STMicroelectronics and Schneider Electric Partner on IoT People-Counting Sensor

The “People-Counting Solution” is designed to understand building-occupancy levels and usage and consequently enable new building-management services and efficiency gains. They showcased the sensor at the ST Live Days event on November 19, 2020.

 

The new sensor. Image courtesy of Schneider Electric/STMicroelectronics.

 

The “People-Counting” Sensor

The companies have developed a new solution to tackle the challenge of monitoring attendance in large spaces with multiple entrance points amidst the Covid-19 pandemic.

“This promising technology opens a new solution for attendance monitoring and people counting in numerous applications such as monitoring queues, building usage, and social distancing,” explained Maxime Loidreau, IoT Sensors Program Manager at Schneider Electric.

The IoT sensor combines the expertise of STMicroelectronics’ AI group with the sensor-application knowledge of Schneider Electric.

“Our innovative demonstration, created with STMicroelectronics, finds applications in various segments, from hotels to offices and retail, and more generally any building where knowing attendance and space occupation has a value,” Loidreau added.

 

A demonstration of how the new sensor works. Image courtesy of Schneider Electric/STMicroelectronics.

 

From a technical standpoint, the system works by identifying and embedding a high-performing object-detection neural network named STM32 AI in the STM32H723 microcontroller unit (MCU).

 

The STM32 AI Ecosystem

The STM32 AI ecosystem supports several deep-learning frameworks and was created to provide foundational blocks for neural networks to run on STM32 MCUs, including Keras, TensorFlow Lite, and ONNX.

The X-CUBE-AI software expansion package is also in the ecosystem, automatically converting pre-trained neural networks, generating optimized libraries for the target MCU, and integrating them into the user's project.

According to STMicroelectronics, the deconvolutional neural network (DNN) approach mapped onto the varied STM32 portfolio allows users to replicate development efforts to create diverse products for multiple markets.

This could translate to hosting AI applications, such as critical performance up to 1Mbyte Flash, and high-speed off-chip memory interfaces. It also comes with integrated features for connecting multiple sensor types.

 

Built on Deep Sensor-Application Expertise

Beyond its AI capabilities, the new prototype sensor packs in some powerful hardware specifications.

A LYNRED ThermEye family thermal imager is integrated into a low-power design by Schneider Electric, while the Yolo-based Neural Network model runs on the STM32H723 MCU from STMicroelectronics.

“This project demonstrates the power of deep learning to enhance embedded data-processing performance, showing how high-value applications can be hosted on a cost-effective microcontroller-based platform,” said Miguel Castro, AI Solutions Business Line Manager at STMicroelectronics. 

 

Miguel Castro, AI Solutions Business Line Manager at STMicroelectronics, and Maxime Loidreau, IoT Sensors Program Manager at Schneider Electric discussing the sensors. Image courtesy of Schneider Electric. 

 

Moving forward, the companies will continue their collaboration to bring the “people-counting” sensor to market and allow customers to design innovative engineering products based on its infrastructure.

“Our STM32Cube.AI ecosystem empowers users to create flexible solutions within a short time-to-market window. Customers can enjoy even greater productivity leveraging the support of our technical team to overcome engineering challenges.”

---

 

For more information about the new sensor, you can watch a video on the STMicroelectronics YouTube channel, featuring Castro and Loidreau explaining the technology in action.