This post is also available in:
עברית (Hebrew)
As robots take on more complex roles in industrial environments, one challenge remains constant: ensuring safe, seamless interaction with human workers. Heavy-duty robotic arms must execute precise movements while maintaining a safe distance, reacting instantly if a person steps into their workspace. Traditional safety systems often rely on external processing and predefined safety zones, which can limit flexibility and slow operations.
A new integrated platform known as NeurOSmart approaches the problem by combining advanced sensing and brain-inspired computing directly within the robot’s perception system.
At its core is a LIDAR-based laser sensor that continuously scans the shared workspace from above. The system emits short pulses of near-infrared light and measures their reflections to build a high-resolution 3D map. Movable MEMS mirrors distribute the laser across the area, using ultra-thin piezoelectric aluminum scandium nitride (AlScN) layers just one micrometer thick. According to TechXplore, the improved mirror design enhances both performance and energy efficiency.
Unlike conventional setups that transmit raw sensor data to external processors, the system performs data processing directly within the sensor module. AI algorithms first preprocess and filter the incoming signals, identifying areas of interest and reducing unnecessary data flow. This lowers power consumption while maintaining real-time responsiveness.
The final evaluation step is handled by a neuromorphic accelerator chip integrated into the system. Designed to function similarly to the human brain, the processor consists of many small interconnected computing units arranged in a matrix. Each unit acts as a decision-making cell, enabling rapid, parallel processing. The result is a reaction time measured in milliseconds between detecting a person and adjusting the robot’s motion. In practice, the system can slow or stop a heavy robot arm if someone enters its proximity.
For defense and homeland security applications, similar architectures could support safer human-machine collaboration in logistics hubs, maintenance facilities or autonomous vehicle operations. Energy-efficient neuromorphic processing may also benefit mobile platforms such as drones or field-deployed sensor systems.
By merging 3D sensing, embedded AI and neuromorphic hardware into a unified platform, the project outlines a pathway toward more adaptive and responsive robotic systems in environments where humans and machines work side by side.
