Cobots: a disruptive innovation in industrial robotics, the opportunity for SMEs Human-Centered AI & Ethics Predictive Maintenance, Data monetization, & AI SHOP4CF (Smart Human Oriented Platform for Connected Factories) is an EU-funded project within the eighth framework program Horizon 2020 that aims to create a unique infrastructure for the convenient deployment of human-centric industrial applications.
SHOP4CF believes that people should not be replaced by automated process but can gain new relevance through usage of their individual abilities within the factory. Therefore, new technologies need to be human-centric.
SHOP4CF (Smart Human Oriented Platform for Connected Factories) is an EU-funded project within the eighth framework program Horizon 2020 that aims to create a unique infrastructure for the convenient deployment of human-centric industrial applications. In the project, 20 partners develop a comprehensive software platform containing a wide range of components that cover a broad spectrum of industrial requirements, especially in the context of modern, flexible, and data-rich manufacturing. Apart from the core consortium who submitted the original SHOP4CF proposal, the project is looking for additional contributors during the project’s runtime via several open calls.
The overall vision of SHOP4CF is that tomorrow’s production requires both machine skills (e.g., high accuracy, precision, or persistence) and human resources (e.g., creativity, adaptability, or tactile sense). Therefore, all components considered in the SHOP4CF software platform aim at the mutual complementation of human labor and machines in order to improve the working conditions by:
AUTOMATING MONOTONOUS / LABORIOUS WORK
INCREASING HUMAN PRODUCTIVITY THROUGH SMART ASSISTANCE
We believe that SHOP4CF can have a strong positive impact on the labor market, as people in production are not simply replaced by automated processes, but rather gain in importance as their individual abilities find new relevance.
An 8-month program in which selected experiments with receive the support of 20 partners who form the SHOP4CF consortium: a mix of research organizations, universities, industry partners and SMEs.
25 components, offered under a common architecture based on RAMI 4.0 and FIWARE Technologies, support humans in production activities and provide basic implementation blocks of human-centered manufacturing solutions.
Up to 100k equity free funding, technical and business support to SMEs, selected through several open calls.
Testing of SHOP4CF technologies in line with the real industry needs, facilitated by pilots at BOSCH, VW, Arcelik and Siemens.
A vibrant and growing community around the platform necessary to create the adoption critical mass and provide support.
Standardisation in the areas of collaborative robotics, AGVs and use of IoT devices on the shopfloor.
Architecture development should include both:
Components are already defined/existing and they should be taken under consideration while developing an architecture.
There should be a unified vision of how architecture should look like, and which technologies it will include for system development. This approach may be (to some extent) similar to architecture development performed in HORSE project. HORSE project is a complete example of an architecture development process with top-down approach, therefore can serve as an inspiration for SHOP4CF. Details on HORSE are available here:
Facilitated by a pan-European network of Digital Innovation Hubs (DIH), RAMP makes the automation value chain effective and connects manufacturing companies with providers of the optimal robotics solutions and vice versa.
Manufacturers gain access to robotics and digitization technologies to improve efficiency and productivity. Automation suppliers can reach their customers faster and access a larger market.
Available online apps for better manufacturing, such as:
Pilots acting as testing facilities and seeds for adoption of the components. 4 large scale pilots lead by industrial leaders in Europe and 30 external pilots selected through 5 open calls
The ROS2 Monitoring component is meant for developers using ROS2: a dashboard for monitoring health, login, examining services, publishers and subscribers associated to ROS2 nodes. The component includes a GUI, which is used to interact with the component, through the GUI the user can setup the ROS components that the user wants to monitor.
Main non-functional requirements
The component has no real-time responsiveness requirements
Software requirements/dependencies
Platform: Ubuntu, MacOS, Windows Requirements: ROS2, Docker
Hardware requirements
64-bit system capable of running: Ubuntu, MacOS, Windows and ROS2
Security threats
The component should operate behind a firewall during production
Privacy threats
No privacy threats have been identified
Execution place
Private Cloud/PC neart production
Deployment instructions
Deployment instructions can be found on a public repository
User interface
A dashboard showing the current status of all ROS2 nodes in the system
Supported devices
Desktop/Laptop, XX
User defined scenarios (non-technical) and relevant pilot cases
The component can be used to monitor a collection of ROS2 nodes
The main functionality of this component is to enable robot control using natural human actions as input, in this case hand gestures using a gesture tracking glove. With sensor glove by CaptoGlove LLC, the operator makes distinguishable hand gestures to command and control the robot in the process. The component is reconfigurable for different controlling scenarios.
Main non-functional requirements
N/A
Software requirements/dependencies
For the CaptoGlove, there should be Capto Suite installed. Docker installed on the host machine
Hardware requirements
20GB Hard drive space, recommended 2GB RAM
Security threats
None
Privacy threats
None
Execution place
CaptoGlove SDK is ran on host Windows PC and all the other parts of M2O2P are docker images
Deployment instructions
Instructions of application are provided in PDF format, later on video too.
User interface
User interface for the component is mainly in Web UI of the component. This can be reached from host machine navigating to localhost:54400 on browser.
Supported devices
Any Windows 10 machine, CaptoGlove
User defined scenarios (non-technical) and relevant pilot cases
Component can be used in any system where there is a need to send commands or finish tasks by human operator using the glove. In the Siemens pilot case, the component was used to complete tasks in a bin picking collaborative robotic application.
The main functionality of this component is to enable the training and support of human workers in collaborative tasks. For doing so, the main activities of the collaborative task and the interaction of worker and robot is created in Virtual Reality (VR). By using a virtual reality headset and equipment the worker can remotely visualize, monitor, and perform the training of collaborative tasks with robots. It should be noted that based on the use case requirements (e.g., workspace and environment, equipment, safety aspects and interfaces to other components), several data inputs might be needed for creation of custom simulations. The component is divided into a sandbox mode (using pre-programmed actions) and a dynamic mode, which depending on configuration could receive data inputs from ROS nodes for on-the-fly creation of tasks
Main non-functional requirements
N/A
Software requirements/dependencies
Windows 10 and compatible browser (Firefox, Chrome, etc)
Hardware requirements
A VR headset supported by A-Frame with controller positional tracking, as listed here: https://aframe.io/docs/1.2.0/introduction/vr-headsets-and-webvr-browsers.html
Security threats
None
Privacy threats
None
Execution place
Private cloud (meaning in pilot premises), cloud
Deployment instructions
Provide information on where deployment instructions for a ready component can be found (e.g. on public or private access repositories or on websites or only upon request, etc.)
User interface
A main configuration page and a sample workcell layout in VR mode.
Supported devices
Desktops, Laptops
User defined scenarios (non-technical) and relevant pilot cases
This component could be used to train workers in a collaborative assembly process by virtualizing the whole procedure in VR and allowing the worker to interact with the robot and components prior to working in the actual setup. It is important to keep in mind that since this application is meant for training, having a concrete step by step process is required to design and fully benefit the collaborative training
The Digital Twin for control and planning (DT-CP) allows the users to create a virtual replica of the production line facilitated by the use case. The component is divided into two parts: A simulator, to experiment with alternative models to be implemented in the real line, and a monitoring dashboard for having an overview of the line.
The time-based simulator takes as input several configuration parameters (e.g., takt time, shift time), process descriptions and resources (e.g., workers) with different skill set. Hence the user can modify and test different production strategies that would be more complicated and time consuming to test in the real line. The monitoring dashboard can provide the user with an overview on the status of production and provide notification mechanism for alerts when implemented. It should be noted that some functionalities (e.g., data sources for monitoring dashboard, data modelling, and configuration parameters for simulator) are dependent on the use case and might require further adaptations for proper integration in the setup.
Main non-functional requirements
N/A
Software requirements/dependencies
N/A
Hardware requirements
Device capable of handling web-based applications
Security threats
None
Privacy threats
None
Execution place
Private cloud (meaning in pilot premises), cloud
Deployment instructions
Instructions will be provided on the Git page
User interface
The component will be divided into two parts: an online dashboard for monitoring the line in real time and a simulator, to experiment with alternative models to be implemented in the real line.
The monitoring dashboard is intended to follow the flow of product from one workstation to another.
The simulator allows to modify and test different production strategies that would be more complicated and time consuming to test in the real line. The simulator setup consists of four steps: the introduction of the initial process execution parameters, design of the layout, process description assignment and allocation of resources.
Supported devices
Desktop, Laptop
User defined scenarios (non-technical) and relevant pilot cases
Coupled with data collection applications, the monitoring part of the component could be used to have an overview on the process and provide notification and control mechanisms. Data inputs, notification handling, and control mechanisms may be to be further adapted as per use case.
The simulator allows the user to define a time-based simulation of a process assembly by setting concrete configuration parameters (e.g., takt time, shift time) and assigning resources (e.g., workers) with additional resource parameters (e.g., skill set). Th result would be a report on the pre-provided production targets. Specific configuration parameters and settings may have to be further adapted as per use case.
The Data Collection Framework (DCF) component collects data from shop floor (field devices, sensors, and controllers) and enterprise resource planning (ERP) systems using data adapters for different use cases. DCF can be used at production/assembly lines to have an overview on the collected data from sensors and workstations. The main function of the component is data collection from systems, data storage into databases if needed, and for engineers/supervisor to review and take appropriate action. It should be noted that some of the data adapters (e.g., ERP adapters) may need to be configured and tested on a use case basis.
Main non-functional requirements
N/A
Software requirements/dependencies
Data Transfer and Communication within DCF and Database requires Python installed and some Libraries (opcua, paho.mqtt, pymongo, pandas, json, flask, requests, cx_Oracle, hbdcli)
Hardware requirements
– Windows 7 or 10
– x86 64-bit CPU (Intel / AMD architecture)
– 4 GB RAM
-5 GB free disk space
Security threats
The component requires authentication from server/database before connecting and collecting ERP or shopfloor data and storing the data in database.
Privacy threats
None
Execution place
Both devices connected with local network and on different host address can be connected via MQTT and OPC-UA
Deployment instructions
DCF component will be deployed in docker and relevant instructions will be provided.
User interface
After specifying necessary connection configuration, the DCF module is monitoring the temperature and pressure reading through opc-ua server (left image). If the temperature or pressure is more than the allowed, it is logging the information (e.g. time, value, description) in the database (right image). These parameters can be changed according to the use case.
Supported devices
Desktop, Laptop
User defined scenarios (non-technical) and relevant pilot cases
DCF component can be used at production/assembly lines to collect data from workstation/sensors and apply event processing. For instance, if more time is being consumed to complete the task at specific workstation, this activity can be monitored, and relevant data can be logged in database for engineers/supervisor to view and take appropriate action
This component creates user interfaces depending on the information collected from the production line devices and the user’s profile. By doing so, relevant task and operation specific user interfaces are composed for the user. Such interfaces could for example display task specific work description (e.g, description of assembly operation) to users and enable the user to confirm completion of task for interaction with external components. It should be noted that if applicable, other interfaces with different functionalities may have to be developed based on the requirements
Main non-functional requirements
N/A
Software requirements/dependencies
N/A
Hardware requirements
Device capable of use web-based application (e.g.: PC or laptop )
Security threats
None
Privacy threats
None
Execution place
Private cloud (meaning in pilot premises)
Deployment instructions
Instructions will be provided on the Git page
User interface
Supported devices
Desktop, Laptop
User defined scenarios (non-technical) and relevant pilot cases
ADIN can be used by workers in an assembly line for assisting them on the task, giving them the specific and relevant information for fulfilling the duty.
It also can be used in collaborative task with cobots where the worker receive instructions on the steps of the collaboration task.
Human-Centered Process Optimization based on RL in which its main functionality is to provide Reinforcement Learning logic wrapped in ROS2 packages
Main non-functional requirements
Requirements for real-time responsiveness depends on the application. However, since reinforcement learning optimize next action and not the current, real-time responsiveness requirements are relaxed.
Software requirements/dependencies
Platform: Ubuntu, macOS, Windows.
Requirements: ROS2, Docker
Hardware requirements
64-bit system capable of running: Ubuntu, macOS, Windows.
High performance PC/Cloud (good CPU and GPU) for training models.
Eg: +10 core count, +64 GB ram, RTX 2080ti for training (depending on the application and model).
Security threats
When deployed on-premise a firewall should be enough. If deployed in the cloud work is required to ensure a secure connection between the cloud and the production equipment/PC.
Privacy threats
No privacy threats have been identified.
Execution place
Private cloud/PC near robot.
Deployment instructions
Deployment instructions for the component can be found on a private access repository.
User interface
The component will have multiple user interfaces:
A common user interface (dashboard) for developers and end-users containing data visualization, selected actions, and other diagnostics.
Developers will furthermore have a GUI for yaml-file system configuration and all available ROS2 tools for visualization and diagnostics.
Supported devices
Desktop/Cloud
User defined scenarios (non-technical) and relevant pilot cases
The component can be used to optimize a work cell process with reinforcement learning. Example: optimize the process control of a vibration feeder, such that that an element always is available for a robot to pick up.
The component is based on a generic add-on force-control for classical industrial and/or collaborative robots.
An innovative force-sensor based strategy is used to fit two or more parts together that require a snap connection.
The component is a ROS based control approach.
Main non-functional requirements
– Low latency required
– The trained assembly skills can be scaled in time and are primarily limited by the force control performance of the robot (F/T sensor)
Software requirements/dependencies
– ROS framework with ROS control (kinetic or melodic)
– FZI Custom extension of ROS Cartesian Motion, Impedance and Force Controllers
– FZI Custom wrappers for external robotic sensors – Robot ROS driver (e.g. ROS UR)
– TensorFlow 2.1 with python 2.7
Hardware requirements
– Robot with wrist force-torque sensor mounted or integrated
– Dedicated Pc (e.g. i7 shuttle PC with 8 GB ram)
Security threats
The component should run on a separate network without access to the public internet or to any other network not authorized to use it (ROS1 security)
Privacy threats
No specific privacy requirements, no personal information logging
Execution place
Local
Deployment instructions
Internal development, deployment instructions only upon (approved) request
User interface
Text configuration files
Supported devices
Any robot which supports ROS control and can measure end-effector forces and torques (intrinsic or integrated)
User defined scenarios (non-technical) and relevant pilot cases
Force-based assembly tasks which require difficult snap fitting of parts by a robot
Pilot cases: Siemens use case 1
Task manager for safe and efficient human-robot interaction
Main non-functional requirements
– Real time responsiveness is fundamental for the task scheduling to work safely and properly. The supervision of the robot pose requires at least 10 checks per second of the environment, for the robot to accurately react if there is any obstacle on its way
– Sensor information (in particular depth information) needs to be as up-to-date as possible
Software requirements/dependencies
– ROS1 Framework
– GPU-Voxels
– FZI Custom extension of ROS Cartesian Motion, Impedance and Force Controllers
– FZI Specific Extension of the FlexBE ROS package or FZI behaviour-Tree Implementation for Task Modelling and Scheduling
– FZI Custom ROS wrappers for external robotic sensors
– FZI Shared workspace (ROS application of GPU-Voxels) for human-robot-collaboration
– FZI Robot Collision Detection ROS package
– FZI Human Pose Prediction and Tracking software (optional)
– Robot ROS Driver
Hardware requirements
– (Depth) Cameras with fast update rate for the images
– Combination of several sensors (one is not enough)
– 1 shuttle PC for robot control with real time optimization (low latency)
– 1 additional PC with GPU for more computational intense tasks (i.e. collision avoidance, human detection)
Security threats
Run on a separate network (ROS1 security) without access to the public internet or to any network not authorized to use it
Privacy threats
No specific privacy requirements. No personal information, camera or 3D data logging
Execution place
Local
Deployment instructions
Internal development, deployment instructions only upon (approved) request
User interface
Text configuration files
Supported devices
Any robot with ROS driver, URDF description and real time joint angles
User defined scenarios (non-technical) and relevant pilot cases
Efficient Human-Robot collaboration on the shop floor, where the robot needs to fulfil tasks in the proximity of the worker
Pilot Use case: Siemens Use Case 1
Main non-functional requirements
– Inputs expected at 10 Hz. Outputs between 2 and 10 Hz
– Lower frequencies will influence safety and acceptability severely
Software requirements/dependencies
– ROS1 framework
– Google Cartographer ROS
– Move_Base and/or Move_Base_Flex ROS packages
– AGV ROS driver
– External ROS sensor drivers (cameras, lasers)
– Open Pose
– Wheel Odometry (ROS Topic)
Hardware requirements
– SICK Lidar (for example, SICK)
– Intel RealSense and/or 2D camera
– Dedicated PC (Intel5, High End GPU)
Security threats
Operates inside of mobile robot or secured WIFI connection
(no off premises connection required)
Privacy threats
No personal information logging
Execution place
Local
Deployment instructions
Present: Internal development available on approved request
Future: Public access repositories
User interface
– Text configuration files
– (optional) GUI
Supported devices
– Specific PC (to be embedded in a compatible AGV)
– Any AGV with ROS driver
User defined scenarios (non-technical) and relevant pilot cases
Mobile Robot evolving in an industrial plant or public area with people
Pilot use case: Bosch use cases 1 and 2
Graphical front end (GUI) to program new robotic applications by quickly creating new control sequences based on ROS tools.
The tool helps to develop or change the collaborative robotic applications, gives monitoring feedback on the status of the process and could be used to model different tasks as well as the interaction between robot and human transparently.
It is an alternative to SMACH and FlexBE using Behavior Trees.
Main non-functional requirements
No real time responsiveness required
Software requirements/dependencies
ROS1 Framework
Hardware requirements
A PC
Security threats
Run on a separate network (ROS1 security) without access to the public internet or to any network not authorized to use it
Privacy threats
No specific privacy requirements, no personal information logging
Execution place
Local
Deployment instructions
Internal development, deployment instructions only upon (approved) request
User interface
HTML editor to control the functionalities of the task-programming tool
Supported devices
GUI: Any device that allows mouse-like controls
User defined scenarios (non-technical) and relevant pilot cases
Any scenario which involves programming of robots
Pilot use cases: Siemens use cases 1 and 2, Bosch use case 1
This component is used to determine whether the chosen robot trajectory & speed is safe and the required separation distance has been chosen adequately and can be covered by the sensor configuration. (This uses a calculation of the size of the required separation distances for robots that use the operating mode speed and separation monitoring.)
Main non-functional requirements
Trajectories should be checked as early as possible to minimize the delay of execution. Ideally, precomputed trajectories are validated in advance.
Software requirements/dependencies
Currently Visual Components together with the IFF Safety Planning Tool are required for setting up the cell layout. In the future, other tools for this process might be available.
To activate all features and use optimized calculations, a valid license can be purchased from Fraunhofer IFF.
The component runs as a Linux Docker Container on Linux and Windows hosts.
Hardware requirements
PC, no special performance features
Security threats
no known issues
Privacy threats
no known issues (no cameras, no collection or processing of personal data)
Execution place
PC next to robot cell or server. Other options possible (e.g. Private cloud).
Deployment instructions
Deployment instructions will be available on the Shop4CF Docker Registry (docker.ramp.eu)
User interface
No user interface available.
The REST API is documented using Swagger.
Supported devices
PC, Docker
User defined scenarios (non-technical) and relevant pilot cases
When operating a robot cell that uses speed and separation monitoring for safety purposes, you have to check if a given trajecory is safe. If the trajectories are fixed or worst case trajectories can be defined, the operator can check them during design phase (e.g. using the IFF Safety Planning Tool). If trajectories can change (e.g. when using dynamic motion planing), the ASA component allows you to check if a trajectory is safe and the separation distance can be monitored by the sensor configuration. If a trajectory is not safe, the user can calculate a different one or reduce the robot’s speed until all safety conditions are met.
The ASA component ist utilized in the Siemens Use Case 1, where collision-free trajectories are calculated at run-time.
The review of the risk analysis supports a safety expert in identifying hazards and estimating risk. The responsible human designer is guided through the formalized process of identifying new hazards based on identified or manually captured system changes (e.g. part changes including geometry and payload; robot changes including speed, reach, tooling; environmental changes including new tables, fencing, etc.). Application highlights where the existing risk estimation requires updates.
Main non-functional requirements
Bidirectional network access from RA server to FIWARE server (if not used as standalone tool)
Port forwarding, firewall configuration
Secret injection via files or environment variables
Software requirements/dependencies
Container runtime, e.g. Docker
Hardware requirements
Server: about 50 MB free RAM / < 1GB disk space.
End-user device: min. 1280×720, ideally 1920×1080 screen, modern web-browser, about 500MB free RAM
Security threats
Production application must use HTTPS reverse proxy. Secrets must be generated and injected securely. Network access should ideally be limited. Additional requirements apply depending on threat model. E.g. when using FIWARE integration and local user management with enabled self-registration, the server must be located inside secure network (on the same permissions level as FIWARE server) due to granted lateral access to FIWARE server (alternatively use authentication via identity server for managing trust).
Privacy threats
Process critical data could be part of the data sent between client and server and this could put partners’ data at risk.
Execution place
Gateway, private cloud (meaning in pilot premises), cloud, etc.
Unrestricted
Deployment instructions
Deployment instructions are a part of the overall documentation and are included in container and available as separate PDF.
User interface
Browser based interface with multiple tabs, available in German and English (with possibility to add additional languages). Documentation (included in container and available as PDF), video see below.
Supported devices
The server host is unrestricted. The end-user device should preferably be laptop or PC (due to size and amount of information on the screen).
User defined scenarios (non-technical) and relevant pilot cases
See ”Main functions”. Additionally, in the FIWARE configuration monitoring mode the RA component will track the resource assignment to the Process on FIWARE server and either automatically communicate previous approval of the configuration or facilitate review by the safety expert.
The aim of the FLINT platform is to facilitate the incorporation of current/future wireless IoT devices (sensors/actuators) in a factory or shopfloor setting, as well as the required local wireless IoT communication infrastructure to connect such devices (e.g. LoRa gateways, BLE gateways). This component requires horizontal integration. At the left side, it will make use of adapters to interact with wireless IoT devices and long-range wireless communication equipment. At the right, after performing the required data transformations, it will either represent the IoT device as a LwM2M compliant device that can interface in a standardized way with a LwM2M back-end platform (for instance the open source Leshan platform) or deliver the data in a suitable format to a broker (e.g. FIWARE context broker).
Main non-functional requirements
N/A
Software requirements/dependencies
– Dependency on data formats used by IoT devices / used wireless IoT infrastructure, which requires the 1-time design of suitable input/processing adapters. Similar dependency for output adapters in case no processing to LwM2M.
– Docker: adapters realized as Docker containers. Implementation of adapters can be done in any language.
– MQTT broker: for the information exchange between adapters
– LwM2M processing adapters: dependency on Anjay, a C client implementation of LwM2M
Hardware requirements
Server/cloud platform supporting deployment/management of Docker containers.
Security threats
Currently, the internal communication between the adapters and MQTT broker is not secured. However most deployments are done on a secure company network, so the security risk should be limited.
Privacy threats
Privacy threats will depend on the type of data that is collected by IoT devices)
Execution place
Private Cloud
Deployment instructions
Deployment instructions can be found on https://github.com/imec-idlab/flint. Customization will be needed depending on the IoT devices/infrastructure to be used/deployed.
User interface
Dashboard for monitoring data received from / sent to IoT devices (see screenshot below). However, the user interface is not the core of the component, as it can operate without any UI.
Supported devices
The aim of the platform is to be extensible to support a wide range of wireless IoT devices and technologies.
User defined scenarios (non-technical) and relevant pilot cases
Industrial monitoring, asset tracking, environmental monitoring, etc.
Supporting human workers on the shop floor by giving them real-time wireless control over aspects such as process management, interactions with robots, collecting sensor data.
Main non-functional requirements
N/A
Software requirements/dependencies
Linux OS, GNU toolchain, Xilinix Toolchain
Hardware requirements
SDR board (e.g. Xilinx ZC706 + FMCOMMS2/3/4 or other compliant board see https://github.com/open-sdr/openwifi )
Security threats
WPA2 encryption is available and should be sufficient. Of course, a network firewall is necessary.
Privacy threats
All data transmitted over the same WiFi network can be seen by all connected clients. So SSL encryption might be necessary.
Execution place
Private cloud (meaning in pilot premises)
Deployment instructions
All information and source code is available on https://github.com/open-sdr/openwifi
User interface
– Developer: interact with openWIFI through linux WiFi driver (e.g. ath9k), and interface to openwifi specific components with a command line program (“sdrctl”)
– User: openWiFi acts as a regular WiFi access point
Supported devices
All 802.11 WiFi enabled devices are supported (smartphones, tablet, laptops, embedded WiFi hardware, WiFi sensors, …)
User defined scenarios (non-technical) and relevant pilot cases
The Wi-POS system is able to accurately determine the position of AGVs, robots or equipment on the SHOP floor. Positioning workers is also possible, but might be difficult for privacy reasons.
Its goal is to enhance the safety of humans and support human workers on the shop floor ( Relieving repetitive and hard tasks such as moving equipment).
Main non-functional requirements
N/A
Software requirements/dependencies
Standalone software (full-stack) is deployed on anchor nodes and mobile tags.
Hardware requirements
Dedicated embedded hardware is needed.
Security threats
No encryption on wireless sensor network, so positions could be retrieved. The server that collects the hardware should be protected by a network firewall.
Privacy threats
If the position of humans is logged, then privacy concerns might arise.
Execution place
Private wireless network is setup by the Wi-POS system (on-site)
Deployment instructions
Deployment instructions can be obtained on request. The instructions will vary depending on the location and the use case. System should be plug-and-play.
User interface
Not available. Measured coordinates are only pushed to FIWARE context broker.
Supported devices
Only dedicated hardware (proprietary) is supported for now. Other UWB enabled hardware (e.g. new iPhone) might be supported in the future.
User defined scenarios (non-technical) and relevant pilot cases
Determining the position of AGVs on the SHOP floor to allow navigation through the factory.
Locating important equipment on the SHOP floor.
Defining safe zones around robots to avoid human injuries.
Automated inventory management.
Code will not be publicly available.
[elementor-template id="14644"]
Supporting human workers in predicting or preventing potential failures and incidents; supporting human workers in planning services and repairs.
Main non-functional requirements
(non) functional requirements, especially with respect to real-time responsiveness
Software requirements/dependencies
Linux, Windows, MacOS
Hardware requirements
Our app consists of several components (docker images). To use it in a comfortable style, we suggest at least 16RAM, and 60gb of disk space (in order to store OracleDB, InfluxDB, Kafka, Orion). OracleDB spaces is increasing in time in estimate way +/- 0,045MB per one operation (unstable value).
Security threats
Due to Volkswagen security politics, we exchange data between microservices with JWT token. All users, who wants to use our app, should be logged in via LDAP server. In develop mode we can use test user. All address and ports are protected. App is running in internal network without access to external network but it is not required.
Privacy threats
LDAP authentication is required.
Execution place
Everywhere when docker images can be hosted – there is no limitation.
Deployment instructions
Instructions can be found in our bitbucket repository with whole project.
Provided upon request.
User interface
The application has a graphical user interface for viewing current waveforms and checking potential anomalous waveforms. The user can view, sort and filter the results. He can also report his own anomaly event if he deems it necessary. The application’s capabilities are limited only to viewing waveforms and metadata resulting from current waveforms. Data is obtained from the backend using REST API and websockets.
Supported devices
Laptop, PC
User defined scenarios (non-technical) and relevant pilot cases
Currently two potential use-scenarios for this component have been identified:
1. Prediction of failures of a car body lift used in production process. Identification of repair time – which should result in reducing unnecessary interventions by human workers and, at the same time, in preventing future failures.
2. Prediction of repair and maintenance (e.g., cleaning) interventions in parts of the paintshop. Detection of dependencies between observed changes in measurements and quality of paint structure. Again the purpose of this scenario is to reduce unnecessary interventions by human workers and, at the same time, to prevent failures.
Available upon request
Supporting human workers at production lines by monitoring quality.
Main non-functional requirements
Input requires template image and test sample.
Software requirements/dependencies
OpenCV, Python, Flask. The component can be run with Docker image, then no additional installations are required.
Hardware requirements
A standard PC have enough computing power for this component, as no machine learning strategy was used.
Security threats
The components’ API should be accessible only in local or private network.
Privacy threats
None
Execution place
PC at the operator stand (in pilot), however this can be also run on the remote server.
Deployment instructions
FIWARE must be already up and running. To deploy the component, you can run the docker image directly or go to project folder and run `docker-compose up -d`.
User interface
There is no user interface – other components should call API function.
Supported devices
In Bosch Pilot AR-CVI was used as GUI.
User defined scenarios (non-technical) and relevant pilot cases
Dedicated application for laptop or desktop PC.
Supporting human workers on production line.
Main non-functional requirements
AR (augmented reality) application precision of measured distances an location of simulated objects – we expect that AR application accuracy will be between 1cm-5cm per 5m
Software requirements/dependencies
Simulation module based on “LogABS” for Windows
AR application – native app for Android or iOS
Hardware requirements
Intel i7 CPU with 16GB RAM or better for LogABS
Mobile device compatible with AR Core or AR Kit for AR application
Security threats
No specific security requirements.
Privacy threats
No specific privacy requirements.
Execution place
PC and mobile device.
Deployment instructions
Instructions can be found in the code repository with whole project.
Otherwise provided upon request.
User interface
Windows GUI for planning and executing logistic simulation, AR app for planning factory locations and equipment.
Supported devices
Laptop, PC, mobile device compatible with AR Core or AR Kit
User defined scenarios (non-technical) and relevant pilot cases
Planning new, safe AGV routes for redesign factory outline e. g. new product is about to enter into production and factory needs to be redesigned to maintain new storage spaces, new machinery etc.
Internal repo, available upon request
Assistance and training for operators during customised product assembly process, and maintenance operations including recognition of objects, sequence of operations and AR guidance to operators.
Main non-functional requirements
– Mobile devices should be compatible with ARCore/ARKit frameworks.
– Wi-Fi connection is required at the shop floor.
– Minimum brightness levels are required for the AR vision algorithms to work correctly.
– If workers are required to wear gloves, specific mobile device models will be required.
Software requirements/dependencies
WebXR, Django, ARCore, Arkit, Microsoft Mixed Reality toolkit
Hardware requirements
Mobile devices/HoloLens. A server
Security threats
The component need an account management. There is one already implemented but in case to be integrated on RAMP or another platform, further adjustments should be needed.
Privacy threats
None
Execution place
Private cloud provided by TECNALIA with access to pilots. In case there is a need for pilot sites to deploy the component there, it can be done using dockers.
Deployment instructions
In RAMP marketplace and upon request.
User interface
– Interface for the developer: an editor to create AR content in an easy way to guide operators in the shop floor.
– Interface for the Operator (customised by the editor): the AR guidance to be visualised with a mobile devices/HoloLens through different steps and different objects (2D/3D objects, images, video, documents, animations, etc.).
Supported devices
Laptop + mobile devices/HoloLens
User defined scenarios (non-technical) and relevant pilot cases
AR guidance to operator during the assembly of the base plate in collaboration with a robot in the Siemens Use case.
– The shop floor manager/developers add a new manual to the system using the editor, adding all the multimedia assets (3D files, pdf, videos, photos, etc.) and defining, step by step, the entire manual.
– The shop floor manager, once the manual has been added, defines through the editor what type of trigger will activate the augmented reality display. In this case, through the Context Broker we will see in a task for HoloLens is raised to show the different steps of the manual.
– Then he publishes from the editor that manual, so that it can be consumed by HoloLens, defining the type of users and roles that will have permission to do so.
– From that moment any worker of the plant with permissions working on the assembly of a base plate will be able to trigger with his device the visualization in augmented reality of that manual based on the status of the collaboration with the robot supporting him/her.
Not available. As stated in the CA, the tool is not open source so no code will be provided.
Supporting human workers on the shop floor: Tele-assistance in maintenance for long distance workers
Main non-functional requirements
An Internet connection (Wi-Fi connection recommended).
Android app: ARCore framework is needed to use the AR functionalities.
Browser client: right now, the best/recommended browser to use it is Firefox.
Depending on the resolution of the real time streaming, the bandwidth has to be in accordance with it, with higher resolution is recommended to use Wi-Fi-connection.
Software requirements/dependencies
Server side: NodeJS, Websockets, Express
Client side: ARCore
Hardware requirements
Server side: UNIX/Linux environment, with enough bandwidth for the number of users to use it
Client side: smartphone device compatible with ARCore; browser can´t be Chrome (Firefox recommended)
Security threats
The component need an account management. There is one already implemented but in case to be integrated on RAMP or another platform, further adjustments should be needed. Server side must use an SSL certificate to send the communication data with https protocol.
Privacy threats
Right now, the authentication part only requires having the client application to log in, so it is necessary not to share them with unknown users.
Execution place
Private cloud provided by TECNALIA with access to pilots. In case there is a need for pilot sites to deploy the component there, it can be done using dockers.
Deployment instructions
In RAMP marketplace and upon request.
User interface
Android app client
Supported devices
Laptop + mobile deviLaptop + mobile devices/HoloLens/HoloLens
User defined scenarios (non-technical) and relevant pilot cases
UC2 in Arcelik for equipment maintenance.
A worker that needs support with any type of physical component or machine, call to an expert colleague in the field. In order to do that, the worker opens the application installed in his smartphone and calls the previously connected expert. The application connects with the expert sharing the back camera of the worker and the frontal camera of the expert. The worker scans the component area or the machine area, and the expert draws through the mobile screen creating indications, as a drawing mode to the worker, giving an augmented reality support. When the support is finished, both exit the application.
Not available. As stated in the CA, the tool is not open source so no code will be provided.
It will allow workers to be trained in the operation of a machine, or manufacturing line for example, through the use of virtual reality. With this web tool it will be possible to create and consume immersive VR experiences (with glasses) oriented to training.
Main non-functional requirements
If video 360 are big files web creator tool will require a PC/Laptop with dedicated (and “good”) graphic card. We will specify more.
Wifi connection is required during training experience creation and consumption.
PC/Laptop and VR HMD should contain a WEBXR compatible browser.
Software requirements/dependencies
WebXR, Django, WebGL
Hardware requirements
PC/laptop and VR HMDs
A Server
Security threats
The component need an account management. There is one already implemented but in case to be integrated on RAMP or another platform, further adjustments should be needed.
Privacy threats
None.
Execution place
Private cloud provided by TECNALIA with access to pilots. In case there is a need for pilot sites to deploy the component there, it can be done using dockers.
Deployment instructions
In RAMP marketplace and upon request.
User interface
N/A
Supported devices
PC/Laptop + VR HMDs
User defined scenarios (non-technical) and relevant pilot cases
Training the operators when using new machines. At this moment we do not have any pilot interested in the component.
Not available. As stated in the CA, the tool is not open source so no code will be provided.
MPMS includes the functionality to design processes and describe agents, and execute in automated way the processes by assigning activities to agents. It provides orchestration of activities in a global level, i.e., covering all work cells/production lines of a factory.
Main non-functional requirements
As a logical functional component, MPMS shall be able to:
Automatically execute a sequence of activities
Monitor agents’ availability
Monitor agent’s performance including at least task estimated completion time and task actual completion time
Monitor process current state
Provide right information to agents to perform a task
Handle exceptions on agent, task and process level by halting/resuming their activities and initiating out-of-normal action processes
(re-)allocate appropriate agents to perform a task based on abilities, skills, authorizations, cumulative workload, overall manufacturing system status and availability
Re-allocate agents in response to external events such as safety alerts or sensor failures.
As a software technical component, MPMS shall be able to:
Provide a modeler application to model processes
Provide a process engine to automatically enact process models
Provide tasklist applications to deliver tasks to human operators
Support integration to custom UIs as tasklist applications
Provide integration to local components to deliver tasks to robotic agents
Support various platform environments
Support various DBMS
Be deployed both on premise/cloud
Provide security/authorisation mechanisms
Integrate to middleware/context broker and other components
Support web services
Support REST/JAVA APIs
Support SOA/Interoperability (NF)
Be robust (NF) be runtime scalable (NF)
Be easy to use by both process modelers, developers and end users (e.g., human operators) (NF)
Software requirements/dependencies
MPMS is built on Camunda Platform 7.15.0, Community Edition and runs in every Java-runnable environment. It can support the following environments:
Container/Application Server for runtime components
Apache Tomcat 7.0 / 8.0 / 9.0
JBoss EAP 6.4 / 7.0 / 7.1 / 7.2
Wildfly Application Server 10.1 / 11.0 / 12.0 / 13.0 / 14.0 / 15.0 / 16.0 / 17.0 / 18.0
Databases MySQL 5.6 / 5.7
MariaDB 10.0 / 10.2 / 10.3 Oracle 11g / 12c / 18c / 19c
PostgreSQL 9.4 / 9.6 / 10.4 / 10.7 / 11.1 / 11.2 postgres:14-alpine (Docker image)
Microsoft SQL Server 2012/2014/2016/2017
H2 1.4
Adminer:4.8.1 (UI for DB management) (Docker image)
Web Browser
Google Chrome latest
Mozilla Firefox latest
Internet Explorer 11
Microsoft Edge
Java
Java 8 / 9 / 10 / 11 / 12 / 13 (if supported by your application server/container)
Java Runtime
Oracle JDK 8 / 9 / 10 / 11 / 12 / 13 IBM JDK 8 (with J9 JVM) OpenJDK 8 / 9 / 10 / 11 / 12 / 13
Openjdk:11.0.13-jre-slim (Docker image)
Camunda Modeler
Windows 7 / 10
Mac OS X 10.11
Ubuntu LTS (latest)
The Camunda Community Platform is provided under various open source licenses (mainly Apache License 2.0 and MIT). Third-party libraries or application servers included are distributed under their respective licenses. Detailed info on licences is provided in T7.2.
Hardware requirements
For deploying MPMS in a local desktop PC, not any special requirements are needed. A powerful processor, a lot of RAM memory and a decent graphics card is sufficient. The following specs should do:
Processor: Intel Core i7-7700 @ 3.60GHz / Intel Core i7-6700K @ 4.00GHz / Intel Core i7-7700K @ 4.20GHz / Intel Core i7-8700K @ 3.70GHz
Storage: SATA 2.5 SSD (e.g. 256 GB) RAM: 32GB DDR4-2133 DIMM (2x16GB) (well, even 16GB will not be a problem)
Graphics Card: Any modern standard graphics card
Monitor, keyboard and mouse are essential. Touchscreen for operators might be handy.
A laptop could also work:
Processor: Intel Core i7-7700HQ @ 2.80GHz / Intel Core i7-6770HQ @ 2.60GHz.
Security threats
MPMS connects to DBs with password access.
Users of web applications have access with password.
Privacy threats
Agents (and specifically human operators) should be described with due diligence wrt privacy data.
Execution place
MPMS shall be deployed on local PCs on premises. It can also be deployed on a cloud server (e.g., on TUE premises) but extra security is required.
Deployment instructions
Deployment instructions and user manuals can be provided upon request.
Can also be uploaded on RAMP if there’s a specific repository.
User interface
Three different types of users:
Process modelers
They use the Modeler application to model manufacturing processes (with BPMN 2.0)
Application developers
They turn the process models designed by the process modelers into executable process models (i.e., the ones that the Process Engine can interpret and enact).
Also, they can build custom applications (e.g. tasklists, cockpits, smartwatch apps)
(Human) Managers and operators (end-users)
Managers use the Cockpit/Dashboard and Admin applications (default or custom) to see the processes state and manage the users of MPMS
Operators use the Tasklist applications (default or custom) to receive tasks and provide input (e.g. task completion confirmation)
For each type of user, there are manuals and webinars to provide instructions.
Supported devices
Modeler runs on PC/Laptop (see SW/HW requirements above)
Process Engine runs on PC/Laptop (see SW/HW requirements above)
Web applications run on PC/Laptop/tablet/smartphones (additionally, a prototype tasklist application has been built for smartwatch)
User defined scenarios (non-technical) and relevant pilot cases
MPMS can be used in any pilot/open call for:
process modelling (for bottlenecks identification and enabling automated execution),
dynamic agent allocation,
process orchestration,
automated process execution,
integration to other IS (e.g., ERP) for getting the right information and providing it to agents during execution,
process status monitoring,
task monitoring for job safety and quality of human operators,
re-allocation of agents when job safety and quality criteria are violated,
etc.
The component provides visual support in manual assembly and inspection tasks. Human worker is guided by the visualized instructions while performing the associated tasks. The component can project the instructions on a surface or a screen. This provides a clean and structured working environment.
Main non-functional requirements
The Fiware messages are checked at 2 Hz.
Software requirements/dependencies
Ubuntu:
Ubuntu 18.04 or later (for running on Ubuntu)
Docker version 20.10.6 (Previous versions later than v19 are also supported)
Windows:
Windows 10
WSL 2
Docker version 20.10.6 (Previous versions later than v19 are also supported)
Windows X Server (for running on Windows)
Hardware requirements
A screen with at least 1920×1080 (HD) resolution support
A projector [Optional] with at least 1920×1080 (HD) resolution. Front surface mirrors might be necessary for projecting down to a table. High brightness according to the illumination of the environment (~4400 Lumens)
A PC or laptop (Preferably a 4-core CPU with 8 GB RAM, 15 GB HDD space)
Security threats
No specific security threats
Privacy threats
No specific privacy threats.
Execution place
The component should be executed in the assembly cell pc (local deployment). Necessary files can be mounted to the Docker container from the host machine or from a remote machine.
Deployment instructions
Currently in the public Github repository: https://github.com/emecercelik/ar-cvi/ Later the instructions will be available in RAMP.
User interface
The instructions are displayed in a full-screen mode.
User can provide inputs to the component using the buttons on the screen if defined with the display messages (templates).
A screen view is provided below:
Instruction images, PCB quality check outputs (provided by another component), a written instruction, and buttons (at the bottom).
For a developer demonstration refer to the following video:
https://syncandshare.lrz.de/getlink/fiC2jAGB9vBsmaRqtVVmzEg4/ar-cvi_demonstration.mp4
Supported devices
PC, Laptop, Projector, Screen
User defined scenarios (non-technical) and relevant pilot cases
Support on the assembly cell for the human worker by displaying manuals of an assembly or inspection task. The operator can provide inputs to start the inspection (provided by another component).
It supports the process management improving the interoperability of the system. It addresses the ability of the system to be standard interoperable.
Main non-functional requirements
In the case of OPC UA communication, the server must be configured and deployed. The Node ID (s) must be provided to be able to read the information in real time.
Software requirements/dependencies
Node JS and/or Java but it can be developer also with other programming languages.
Docker and Docker compose
Orion LD Context Broker
Hardware requirements
16 Gb RAM, 10 Gb HD
Linux – Ubuntu
Security threats
It wraps REST APIs in another descriptor, the component does not manage the security of the APIs described. The server that serves the descriptor will be secured through HTTPS and certificate.
Privacy threats
No specific privacy threats.
Execution place
The component should be executed in a local PC on the shopfloor
Deployment instructions
We provide a docker container with a configuration file.
User interface
In this version a Command Line Interface has been implemented. It is expected to develop a graphical interface for the next version.
Supported devices
PC/Laptop
User defined scenarios (non-technical) and relevant pilot cases
A pilot or a SHOP4CF component owner wants to provide to the external word some functionalities he wants to test with external users (for instance for open calls), he builds a WoT (Web of Things) interface with our component that wraps its component to be used by third party developers.
The main functionality of this supporting tool is to allow robotics applications designers and system integrations to also analyse the required size of the minimum separation distance for cobot applications featuring Speed and Separation Monitoring. The minimum separation distance is determined by a combination of the following parameters:
The supporting tool integrates user interfaces for specifying these values into the overall simulation environment and performs the calculations to show the instantaneous and accumulated minimum safety distances, in 2D (projected on the floor) and in 3D (around the robot).
The main functionality of this supporting tool is to help designers of collaborative robotics applications ensure the safety of their applications. It provides examples of best practices, how-to guides, information related to legal issues, and information related to relevant standards and regulations.
The main functionality of this supporting tool is to help system integrators and designers of collaborative robot applications validate the application at a system level in the form of a test, to prove that the system is safe. This support is in the form of a step-by-step guide (a pdf document), specifying which relevant information the user needs, which measuring equipment to use, and how to execute and evaluate the test results. There are a wide variety of different protocols available, as these are specific to robot device type and safeguarding strategy.
The main functionality of this supporting tool is to to facilitate the incorporation of innovation and creative problem-solving methodologies into industrial processes. The tool combines common visual resource management tools (such as Kanban) with collaborative creation methodologies such as design thinking. In this way, the iterative refine process incorporated by these collaborative techniques is expected to increase daily production problems of workers, incorporating, in a natural way (without extra work), skills, and tools to improve. The expected results systematically introduce this iterative process for the solution of daily production problems, in a way that workers naturally understand the specific needs of users of services / products and redefine problems in an attempt to identify alternative solutions more productive, more attractive and more efficient, without a high investment in resources.
The main functionality of this supporting tool is to provide insight into the composition of the flexible manufacturing process in terms of the availability of resources and correctness of the inputs and outputs flow.
This is achieved by facilitating the step-by-step gathering of information related to manufacturing tasks and potential actors involved in the production process. The outcome may be used to assign workers to tasks fitting their qualifications, allocate robotic agents to tasks that may harm humans, and adequately plan upskilling of the personnel. MPDesign is also an excellent means to document the manufacturing process’s design for future reference.
The main functionality of this supporting tool is to provide a set of recommendations (based on interaction patterns and design guidelines) that designers and developers should implement in their interfaces to ensure usability, accessibility, and UX aspects. This tool would be applicable to designers/developers at an early stage of product development as well as evaluating the interface of an already developed solution.
The users would have to answer a questionnaire covering relevant aspects of the human-machine interactions, such as the type of devices to be used (e.g., tablet, PC), the type of technology (e.g., Augmented Reality, Web), and the target user profile.
The main functionality of this supporting tool is to give solution providers or developers a possibility to collect feedback from users of a technical solution with a short, ready-made questionnaire. The questionnaire is based on a design and evaluation framework of the SHOP4CF project and it addresses seven themes: user experience, usability, user acceptance, usefulness, ergonomics, safety, and ethics. For solution providers, the tool provides an easy way to collect holistic information of their solutions, and for workers, it provides an effortless way to participate in the design of work tools.
The use of the questionnaire can be started by creating a project, adding an expiration date to the ready-made questionnaire, and sending the questionnaire link to the respondents. After four respondents have answered the questionnaire, it is possible to see the results (e.g., an overview of the human factors topics and detailed information for each question). The results are automatically updated when the questionnaire receives more responses.
The main functionality of this supporting tool is to supports the rapid assessment of various factors affecting human performance and part quality in industrial operations. In particular, the assessment reviews factors such as time deviations (faster or slower executions), assembly sequence modifications or errors (forgotten parts, incorrect operations, etc.). Beyond the rapid assessment of human performance factors, the tool proposes the potential areas of improvement. A SSH (social sciences and humanities) professional is required for full assessment before engaging further actions.
PATH is a web-based application that can potentially run from telephones, tablets or PCs. It enables workers and managers to provide feedback about the tasks they perform, the working conditions or their attitudes. Privacy aspects have been included into the tool: the questionnaire is anonymous and only statistical results are provided.
Your are now accessing the Getting Started Guide
