approach is represented in the bottom figure (b), where the picking strategy is decomposed in 6 states. When a contact between the hand and the bucket is detected (the force exceeds a given threshold), a transition from Approach (A) to Manipulation (M), (A to M), is triggered.

WRAPP-up picking thin boxes using the horizontal rotation strategy.

WRAPP-up picking thick boxes using the sliding primitive.

WRAPP-up picking thick boxes using the vertical rotation primitive.

WRAPP-up picking cylindrical objects.

Photo sequence of the autonomous pick-and-palletizing task in the novel facility at UNIPI.

Video Presentation

Videos

Paper

BibTeX

@article{garabini2020wrapp,
        title={WRAPP-up: A dual-arm robot for intralogistics},
        author={Garabini, Manolo and Caporale, Danilo and Tincani, Vinicio and Palleschi, Alessandro and Gabellieri, Chiara and Gugliotta, Marco and Settimi, Alessandro and Catalano, Manuel Giuseppe and Grioli, Giorgio and Pallottino, Lucia},
        journal={IEEE Robotics \& Automation Magazine},
        volume={28},
        number={3},
        pages={50--66},
        year={2020},
        publisher={IEEE}
      }

Acknowledgement

This work was supported in part by the European Unions Horizon 2020 research and innovation program as part of the projects ILIAD (Grant no.732737), and in part by the Italian Ministry of Education and Research in the framework of the CrossLab project (Departments of Excellence).