AI.Society: Robotics and AI applications of the future

Automated guided vehicles (AGVs) are an integral part of automated intralogistics systems. Even though, AGVs can be used effectively for the transport of standardized load carriers, they are not capable of performing other activities during transport to ultimately add value to the transport process. The aim of the project “Adaptive process planning using mobile pre-assembly systems” is to expand the range of tasks performed by AGVs to include complex handling and pre-assembly processes, which can be performed while the vehicle is in motion. By combining a mobile robot with two collaborative manipulators tasks currently performed manually on mobile platforms can be automated. In turn, stationary work for humans even in flexible production processes is enabled. The AI-based control system is designed to react flexibly to load peaks and errors.

Using technology developed in the Horizon 2020 European project I.AM. (Impact-Aware Manipulation), we show one of the use cases that could benefit from the pursuit of intentional impacts in robotic manipulation tasks. This concerns the rapid grasping of objects using a robotic dual-arm setup, where contact between the end effector and parcel is established with nonzero relative velocity. In our demo, we show how this idea can help decrease the cycle time for robotic depalletization, using two Franka Emika robots with custom silicon flat end effectors, and a set of test parcels with varying weights, up to 2 kilograms

Roundpeg presents the first robot with an integrated human detection system that can work fence-free at speeds that match industrial robots.

When robots learn by observing humans, it is usually not enough to just track the human's motion, but further information about the human's interaction with the environment must be determined. The demo shows how a manipulator can successfully determine dynamic parameters of objects by interacting with them. Mass, center of gravity and inertia of the object are of great importance for a successful execution of the task to be learned.

Network management in AI Factory. We will show how AI can provide deterministic guarantees in network latency in a factory.

Digital twins in the KI.FABRIK are created using engineering data as well as data from the operation. A prototype digital twin for the WITTENSTEIN gearbox was developed to describe an asset electronically in a standardized manner by relying on the Asset Administration Shell. The resulting Digital Twin enables the exchange of asset-related data among industrial assets and between assets and production orchestration systems or engineering tools.

CENTRAL.AI and automatic planning & reconfiguration enable competitive automated manufacturing (Production-as-a-Service) in Bavaria even for batch size 1 for complex products.

This demo aims to approach the problem of deformable objects manipulation in the production process, which currently remains to be handled only manually. Visual and tactile information are provided for detection and tracking, and the wiring task is achieved autonomously through collaboration of two panda robots.

Hence, a powerful Digital Twin provides an innovative way to improve production management and accessibility for human operators.

The mobile robot demo in the AI factory showcases the capabilities of an advanced autonomous robot in a manufacturing environment. The robot is equipped with cutting-edge AI technologies and sensors that enable it to navigate through the factory, identify objects, and perform tasks with high precision and efficiency.

Collective Learning is part of the KI.FABRIK. It learns the skills needed for assembling or fabricating object inside a industrial environment. The robot collective consists of 100 robots that are learning a wide range of insertion tasks. After an adequat number of tasks are learned, the system has enouth knowledge to imideatly solve a new previously unknown task. Collective Learning provides the lowlevel parametrisation of skills needed by the Central AI of KI.FABRIK to produce new products.

We leverage state-of-the-art generative Artificial Intelligence (AI) algorithms and Additive Manufacturing (AM) to optimize product design for small production sizes. Our program enables seamless interaction between human designers and AI, facilitating a natural and intuitive design process. By iteratively selecting AI-generated design suggestions that meet specific criteria, designers can guide the AI through the generation process. The use of AM allows us to print and immediately deploy the generated design. This process is fast, cost-effective and highly adjustable, making it ideal for the development of customized products or single-use equipment.

With the Robosphere, Hydrabyte presents a system for optimal near-field movement planning of manipulator groups. Hydrabayte's unique multi-robot technology allows multiple assembly processes to be carried out in parallel in the smallest of spaces. Together with Olive Robotics GmbH, the assembly of robotic systems using the modular robotic components from Olive Robotics will be demonstrated at the booth.

Mimetik develops solutions for human-centric automation and robot/machine control. The know-how of Mimetik is an IIoT sensor of human activity. It is based on AI-powered industrial-quality data gloves that track hand/finger movement and extract an intelligence from the motion data. Proprietary AI achieves high accuracy and quick deployment. With a focus on industrial applications, Mimetik improves the training of workers at assembly lines, optimises efficiency and ergonomics of workplaces, and improves automated quality control.

The demo is aimed to provide an overview of how the Meshmerize network can help to quickly provide network coverage in the areas that pose a challenge for the cellular networks at the moment. We are excited to show in our scenario just how easy it is to deploy a reliable network and maintain strong connectivity to greatly benefit firefighting efforts. Not being dependent on cellular network infrastructure means it can be used in a remote forest, as well as in a buzzing city. UAVs and mesh can give firefighters an extra pair of eyes to help fight the flames, ensuring safe site inspection, prioritized and strategic disaster response, efficient search and rescue operations, information sharing, and keeping track of team members locations.

The Collective Learning demo presents the superior learning speed of an interconnected group of robots compared to a single robot. Where a single robot would take several hours to learn a task like insertion, the group can benefit from the experience of each member – and thus de-velop a superior solution faster.

Future of Workshop is about understanding and testing skill learning, human-robot collaboration, force-impedance control, and motion planning.

The Remote Robotic Teleoperation demonstrator addresses the challenges of a human-centered teleoperation station for production tasks such as assembly, which are still largely performed manually. Visual and bilateral haptic feedback in human-robot interaction enables remote support of assembly tasks.

Digital Twins in the KI.Fabrik are created using engineering data as well as live operational data. A prototype Digital Twin of the Wittenstein gearbox was developed to create a standardized electronic asset description using the Asset Administration Shell. This enables data exchange among industrial plants as well as between plants and production orchestration systems or engineering tools. Due to the great manual effort associated with extracting information from the available engineering documents, this demonstration presents a tool for automatic Asset Administration Shell generation from .stp files (CAD) of the Wittenstein gear. Additional automated information extraction tools are developed for formats such as .pdf and .xml. This is the first step towards an industrial real-time Digital Twin of the entire KI.Fabrik setup.

Automated Cable Insertion targets the issue of manipulating deformable objects in the production process – a problem that currently requires manual solutions. Visual and tactile information is provided for recognition and tracking purposes; and the wiring process is handled autonomously by two collaborating Panda robots.

In nature, the movement of quadrupeds is achieved through combined actions performed by the spine and legs. This provided the inspiration to design a biomimetic mouse with a lateral flexing spine to optimize its locomotion. The mouse robot benefits from the regular lateral flexion of a soft spine – enabling enhanced agile locomotion when moving through a maze scenario.

The start-up Olive Robotics will present smart and modular robot components with native ROS support for easy prototyping as well as flexible extension and augmentation of ROS-based robotic and automation systems.

RobCo has a vision to make robots available to everyone. With their tailored all-in-one robot solution, they can accompany SMEs during their transformation process towards Industry 4.0.

No one likes to spill a coffee — neither do robots. At this demo, we want to demonstrate that robots can transport and manipulate liquids with safety guarantees. That is, we present robots with spill-free capabilities! We would also like to invite all participants to interact with and telecontrol our robotic arms safely transporting your favourite drink. Would you allow a robot to carry your favourite beverage? Visit our booth and our robots at Automatica Munich_i 2023!

Real-time visualization of muscle efforts during a series of movements in a human musculoskeletal model.

For ubiquitous use in modern households, robots need to learn new skills, represent what are the skill's relevant features, and adapt the skill to their own robot kinematics. The goal being to execute a skill with (other) objects in a new environment with different constraints or geometry than the environment in which the skill was learned. We are proposing a two-phase system for first learning and consequently adapting skills. In the learning phase, a model- and knowledge-based approach is compared to an experience-based approach by their representational qualities. And in the adaptation phase, a robot executes one of the learned tasks (i.e. skill sequence) on its own.

We are presenting new skills of GARMI that have been developed since our presentation last year. In addition to enhanced telemedicine skills, we will show how GARMI can function as a teleoperated avatar using various input devices, such as our self-developed exoskeleton, Omega haptic controller and a full VR environment. We will also demonstrate some of the autonomous service features of GARMI.

Exoskeleton to remotely control the humanoid avatar GARMI and interact with patients.

The Telemedicine demo will show how service humanoid robot GARMI can be operated in using a haptic input device (Omega.7) or HTC vive controller.

Hands play an important role in human life for prehensile, proprioceptive, and communication purposes. Compensating for losing fine and coordinated function of upper extremities with prostheses is a medical, technological, psychological, and social challenge. Even though artificial limbs open up the prospect of restoring some missing capabilities, there is still a wide gap between available commercial devices and the perceived demands of prosthesis users. This booth includes novel solutions for more intelligent neuroprostheses, both at their structure and in the control loop. We will present demos of commercially-available hands, explorative and bio-inspired designs through soft robotics, and different control modalities.

The research project AURORA aims at developing a mobile service robot for the OR wing. This „robotic circulating nurse“ moves freely within non-sterile OR areas and autonomously supports the surgical team with context-dependent tasks, such as the provision of sterile goods or the adjustment of medical devices. Thereby, today‘s severe shortage of nurses can be alleviated. The system is also capable of collaborating with other robots, which includes sterile handover interactions with the instrumentation robot SASHA-OR.

In the SASHA-OR research project, an intelligent assistance robot is being developed for the surgical operating room, designed for sterile use at the operating table. The system supports the surgical team with a flexible instrument and object management. By incorporating tactile sensor technology and computer vision methods, the robot can verbally interact with the surgical team, recognize, hand over and receive requested laparoscopic instruments and sterile goods. In addition to robot-human interaction, the system also aims at robot-robot interaction to enable sterile handover actions with the robotic circulating nurse AURORA.

We are excited to present our research on some of the latest advancements in catheter tracking technologies and their potential impact on future minimally invasive endovascular procedures. These technologies utilize advanced algorithms and a variety of imaging and sensing techniques to provide real-time guidance during catheter placement. The integration of these technologies has the potential to enhance the accuracy and efficiency of medical procedures, even in complex anatomical structures. By reducing errors and improving patient outcomes, catheter tracking technologies represent an exciting area of growth in medical technology. Their incorporation into medical interventions underscores the potential for artificial intelligence to transform the field of interventional medicine and improve patient care.

Patients in the intensive care unit require constant monitoring to prevent untoward incidents such as falling down and removal of catheters. However, due to the shortage of healthcare workers and an ever-increasing amount of paperwork, it is not possible to have a healthcare worker by a patient’s side all the time. To overcome this problem, we propose to use cameras to monitor patients by tracking their 3D skeletons. The actions of the patients can then be recognized using the tracked skeletons of the patients. With the information, a system can be built to notify healthcare workers if something has happened to a patient.

In mirror therapy, a mirror is placed into body-midline. Looking into the mirror, the reflection of the unimpaired hand is seen at the position of the non-seen impaired hand behind this mirror. Observing a seemingly normal moving impaired limb has a strong stimulating effect on the motor system and has been shown to be highly effective for treatment in stroke. You will be able to perceive this effect in the VR demo when movements of your virtual hand are controlled by real movements of your opposite hand. During bilateral or unilateral games you will become confident with using your avatar hand.

The Geriatronics Lighthouse Initiative in Garmisch-Partenkirchen develops adaptive humanoid service robots to enable the elderly to live independently. 3D navigation, environmental monitoring and object manipulation are used to create reliable patient monitoring tools that provide greater safety for the elderly and support caregivers. The goals: to preserve mobility and interpersonal interaction in old age as well as to support independent living through technical assistance systems with intuitive operation. Geriatronics shows the interaction between different robots: e. g. the prototypes by GARMI and its partners MUCKI and PARTI, which were designed for different tasks such as prevention, diagnostics, or rehabilitation.

While teleconsultation has evolved rapidly over the past decades, telediagnostics and telemanipulation are still lagging behind. To change this, ProteCT has developed a robot-based telediagnostics station enabling contactless examinations ranging from basic vital parameter tests to examinations requiring more intense physical interaction – such as throat swabs or palpation of the abdomen. The system is teleoperated by a physician and thus guarantees contactless interaction while ensuring a trusting doctor-patient relationship.

The goal of the AURORA project is to reduce the workload of OR teams while performing surgical procedures. The context-sensitive mobile assistant robot is able to move autonomously within the non-sterile area of the operating room where it can fetch required instruments or drugs etc. for the sterile OR team. To do this, the system combines elements of autonomous driving with domain knowledge and situational awareness, and augments these by way of intelligent object manipulation.

The startup Enari will showcase its MPER Lab motion profiling system that captures and monitors human musculoskeletal activity. Through the creation of a Digital Twin, human movements can be interpreted and predicted using special sensors. Furthermore, the sensors provide information about tendons and ligaments, muscle innervation, and the muscle/fat tissue ratio.