Events and Trade Shows

WS#32 From Lab to Fab: Advanced Control Strategies, RL, and Interfaces for Industrial Robotics

Moderator: José Saenz, Fraunhofer IFF

Main questions to be answered

• What to be considered when automating a process
• When accuracy is key – how to solve it?
• Impact of AI – benefits in a robot cell.
• Grip it – but how ?
• From simulation to factory floor: solving accuracy, safety, and real impact of AI in robotics

WS Content

The workshop explores three complementary strands of industrial robotics. First, it addresses end‐of‐arm applications from a requirements‐driven perspective: starting with pick, move, and lift tasks, it outlines how to select and size grippers and/or vacuum components based on material properties, surface conditions, payload, and cycle constraints, and discusses integration with robot and cobot arms with implications for reliability and maintainability. Second, it examines how Reinforcement Learning can be applied in production settings by training in realistic simulations and refining policies with real‐world data, explicitly confronting challenges such as sim‐to‐real transfer, data efficiency, safety, and process constraints, with the goal of informing a practical roadmap for future adoption on the shop floor. Third, it presents a robotic control approach that shifts the reference from individual motors to the robot’s center of mass and tool position, naturally compensating for axis interactions to reduce unwanted oscillations, improve motion quality, shorten development time and lower costs, while enhancing precision and stability in applications that demand high trajectory fidelity.

WS Organisation

Festo (Tobias Lidén) will present how to solve end‐of‐arm applications starting from the premise that a robot/cobot arm must pick, move, or lift material, guiding users in selecting suitable grippers and/or vacuum components. Adeptic Reply (a Reply group company) will show how Reinforcement Learning (RL) can be applied in industrial contexts, training robots in realistic simulations and refining them with real‐world data, while addressing challenges such as sim‐to‐real transfer, data efficiency, and safety to inspire reflections on future integration into industrial processes and a potential application roadmap. Part of those features have been developed in the content of IPCEI‐CIS program in which Adeptic Reply is actively involved.

Comau will introduce a new robotic control method that shifts the reference from individual motors to the robot’s center of mass and tool position, naturally compensating for axis interactions; this reduces unwanted oscillations, improves motion quality, lowers development time and costs, and enhances precision and stability, especially in applications requiring high trajectory fidelity. To deepen the exchange, the session will conclude with a moderated fishbowl discussion: the three speakers remain on stage with one or two empty chairs; audience members are invited to take an empty chair to ask their question or make a point, then return to the audience to free the seat for the next participant—keeping the dialogue focused, dynamic, and inclusive.

Intended outcome

• Technical checklist for selecting grippers and vacuum systems based on material, surface, payload, and cycle constraints.
• Insights on the challenges and opportunities of Reinforcement Learning (RL) in industrial robotics, including elements for a realistic adoption roadmap.
• Exploration of a new robot control method (center‐of‐mass & tool‐based), highlighting benefits in motion quality, stability, and cost reduction.

People actively involved (e.g. speakers, panelists, moderators)

Simone Voto, Roboverse Reply, Giuliano Giampietro, Adeptic Reply, Giuseppe Parlato, Comau, Tobias Lidén, Festo, José Saenz, Fraunhofer IFF (Moderator)

Topic Groups and/or Innovation Networks involved

Industrial Robotics