Real-time order dispatching for a fleet of autonomous mobile robots using multi-agent reinforcement learning

Malus, Andreja; Kozjek, Dominik; Vrabič, Rok

doi:10.1016/j.cirp.2020.04.001

Cited by 49 publications

(15 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Similarly, [60] states that Hopfield networks had previously been used as TSP solutions. Given that Reinforcement Learning applications can be found in job-shop scheduling problem literature [81,82,83,84], AGV routing [85] and have further been proposed as a future research topic on material handling system location planning [86], adopting RL as resolution technique for FLP, based on re-formulating classic layout formulations as Markov Decision Process, certainly stands to reason. Fig.…”

Section: B Discussion Of Findingsmentioning

confidence: 99%

Bibliometric Study on the Use of Machine Learning as Resolution Technique for Facility Layout Problems

2021

View full text Add to dashboard Cite

Facility Layout Problems (FLP) are concerned with finding efficient factory layouts. Numerous resolution approaches are known in literature for layout optimization. Among those, intelligent approaches are less researched than solutions from exact or approximating approaches. The recent surge of research interest in Artificial Intelligence, and specifically Machine Learning (ML) techniques, presages an increase of such techniques' usage in FLP. However, previous reviews on FLP research induce that, to date, this trend has not yet emerged. Utilizing a systematic literature review coupled with a k-Means based clustering algorithm, we analyzed 25 relevant publication full-texts from an original sample of 1,425 papers. Our findings corroborate the statement that ML techniques have attracted substantially less research interest than most other resolution approaches. While a few papers used Unsupervised Learning algorithms directly as a solution to the FLP, Supervised and Reinforcement Learning were found to be practically irrelevant. ML usage was significantly higher in FLP-adjacent planning tasks such as group technology. Drawing from experiences with other NP-hard combinatorial optimization problems in manufacturing research, we conclude that Reinforcement Learning is most promising to bridge the evident gap between FLP and ML research. Our study further contributes to FLP research by extending established classification frameworks.

show abstract

Section: B Discussion Of Findingsmentioning

confidence: 99%

Bibliometric Study on the Use of Machine Learning as Resolution Technique for Facility Layout Problems

2021

View full text Add to dashboard Cite

show abstract

“…The review results for intralogistics are briefly summarised in Table 4. Beginning with Malus, Kozjek, and Vrabič (2020), an intralogistics-related dispatching solution was implemented to meet real-time requirements and handle a rapidly changing production by utilising autonomous mobile robots (AMRs). Based on the observations of the individual agents, they could negotiate with each other and virtually raised bids for orders.…”

Section: (Intra-) Logisticsmentioning

confidence: 99%

Deep reinforcement learning in production systems: a systematic literature review

Panzer

Bender

2021

International Journal of Production Research

View full text Add to dashboard Cite

Shortening product development cycles and fully customisable products pose major challenges for production systems. These not only have to cope with an increased product diversity but also enable high throughputs and provide a high adaptability and robustness to process variations and unforeseen incidents. To overcome these challenges, deep Reinforcement Learning (RL) has been increasingly applied for the optimisation of production systems. Unlike other machine learning methods, deep RL operates on recently collected sensor-data in direct interaction with its environment and enables real-time responses to system changes. Although deep RL is already being deployed in production systems, a systematic review of the results has not yet been established. The main contribution of this paper is to provide researchers and practitioners an overview of applications and to motivate further implementations and research of deep RL supported production systems. Findings reveal that deep RL is applied in a variety of production domains, contributing to data-driven and flexible processes. In most applications, conventional methods were outperformed and implementation efforts or dependence on human experience were reduced. Nevertheless, future research must focus more on transferring the findings to real-world systems to analyse safety aspects and demonstrate reliability under prevailing conditions.

show abstract

“…The SDR is a 1:1 replica of the real environment, able to simulate robot as well as human motion. It is built, among others, around the GAZEBO simulation software [25] and MoveIt! planning framework [26].…”

Section: Approach and Architecturementioning

confidence: 99%

Seamless Human–Robot Collaborative Assembly Using Artificial Intelligence and Wearable Devices

et al. 2021

View full text Add to dashboard Cite

Seamless human–robot collaboration requires the equipping of robots with cognitive capabilities that enable their awareness of the environment, as well as the actions that take place inside the assembly cell. This paper proposes an AI-based system comprised of three modules that can capture the operator and environment status and process status, identify the tasks that are being executed by the operator using vision-based machine learning, and provide customized operator support from the robot side for shared tasks, automatically adapting to the operator’s needs and preferences. Moreover, the proposed system is able to assess the ergonomics in human–robot shared tasks and adapt the robot pose to improve ergonomics using a heuristics-based search algorithm. An industrial case study derived from the elevator manufacturing sector using a high payload collaborative robot is presented to demonstrate that collaboration efficiency can be enhanced through the use of the discussed system.

show abstract

Real-time order dispatching for a fleet of autonomous mobile robots using multi-agent reinforcement learning

Cited by 49 publications

References 12 publications

Bibliometric Study on the Use of Machine Learning as Resolution Technique for Facility Layout Problems

Bibliometric Study on the Use of Machine Learning as Resolution Technique for Facility Layout Problems

Deep reinforcement learning in production systems: a systematic literature review

Seamless Human–Robot Collaborative Assembly Using Artificial Intelligence and Wearable Devices

Contact Info

Product

Resources

About