MEMS BASED SMART SENSOR SYSTEM AND ITS FUTURE
Shanshan Gu-Stoppel
Fraunhofer Institute for Silicon Technology ISIT

Abstract
As manufacturing technologies for MEMS devices become more and more established, the challenge for the MEMS industry is to stay innovative and maintain access to new technologies. Today we will present an example of a smart sensor system that combines MEMS devices with new materials and machine learning for data processing. With the increasing autonomy of mobile robotic systems, the number of sensors, the effort to link their data and thus the need for computing power to ensure reliable and safe real-time operation is increasing. The scalability of the architecture, sufficient transmission bandwidth between sensor and data processing and the minimization of energy requirements are the greatest challenges for their uses in mobile systems, which is why the development reported here focuses on the direct integration of data processing intelligence into the sensor system: A MEMS-scanner-based LiDAR system enables direct access to the incoming data stream, whereupon an analog neuromorphic HPC chip is coupled with an AI-assisted pre-processing pipeline to interpret the data directly at the sensor. In the end, the decisions based on the data analysis are fed back to the sensor systems to support the subsequent operations. The talk will also present applications scenarios for such a smart sensor system and discussion about the development roadmap for smart MEMS sensor systems.

Biography
Dr. Shanshan Gu-Stoppel is a senior scientist at Fraunhofer Institute for Silicon Technology (ISIT) and heads the research group Optical Systems. She is an expert in development of MEMS devices, systems and manufacturing process. Her expertise covers products such as MEMS mirrors, IMU devices, MEMS speakers and microfluidics.