Edge computing and machine learning on embedded systems
Running Machine Learning (ML) in embedded systems has fueled the rush for edge computing, where machine learning runs in edge devices. This approach to ML yields many results, such as lower latencies and reduction of network traffic and bandwidth. This project set out to explore machine learning in embedded systems. The Edge Impulse Platform was used to collect data and to a neural network. The neural network created was able to distinguish between five classes of motion. The Neural Network created was tested on two microcontrollers and a desktop. Inferencing on the Arduino Nano BLE took 24ms, and on the desktop, it took 271.8 μs.
Capstone Project submitted to the Department of Engineering, Ashesi University in partial fulfillment of the requirements for the award of Bachelor of Science degree in Computer Engineering, May 2022