FAQ

The primary goal of the LEARN project is to develop innovative wearable exoskeletons for motor assistance, leveraging machine learning (ML) as a key technology to significantly improve their controllability and usability for end users. The project aims to integrate ML and edge computing to create semi-autonomous devices capable of understanding and responding to user needs more intuitively and effectively.

The LEARN project relies on a combination of advanced technologies, including:

• Machine Learning (ML): For processing biometric signals, detecting movement intention, and semi-autonomous device control.
• Edge Computing: For real-time data processing directly on the device, reducing reliance on cloud computing.
• Artificial Vision: For affordance segmentation, enabling the device to recognize objects and their possible interactions.
• Biometric Sensors: For detecting muscle signals (surface electromyography, sEMG) and other physiological signals for more intuitive control.
• Robotic Exoskeletons: For providing motor support and assisting upper limb movements.

ML plays a crucial role in the LEARN project by enabling several advanced functionalities:

• Improved Control: ML interprets user biometric signals, such as muscle signals (sEMG), for more precise and intuitive device control.
• Operational Awareness: ML allows the device to “understand” the operational context, such as the type of object the user is trying to grasp, and adapt its behavior accordingly.
• Shared Autonomy: ML enables semi-autonomous systems capable of collaborating with the user, providing assistance only when necessary and allowing full user control when possible.

Using edge computing in the LEARN project offers several significant advantages:

• Real-time Processing: Processing data directly on the device allows for faster response times and smoother exoskeleton control.
• Privacy and Security: Local processing of sensitive data, such as biometric signals, reduces privacy and security risks.
• Reduced Latency: Processing data on the device eliminates the need to send data to the cloud, reducing latency and improving responsiveness.
• Offline Operation: Edge computing enables the device to function even without internet connectivity.

Affordance segmentation is the ability of an artificial vision system to identify different parts of an object and understand their possible interactions. In the LEARN project, affordance segmentation is used to allow the exoskeleton to:

• Recognize Objects: The device can identify objects in the surrounding environment, such as cups, pens, or tools.
• Predict Object Use: The device can “understand” how an object can be grasped or used, for example, by gripping a cup by the handle.
• Adapt Grip: The exoskeleton can adjust its grip based on the shape and size of the object, ensuring a secure and stable hold.

The LEARN project has the potential to significantly improve the quality of life for patients with motor disabilities by enabling them to:

• Regain Autonomy: AI-enhanced exoskeletons can help patients perform daily activities that would otherwise be difficult or impossible, such as eating, dressing, or writing.
• Participate More Actively in Social Life: Greater autonomy can enable patients to participate more actively in social and work life.
• Improve Quality of Life: The ability to perform daily activities more independently can have a positive impact on patients’ mental health and overall well-being.