Sensing the Future: Integrated MEMS and Edge Intelligence for Autonomous Robotics

The rapid evolution of intelligent robotics is driving stringent requirements for compact, energy efficient, and highly integrated sensing systems. This presentation presents recent advancements from CEA-Leti and CEA-List in the development of multifunctional MEMS-based sensing platforms combined with embedded data processing, targeting next-generation autonomous robotic applications. 

Our work focuses on three key sensing domains: (1) high-performance acoustic MEMS sensors with associated signal processing, enabling robust environmental perception; (2) integrated inertial measurement units combining gyroscopes, accelerometers, and air pressure sensors for precise motion tracking and localization; and (3) advanced image sensors embedding on-chip data processing to reduce latency and bandwidth constraints. These sensing modalities are complemented by edge AI capabilities developed within the CEA-List, including ultra-efficient embedded intelligence frameworks such as the NeuroCorgi architecture, enabling real-time interpretation and decision-making directly at the sensor level. 

The objective of this contribution is to demonstrate how tight co-integration of heterogeneous MEMS sensors with localized processing can significantly enhance robotic autonomy while minimizing power consumption and system complexity. The significance lies in bridging the gap between raw data acquisition and actionable insights, a critical requirement for scalable robotic systems operating in dynamic and constrained environments. 

Although current implementations have not been exclusively validated in robotic platforms, we extrapolate their potential through representative use cases such as navigation, obstacle detection, and human-robot interaction. These results highlight the relevance of advanced semiconductor technologies in enabling intelligent, perception-driven robotics, aligning with the future vision of highly autonomous systems.

Key Technologies Covered

• Advanced Microelectromechanical Systems (MEMS) acoustic sensing and signal processing  
• Multi-sensor fusion combining inertial (gyroscope, accelerometer) and pressure sensors  
• High-performance inertial measurement units (IMU) for motion tracking and localization  
• Edge AI for embedded, real-time data processing at the sensor level  
• Ultra-low-power sensor design for autonomous and mobile robotic platforms  
• Smart image sensors with on-chip preprocessing and feature extraction  
• Heterogeneous integration of sensors and processing units in semiconductor platforms  - - - 
• Embedded intelligence frameworks (e.g., NeuroCorgi) for adaptive decision-making  
• Sensor data optimization: noise reduction, calibration, and contextual interpretation  
• Applications in intelligent robotics: navigation, perception, and human–machine interaction