AI augmented Asset-tracking

In global manufacturing and commerce, raw materials, components, and final goods can travel long distances through warehouses and logistic hubs before arriving at their final destination. More often than not, the direct or indirect value of any particular item warrants certain costs associated with monitoring its condition and position. In the past years Artificial Intelligence has become a key factor to empower the transformation of asset-management. In fact, augmented assettracking represents a good solution to keep trace of physical assets, providing the location, the status and other important information. The main goal of this thesis is to implement an asset-tracking system augmented by Artificial Intelligence, based on STMicroelectronics hardware platforms and software already available. The project consists in the development of a monitoring system that, with the help of Machine Learning, can classify complex movements while carrying a package/good throughout a logistic environment. The workflow is centered in the development of two ML models: one running on the Machine Learning Core present on one of the on-board sensors and the other one running on the MCU. Then a big part of the project is the customization of the firmware, which must be done in order to be able to integrate the developed models on the hardware and process the results. In more details, thanks to a particular STM electronic board, designed for asset-tracking purposes, it is possible to acquire the samples describing the intended movements to be classified. Then, by following the usual steps adopted for a machine learning project, the network can be trained, tested and integrated in the available firmware by means of a set of STM tools. The results can be made visible in real time, while performing different movements as a simple simulation to solve and detect a specific use-case. The last part described in this thesis reports the final outcomes viewable through a simple application and web interfaces, over the Bluetooth and LoRa wireless standards.