Leader
Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum.
Hosting Facilities
Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum.
HFS4HRC
Automation
Using high-fidelity simulations to train and develop a human-robot collaborative assembly process
Challenge: Co-bot refrigerator door assembly solutions
Despite major advancements in robotics, many manufacturing tasks are still done manually. One such job is fitting refrigerator door gaskets, which is physically demanding, repetitive and requires a relatively high degree of precision. This project introduces collaborative robots designed to assist operators with this physically demanding, repetitive work. By merging advanced robot control algorithms with an AI overseer, our system will actively infer and adapt to human intentions. To accelerate development, we leverage next-generation simulation tools to both design robot behaviors as well as generate the synthetic data needed to train our AI vision models.
Background
A major appliance manufacturer, an AID4SME partner, identified an optimization opportunity: automating the installation of flexible elastomer gaskets into refrigerator doors. Currently, this task requires two operators working in unergonomic positions to handle a high volume of parts. The physical strain directly impacts worker well-being, leading to lower job satisfaction and higher product defect rates. By adopting adopting modular, energy-efficient collaborative robots, the facility aims at improving operator ergonomics, enhance job attractiveness, and reduce manufacturing waste and environmental impact.
Solution
To address the challenge of assisted gasket installation we are developing a robotic system using collaborative robots equipped with AI perception models. By observing human motion, these perception models act as an intelligent overseer that anticipates operator intentions, enabling seamless, side-by-side human-robot collaboration where both can excel at their respective assembly tasks. Beyond the physical deployment in a laboratory, this project emphasizes how modern simulation software and open-source programming frameworks drastically reduce development and deployment timelines. Our workflow begins in NVIDIA Isaac Sim, where we will build a digital twin of the robot cell, including the collaborative robots, structural hardware, and cameras. Using ROS 2, we will develop the control algorithms and orchestrate the required assembly sequences. To train our AI models, we will leverage Isaac Sim’s photorealistic rendering and automatic data labeling to generate high-quality synthetic datasets. Finally, we will transition to physical hardware, utilizing out own b»controlled box to achieve tightly coordinated execution between the robots and peripheral equipment.
Objectives
- Achieve the target cycle time while solving the primary assembly challenge.
- Deploy a simulation-developed and simulation-tested, force-based control system for real-world gasket insertion.
- Train an AI perception model via photorealistic synthetic data to track human motion and infer assembly intentions.
- Align robot system behaviors dynamically with inferred human actions.
- Validate simultaneous dual-arm and periphery control using the b»control box (developed by b»robotized)
KPIs achieved
Development
Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum. Sed sit amet consequat ex. Pellentesque id odio felis. Etiam tellus dui, maximus non vehicula ut, suscipit vel ex. Maecenas quis lorem ut sem aliquet fermentum. Duis vel nulla hendrerit augue aliquet posuere. In erat arcu, accumsan ut rhoncus in, fringilla at elit. Morbi non gravida ligula. Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum. Sed sit amet consequat ex. Pellentesque id odio felis. Etiam tellus dui, maximus non vehicula ut, suscipit vel ex. Maecenas quis lorem ut sem aliquet fermentum. Duis vel nulla hendrerit augue aliquet posuere. In erat arcu, accumsan ut rhoncus in, fringilla at elit. Morbi non gravida ligula.
Feedback
Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum. Sed sit amet consequat ex. Pellentesque id odio felis. Etiam tellus dui, maximus non vehicula ut, suscipit vel ex. Maecenas quis lorem ut sem aliquet fermentum.
Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum. Sed sit amet consequat ex. Pellentesque id odio felis. Etiam tellus dui, maximus non vehicula ut, suscipit vel ex. Maecenas quis lorem ut sem aliquet fermentum.
Mauris et nulla tellus. Nulla consectetur placerat nunc placerat bibendum. Sed sit amet consequat ex. Pellentesque id odio felis. Etiam tellus dui, maximus non vehicula ut, suscipit vel ex. Maecenas quis lorem ut sem aliquet fermentum.