Advanced Edge-Computing Solution for Intelligent Material Identification in Construction: WebAssembly Integration and Performance Analysis

This project focuses on developing an edge-computing solution for conventional construction tools that can intelligently identify materials based on vibration analysis. The system is designed to be modular and adaptable to different use cases, with the integration of WebAssembly allowing for the same application to be executed on different platforms with different operating systems and programming languages. The system also allows for wireless firmware updates and stores data in AWS S3 for long-term archiving and analysis. Tests were conducted to verify the system’s effectiveness and reliability, including evaluating the inferencing duration for the task identification machine learning model based on different window sizes. The results show that a window size of 335 milliseconds achieves higher accuracy of 72.73% compared to a window size of 4000 milliseconds with an accuracy of 70.59%. The machine learning model is capable of achieving an accuracy of 69.09% with all features integrated into the system of a microcontroller.