PRESENTATION
Distributed Edge AI for Sensors – towards safe & secure embedded AI sensors
Why not have IIoT sensors transmit high-level, context-aware information instead of raw measurement values? AI based processing of sensor data within sensor modules allows to mitigate cloud connectivity dependencies. Systems are enabled to respond in real-time to events more complex than a simple threshold. Handling of cross-sensitivities and the calibration process can be handled more cost-efficient and also happen adaptively in the field, providing a good reason to investigate AI even for legacy applications. Unfortunately, information security and fault-tolerance become an increasingly important topic for such self-contained Edge AI sensors, with cost requirements demand for resource constraint hardware. The interaction of the distributed AI framework for embedded Systems „AIfES“ with custom computing hardware based the „AIRISC“ family of RISC-V softcores, supported by our rigourous formal safety- and security-by-design approaches can serve as an enabling technology, as will be shown in the process of the talk. Examples will include medical wearables as well as power electronic devices and LIDAR sensors. References: www.airisc.de, www.aifes.de, www.ims.fraunhofer.de
Alexander Stanitzki
Fraunhofer IMS
Mr. Stanitzki is head of the industrial sensors business unit within Fraunhofer IMS. After graduating from the Ruhr-University Bochum in 2008 he worked as an IC design engineer for radiation hard communication ICs. In 2011 he joined Fraunhofer IMS in Duisburg and was head of the mixed-signal ASIC design group until he became responsible for the industrial business unit and strategic planning in 2021. He is actively advocating the use of Open Hardware and Open Source in sensors development and became a RISC-V ambassador in 2019.