NeurOSmart

The high practical relevance of energy-efficient and intelligent sensors for autonomous systems is of far-reaching importance. It continuously increases as intelligent machines become an increasingly important part of our everyday lives. Some examples of this are autonomous driving or modern robot vacuum cleaners in the home and autonomously operating robots in logistics centers or manufacturing environments.

Against this increasing relevance, the institutes Fraunhofer ISIT, IPMS, IMS, IWU, and IAIS are jointly researching energy-efficient and intelligent sensors for the next generation of autonomous systems in the NeurOSmart leading project . The aim is to increase the energy efficiency of sensor-related data processing for mobile autonomous systems. For the first time, the project will combine state-of-the-art digital data processing with an analog neuromorphic accelerators specially adapted to the sensor system and evaluate it in an application environment. The novel hybrid computing architecture will be tailored to the sensor system and co-integrated. As a result, a significant proportion of the data processing and interpretation up to object classification is already realized directly in the sensor.

The human brain serves as a model for the so-called neuromorphic electronics to be developed, as it is very energy-efficient when making decisions despite its enormous computing power. Although the computing power of neuromorphic electronics cannot yet surpass that of the brain, the application specifically to a sensor enables the development of small and efficient models for object recognition and classification.

Advantages compared to alternative methods

• Easier sensor integration in complex systems
• Increased data protection
• Significant energy savings

Participating Fraunhofer institutes:

ENAS, IAIS, IMS, IPMS, ISITand IWU