Beyinden Esinlenen NeurOSmart Sistemi Robot Güvenliğini Artırıyor

[Ahmet Ö.]

Fraunhofer's NeurOSmart system enhances safety and efficiency in human-robot collaboration through neuromorphic computing and integrated LiDAR sensors.

Researchers from the Fraunhofer Society have introduced the NeurOSmart system, a neuromorphic computing platform developed to improve safety and efficiency in human-robot collaboration. Thanks to LiDAR sensors integrated with an analog processor inspired by the brain, environmental data is processed directly at the sensor level. This edge-based approach reduces latency and energy consumption, allowing robots to perceive and react to human movements in real-time. This enables safer and more responsive industrial automation systems.

### Technology Inspired by the Human Brain

Fraunhofer draws inspiration from the functioning of the human brain to enhance the efficiency and responsiveness of robotic systems. The human body receives, processes, stores, and predicts future movements from a multitude of sensory data every second; this biomechanical excellence serves as a significant reference for robotic systems.

### New Horizons in Robotic Systems

Significant advancements exist in robots equipped with sensors, visual systems, and AI-powered processors. However, achieving the processing power and efficiency of the human brain remains a challenge. Large language models and advanced deep learning algorithms can, in some areas, surpass the human brain in analyzing complex data.

### NeurOSmart Project

The NeurOSmart project, a collaboration of five Fraunhofer institutes (ISIT, IPMS, IMS, IWU, and IAIS), focuses on developing hybrid intelligent architectures with sensor-based data processing. The goal is to increase the dynamic response of robotic systems while improving energy efficiency.

### NeurOSmart Features

• Use of analog neuromorphic HPC (High-Performance Computing) chip
• Modeling the biomechanics of the human brain with analog circuits
• Eliminating the RAM-CPU data transfer delay known as the "Von Neumann bottleneck"
• Use of analog circuit elements like memristors for data storage
• High-resolution 3D imaging with LiDAR systems integrated into edge processors
• Laser scanning with MEMS (Micro-Electro-Mechanical Systems) mirrors
• More efficient movement and lower energy consumption using piezoelectric aluminum scandium nitride (AlScN) material
• AI-powered pre-processing to distinguish between a robot arm and a human hand or arm at the sensor level
• Reduced processing load on the factory network due to sensor and processor integration
• Energy consumption reduced to almost the level of a light bulb

### New Safety Standards in Human-Robot Collaboration

NeurOSmart's core achievement is enabling robotic systems to "think" and react like humans. The HPC chip structure processes data within milliseconds by mimicking the brain's neural connection networks. Fraunhofer engineers have established high standards for worker safety by training AI in real working environments.

This innovative system significantly contributes to the automation and robotics sector by enhancing safety in human-robot collaboration and improving energy efficiency.