STMicroelectronics, Schneider Electric and Lynred presented the first integration of Edge AI in a high-performance people-counting sensor developed in the frame of Nanoelec in November 2020.
While monitoring occupancy in large spaces with multiple entrances and exits is a significant challenge in any closed environment, it potentially provides significant value to hotels, offices, retail businesses and building managers. Three core partners of Nanoelec – STMicroelectronics, Schneider Electric, and Lynred – recently demonstrated a prototype of a high-performance people-counting sensor that overcomes the challenges. This sensor provides the opportunity to benefit from the data to optimize room occupancy, anticipate energy consumption, reduce waiting and queuing time, more efficiently manage social distancing, and more.
Since 2015, the Nanoelec/Pulse program partners have been working on attendance monitoring in public areas using digital devices. The 2020 demonstrator is the first device developed with Artificial Intelligence (AI) at the Edge. In the Nanoelec collaborative environment, STMicroelectronics, Schneider Electric, and Lynred, with scientific and technological inputs of CEA, Inria and UGA, found a strong mix of hardware and software options with the appropriate cybersecurity, reliability, and energy specifications. At the same time, Pulse/Nanoelec provided the technological platform to setup and test the different configurations of devices and algorithms to explore a range of suitable innovative solutions.
The people-monitoring device combines the AI expertise of STMicroelectronics with the deep sensor-application knowledge of Schneider Electric to identify and embed a high-performing object-detection neural network in a small microcontroller (MCU). Schneider Electric realized an increase in design productivity from its use of the STM32Cube.AI toolchain, which has mature capabilities for developing AI applications targeting the broad portfolio of STM32 MCUs. The STM32Cube.AI allowed Schneider Electric to gain valuable flexibility and efficiency in hardware design from the sophistication and ease of use of the entire STM32Cube software-development ecosystem.
The final prototype of the people-counting sensor includes a Lynred ThermEye family thermal imager into a unique ultra-low-power design created by Schneider Electric, with a Yolo-based Neural Network model running on the recently introduced high-performance STM32H723 MCU.