Pick and Place Operations in Logistics Using a Mobile Manipulator Controlled with Deep Reinforcement Learning

Iriondo, Ander; Lazkano, Elena; Susperregi, Loreto; Urain, Julen; Fernández, Ane; Molina, Jorge

doi:10.3390/app9020348

Cited by 57 publications

(28 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep reinforcement learning is another research line that should be explored here, as it is having promising results in many fields and the ASLAM model fits with the problem type to which it gives response [ 150 , 151 , 152 , 153 , 154 , 155 ]. Although not many machine learning approaches have been published yet in the ASLAM field, the proposals of recent years have drawn the attention of the community [ 156 , 157 , 158 , 159 , 160 ].…”

Section: On Going Developmentsmentioning

confidence: 99%

Active Mapping and Robot Exploration: A Survey

Lluvia

Lazkano

Ansuategi

2021

Sensors

Self Cite

View full text Add to dashboard Cite

Simultaneous localization and mapping responds to the problem of building a map of the environment without any prior information and based on the data obtained from one or more sensors. In most situations, the robot is driven by a human operator, but some systems are capable of navigating autonomously while mapping, which is called native simultaneous localization and mapping. This strategy focuses on actively calculating the trajectories to explore the environment while building a map with a minimum error. In this paper, a comprehensive review of the research work developed in this field is provided, targeting the most relevant contributions in indoor mobile robotics.

show abstract

Section: On Going Developmentsmentioning

confidence: 99%

Active Mapping and Robot Exploration: A Survey

Lluvia

Lazkano

Ansuategi

2021

Sensors

Self Cite

View full text Add to dashboard Cite

show abstract

“…Following this same path, Wang et al [ 20 ] present a novel mobile manipulation system that decouples visual perception from the deep reinforcement learning control, improving its generalization from simulation training to real-world testing. Additionally, Iriondo et al [ 21 ] include a deep reinforcement learning (DRL) approach for pick and place operations in logistics using a mobile manipulator. Moreover, deep learning algorithms have also been used in force-based human-robot interaction, specifically for the identification of robot tool dynamics [ 22 ], allowing a fast computation and adding noise robustness.…”

Section: Related Workmentioning

confidence: 99%

Path Driven Dual Arm Mobile Co-Manipulation Architecture for Large Part Manipulation in Industrial Environments

Ibarguren

Daelman

2021

Sensors

View full text Add to dashboard Cite

Collaborative part transportation is an interesting application as many industrial sectors require moving large parts among different areas of the workshops, using a large amount of the workforce on this tasks. Even so, the implementation of such kinds of robotic solutions raises technical challenges like force-based control or robot-to-human feedback. This paper presents a path-driven mobile co-manipulation architecture, proposing an algorithm that deals with all the steps of collaborative part transportation. Starting from the generation of force-based twist commands, continuing with the path management for the definition of safe and collaborative areas, and finishing with the feedback provided to the system users, the proposed approach allows creating collaborative lanes for the conveyance of large components. The implemented solution and performed tests show the suitability of the proposed architecture, allowing the creation of a functional robotic system able to assist operators transporting large parts on workshops.

show abstract

“…The article written by Iriondo et al (2019) is focused on operations performed by a mobile manipulator, in this case a Kuka iiwa, in areas such as logistics which include object picking in an unstructured area. Regarding how that is done, Iriondo et al (2019) states that a "robot manipulator has to be able to place its base in a correct place, so the arm is able to plan a trajectory up to an object in a table."…”

Section: Mobile Cobotmentioning

confidence: 99%

Development of a solution for adding a collaborative robot to an industrial AGV

D'Souza

Costa

Pires

2020

View full text Add to dashboard Cite

Purpose The Industry 4.0 initiative – with its ultimate objective of revolutionizing the supply-chain – putted more emphasis on smart and autonomous systems, creating new opportunities to add flexibility and agility to automatic manufacturing systems. These systems are designed to free people from monotonous and repetitive tasks, enabling them to concentrate in knowledge-based jobs. One of these repetitive functions is the order-picking task which consists of collecting parts from storage (warehouse) and distributing them among the ordering stations. An order-picking system can also pick finished parts from working stations to take them to the warehouse. The purpose of this paper is to present a simplified model of a robotic order-picking system, i.e. a mobile manipulator composed by an automated guided vehicle (AGV), a collaborative robot (cobot) and a robotic hand. Design/methodology/approach Details about its implementation are also presented. The AGV is needed to safely navigate inside the factory infrastructure, namely, between the warehouse and the working stations located in the shop-floor or elsewhere. For that purpose, an ActiveONE AGV, from Active Space Automation, was selected. The collaborative robot manipulator is used to move parts from/into the mobile platform (feeding the working stations and removing parts for the warehouse). A cobot from Kassow Robots was selected (model KR 810), kindly supplied by partner companies Roboplan (Portugal) and Kassow Robotics (Denmark). An Arduino MKR1000 board was also used to interconnect the user interface, the AGV and the collaborative robot. The graphical user interface was developed in C# using the Microsoft Visual Studio 2019 IDE, taking advantage of this experience in this type of language and programming environment. Findings The resulting prototype was fully demonstrated in the partner company warehouse (Active Space Automation) and constitutes a possible order-picking solution, which is ready to be integrated into advanced solutions for the factories of the future. Originality/value A solution to fully automate the order-picking task at an industrial shop-floor was presented and fully demonstrated. The objective was to design a system that could be easy to use, to adapt to different applications and that could be a basic infrastructure for advanced order-picking systems. The system proved to work very well, executing all the features required for an order-picking system working in an Industry 4.0 scenario where humans and machines must act as co-workers. Although all the system design objectives were accomplished, there are still opportunities to improve and add features to the presented solution. In terms of improvements, a different robotic hand will be used in the final setup, depending on the type of objects that are being required to move. The amount of equipment that is located on-board of the AGV can be significantly reduced, freeing space and lowering the weight that the AGV carries. For example, the controlling computer can be substituted by a single-board-computer without any advantage. Also, the cobot should be equipped with a wrist camera to identify objects and landmark. This would allow the cobot to fully identify the position and orientation of the objects to pick and drop. The wrist camera should also use bin-picking software to fully identify the shape of the objects to pick and also their relative position (if they are randomly located in a box, for example). These features are easy to add to the developed mobile manipulator, as there are a few vision systems in the market (some that integrate with the selected cobot) that can be easily integrated in the solution. Finally, this paper reports a development effort that neglected, for practical reasons, all issues related with certification, safety, training, etc. A future follow-up paper, reporting a practical use-case implementation, will properly address those practical and operational issues.

show abstract

Pick and Place Operations in Logistics Using a Mobile Manipulator Controlled with Deep Reinforcement Learning

Cited by 57 publications

References 23 publications

Active Mapping and Robot Exploration: A Survey

Active Mapping and Robot Exploration: A Survey

Path Driven Dual Arm Mobile Co-Manipulation Architecture for Large Part Manipulation in Industrial Environments

Development of a solution for adding a collaborative robot to an industrial AGV

Contact Info

Product

Resources

About