Register now After registration you will be able to apply for this opportunity online.
Smart Sensor Fusion and TinyML for indoor People Counting using Low-Power Ultrasound Sensors
The project's goal is to create cutting-edge algorithms that use low-power ultrasonic sensors with TinyML for precise presence detection and indoor people counting. Additionally, the study intends to investigate sensor fusion with CO2, VOC, noise, or other data to improve accuracy. The study assesses the practicality of the technology, taking into account the trade-offs between precision, power consumption, and system limitations.
Keywords: Embedded Systems, Low-power Sensing, Signal Processing, Machine Learning, Energy Harvesting, Wireless Communication, Sensor Fusion
We at Anevo, a proud startup of PBL, are working on a completely self-sustaining sensor platform for occupancy intelligence that can be integrated into building automation systems to improve HVAC control and building operations. The goal is to save energy by understanding and anticipating the occupancy of rooms or open spaces inside buildings. We connect novel, ultra-low-power sensor capabilities to anonymously estimate interior occupancy.
Your task in this project will be to develop algorithms for a low-power ultrasonic sensor that can identify and count people in the surrounding area using TinyML. A prototype of the sensor devices, with different indoor sensing capabilities, has already been developed and tested. If time allows, sensor fusion can be examined by combining ultrasonic sensor data with CO2, VOC, noise, or other indoor sensor data to provide a more precise estimation of occupancy. The technology's overall feasibility, as well as the trade-offs between precision, power consumption, and system restrictions, are of interest for this project.
**Prerequisites**
- Embedded Systems - MCU programming, sensor interfaces (I2C, SPI etc.)
- MATLAB or Python for evaluation and data processing
- Basic knowledge in Machine Learning
- Wireless Communication (Bluetooth Low-Energy or LoRa) is a plus
- Circuit design (Altium, KiCad) is a plus
**Character**
- 10% Literature study
- 70% Signal processing, data acquisition, algorithm development, and potentially hardware design
- 20% Evaluation and validation
We at Anevo, a proud startup of PBL, are working on a completely self-sustaining sensor platform for occupancy intelligence that can be integrated into building automation systems to improve HVAC control and building operations. The goal is to save energy by understanding and anticipating the occupancy of rooms or open spaces inside buildings. We connect novel, ultra-low-power sensor capabilities to anonymously estimate interior occupancy. Your task in this project will be to develop algorithms for a low-power ultrasonic sensor that can identify and count people in the surrounding area using TinyML. A prototype of the sensor devices, with different indoor sensing capabilities, has already been developed and tested. If time allows, sensor fusion can be examined by combining ultrasonic sensor data with CO2, VOC, noise, or other indoor sensor data to provide a more precise estimation of occupancy. The technology's overall feasibility, as well as the trade-offs between precision, power consumption, and system restrictions, are of interest for this project.
- MATLAB or Python for evaluation and data processing
- Basic knowledge in Machine Learning
- Wireless Communication (Bluetooth Low-Energy or LoRa) is a plus
- Circuit design (Altium, KiCad) is a plus
**Character**
- 10% Literature study
- 70% Signal processing, data acquisition, algorithm development, and potentially hardware design
- 20% Evaluation and validation
**Project Tasks**
- Acquire a dataset of ultrasound sensor data in various indoor environments
- Develop and evaluate different algorithms for presence detection and people count estimation using a low-power ultrasound sensor
- Compare accuracy, power consumption, and other characteristics to existing solutions
- Apply sensor fusion with additional sensors to have a precise people count estimation
**Project Tasks**
- Acquire a dataset of ultrasound sensor data in various indoor environments
- Develop and evaluate different algorithms for presence detection and people count estimation using a low-power ultrasound sensor
- Compare accuracy, power consumption, and other characteristics to existing solutions
- Apply sensor fusion with additional sensors to have a precise people count estimation
- Tiago Salzmann (tiago.salzmann@pbl.ee.ethz.ch)
- Yvan Bosshard (y.bosshard@anevo.ch)
- Dr. Michele Magno (michele.magno@pbl.ee.ethz.ch)
- Tiago Salzmann (tiago.salzmann@pbl.ee.ethz.ch)
- Yvan Bosshard (y.bosshard@anevo.ch)
- Dr. Michele Magno (michele.magno@pbl.ee.ethz.ch)