As part of the MOSES project, under the EU's Horizon 2020 framework, CORE has pioneered the transition from traditional manual docking procedures to an autonomous swarm of tugboats. These advancements were made possible by creating a sophisticated simulation environment and training the agents through deep reinforcement learning techniques to optimize docking strategies. This digital twin technology, coupled with an AutoPilot control system, exemplifies a significant leap forward in maritime operations, reducing docking time and enhancing port service availability and environmental sustainability.

The docking procedure of large vessels and containerships is a very important and complex operation in the maritime sector and typically involves a series of different subprocesses for its execution. Existing docking procedures rely on manually operated conventional tugboats, where the tugboat captain has to detect the vessel and its approach point, approach delicately with an allowed velocity, and berth the ship in a collaborative way while interacting with fellow tugboat captains. During all these actions, the captain must deal with a dynamic environment and many uncertainties, which increases the docking time and, hence, affects the port services availability and port emissions.

As part of the MOSES project (H2020, 861678), a cross over from the current manual system to an autonomous tugboats swarm system has been accomplished, to assist in the manoeuvring of a containership and docking at berth. As a first step, a virtual environment had been created, utilising the Unity3D software and its associated modules, to simulate the real-life components, such as the port, water mass, tugboats and containership. To create a virtual environment as close as possible with the real one a series of hydrodynamic simulations were conducted to analyse the navigation and evaluate the hydrodynamic parameters, such as the friction resistances for each ship object separately. Additionally, Finite Element Model simulations (FEM) were employed to assess the interactions between the tugboats and the containership, by evaluating the force-reactions and stresses.

The simulation environment served as a training environment for the developed swarm intelligence machine learning algorithm by allowing agents to learn from their experiences. Specifically, through the ML-Agents toolkit the agents (tugboats) were extensively trained utilising deep reinforcement learning techniques, where the learning procedure is based on the interaction of the agents with the environment and the accumulation of feedback (rewards or penalties), while the agents collected observations through LiDAR and GPS sensors. The goal is to discover optimal strategies that maximize cumulative rewards over time. The digital twin was deployed at the edge, along with an AutoPilot system to control the steering and thrust of the tugboats based on the digital twin’s inference. The digital twin was successfully demonstrated at the Faaborg port in Denmark (at TRL6), when applied in a swarm of two tugboats achieving over 25% reduction of the manoeuvring and docking time and, hence, a corresponding reduction of port emissions and increase of port services availability.

<p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB"><a href="https://moses-h2020.eu/" style="color: rgb(70, 120, 134); text-decoration: underline;">https://moses-h2020.eu/</a><o:p></o:p></span></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB">&nbsp;</span></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB"><a href="https://www.youtube.com/watch?v=28P-BRpVXRY&amp;ab_channel=MOSESProject" style="color: rgb(70, 120, 134); text-decoration: underline;">https://www.youtube.com/watch?v=28P-BRpVXRY&amp;ab_channel=MOSESProject</a><o:p></o:p></span></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB">&nbsp;</span></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB"><a href="https://www.researchgate.net/publication/371260713_MOSES_Autonomous_tugboat_swarm_operation_Operational_scenarios_requirements_and_architecture" style="color: rgb(70, 120, 134); text-decoration: underline;">https://www.researchgate.net/publication/371260713_MOSES_Autonomous_tugboat_swarm_operation_Operational_scenarios_requirements_and_architecture</a><span class="MsoHyperlink" style="color: rgb(70, 120, 134); text-decoration-line: underline;"><o:p></o:p></span></span></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB">&nbsp;</span></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-GB"><a href="https://www.core-innovation.com/" style="color: rgb(70, 120, 134); text-decoration: underline;">https://www.core-innovation.com/</a>&nbsp;<o:p></o:p></span></p>

The SMC FIC-400 Teaching Learning Factory (TLF), part of PBN’s Digital Innovation Hub in Hungary, has integrated an innovative Digital Twin of its new recycling module, under the DigiTwinGreen project funded by the EIT Regional Innovation Scheme. The Digital Twin simulates the recycling process in a Mixed-Reality environment, allowing for hands-on training and optimization of the recycling module's efficiency.

The SMC FIC-400 Teaching Learning Factory (TLF), located in PBN’s Digital Innovation Hub in Hungary, offers the opportunity to design, develop, and test successfully working solutions for the industry. The TLF is a critical component of a 15-module system including 5 Industry 4.0 modules. This assembly line system features a pallet and container feeder, pellet feeder, cupping station, warehouse, as well as a labelling and dispatching station. These are all connected to a Manufacturing Engineering System, capable of managing orders, initiating manufacturing process, monitoring warehouse state, etc.

As part of the DigiTwinGreen project, funded under the EIT Regional Innovation Scheme, the TLF has been extended with a new recycling module capable of sorting balls of different colour. Specifically, balls from the container disk fall into the nests of a rotating dosing disc, where a sensor identifies the colour of the balls triggering the corresponding latch to open. Subsequently, the balls are delivered to the appropriate container by pipes that continue beneath the worktable. The recycling module also includes a dedicated User Interface (UI), featuring a slider to set the speed of the disk (in RPM), start/stop buttons and a numeric display which displays the current speed value (RPM), the cycle time and the number of sorted pieces (per colour and total number).

Under this context, a Digital Twin simulation environment for the recycling module has been developed to enable the digital representation and operation of the recycling procedure. For the visualisation and control of the virtual twin module, a Mixed-Reality (MR) approach has been employed to blend the real and virtual aspects of the module and facilitate their interaction in semi real-time. In this way, the Digital Twin can provide hands-on training to new workers, enabling them to learn the manufacturing process and gain practical knowledge quickly and efficiently. The MR technology has been implemented through the Oculus Quest Pro headset, while the developed application carried out in Unity 3D software. The whole solution was deployed at the edge. For the seamless accommodation of the virtualization and simulation task’s demands, a powerful workstation has been utilised, encapsulating processing power for both CPU and GPU, as well as meeting the essential memory requirements for deploying MR applications.

Furthermore, to ensure the green modality of the recycling module, a solar panel was utilised to supply the energy along with storage unit to store the energy produced. Through the simulation environment the process efficiency of the module was enhanced by identifying the theoretical optimal rotational speed (RPM) of the rotating disk which maximizes the throughput (pieces/time) of the module. Finally, CORE Innovation Centre has verified that the Digital Twin’s features comply with the user requirements, and the Digital Twin’s accuracy has been validated by data-driven techniques, proving that it is capable of accurately replicating the operational behaviour of the real-world recycling module.

<p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><a href="https://digitwingreen.eu/" target="_blank">https://digitwingreen.eu/</a><a href="https://digitwingreen.eu/" target="_blank"></a></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><br></p><p class="MsoNormal" style="margin: 0cm; text-align: justify;"><a href="https://www.pbn.hu/" target="_blank">https://www.pbn.hu/</a><font face="Aptos, sans-serif"><span style="font-size: 14.6667px;">&nbsp;</span></font></p><p class="MsoNormal" style="margin: 0cm; text-align: justify;"><font face="Aptos, sans-serif"><span style="font-size: 14.6667px;"><br></span></font></p><p class="MsoNormal" style="margin: 0cm; text-align: justify;"><font face="Aptos, sans-serif"><span style="font-size: 14.6667px;"><br></span></font><br></p><p class="MsoNormal" style="margin: 0cm; text-align: justify;"><font face="Aptos, sans-serif"><span style="font-size: 14.6667px;"><br></span></font></p><p class="MsoNormal" style="margin: 0cm; text-align: justify;"><font face="Aptos, sans-serif"><span style="font-size: 14.6667px;"><br></span></font><br></p><p class="MsoNormal" style="margin: 0cm; text-align: justify; font-size: 11pt; font-family: Aptos, sans-serif;"><span lang="EN-US"><o:p></o:p></span></p>

A digital twin of the Hyundai Konas battery pack was created to make data sets for computer vision models and teach the battery pack's robotic disassembly

There is a pressing need to develop an efficient method for disassembling battery packs and reclaiming rare earth materials for recycling. Despite this imperative, numerous challenges persist, requiring significant advancements to establish an effective solution. Currently, battery pack disassembly is predominantly a manual process, a practice that necessitates reconsideration in the near future.

Map of battery recycling

To overcome this problem, we are using metaverse and digital twins to teach the robot to disassemble the batteries. This is made possible by Using Nvidia Omniverse software, which enables the creation of realistic-looking renders and data sets and teaches robots with reinforcement learning.

For Omniverse, the digital twin just needs to be a CAD file of the system, and in our case, we created the CAD by reverse engineering it to Onshape, which is browser-based CAD software that allows users to directly import the CAD models to Omniverse.

The CAD models

After the import to Omniverse, it's possible to put realistic-looking materials to the part and create date sets from the model with Omniverses replicator and simple Python scripts. And to simulate complex geometries like unscrewing with the signed distance field collisions.

Battery pack in omniverse

<p>Switzerland battery research center<br></p>

A digital twin has been created for the process of thermo-moulding within the manufacturing of shoe insoles. The large number of orders, added to the large amount of variety that an order can have, added to the numerous customizations that customers want for each of their orders makes our resource management impossible to manage by human beings. As a result, every day we have more than 21,000 different orders going around our warehouse and being processed by our machines. Achieving a perfect organization of resources and time has become a huge challenge that is collapsing one of our main processes.

The objective is to develop a scalable algorithm fed with information in real time that is capable of distributing orders in the machines, optimizing waiting time, electricity consumption and the availability of limited resources. All these variables and casuistics make it impossible for a conventional mathematical algorithm to find a favorable solution in a short time. So we need computing power and more sophisticated algorithms and techniques. In principle we are going to try to address the problem with genetic algorithms, they are a type of algorithms that can explore many search spaces in a short time and can address multifactorial problems.

After developing the solution, we will have a great impact at the organizational level, so we will not need so many resources for the management and organizational part and thus allocate more resources to what is really important, which is to scale and increase manufacturing. We will reduce the use of electricity, which is a great incentive in these times of energy crisis. We will optimize the use of space in our warehouses, since everything will be cleaner and orderlier and it will be easier to locate the assets since we will gain more speed in the manufacturing flow.

<p><a href="https://digitbrain.eu/3rd-wave-of-digitbrain-experiments/experiment-18-insotwin-an-algorithm-to-autonomously-and-intelligently-plan-all-the-tasks-that-our-machines-have-to-perform/" target="_blank">https://digitbrain.eu/3rd-wave-of-digitbrain-experiments/experiment-18-insotwin-an-algorithm-to-autonomously-and-intelligently-plan-all-the-tasks-that-our-machines-have-to-perform/</a>&nbsp;</p><p><a href="https://www.itainnova.es" target="_blank">https://www.itainnova.es</a>&nbsp;<br></p>

A rotary dryer reduces the humidity in particulate matter through direct contact with combustion gases from fossil or renewable fuels (biomass). The design and operation of a rotary dryer present significant challenges, mainly because it must be flexible enough to adapt to a multitude of operating modes. Under this context, we want to prove the benefits for the manufacturer of the industrial product of the development of a digital twin for the rotary dryer. The digital twin will support the design, production and operation phases of the rotary dryer. It will contribute to the faster and more reliable design, better integration of the equipment in the customers’ general process and provide an optimal configuration of the process.

This DT uses the detailed simulation of a process to determine the optimal control of an installation. This methodology, applied to continuous processes, we believe is innovative and applicable to multiple manufacturing operations. From the economic point of view, as indicated above, the proposed tool is positioned in markets that will experience considerable growth in the coming years. In addition, this digital twin has multiple types of users: from rotary dryer manufacturing companies to companies that use these dryers in their processes. This will facilitate the exploitation of the tool. Finally, this DT is built under the concept of a sustainable and digital transition since the original application of the digital twin will focus on a rotary dryer to produce biomass pellets, one of the expected impacts being the improvement of the energy efficiency in the process.

<p><a href="https://youtu.be/LGz12-6UDb8" target="_blank">https://youtu.be/LGz12-6UDb8</a><a href="https://youtu.be/LGz12-6UDb8" target="_blank"></a>&nbsp;</p><p><a href="https://digitbrain.eu/2nd-wave-of-digitbrain-experiments/dt4dryer/" target="_blank">https://digitbrain.eu/2nd-wave-of-digitbrain-experiments/dt4dryer/</a>&nbsp;</p><p><a href="https://www.itainnova.es" target="_blank">https://www.itainnova.es</a>&nbsp;<br></p>