|
Indoor tracking for factory environment |
|---|---|
| รหัสดีโอไอ | |
| Title | Indoor tracking for factory environment |
| Creator | Karishma Agrawal |
| Contributor | Supachai Vorapojpisut, Advisor |
| Publisher | Thammasat University |
| Publication Year | 2566 |
| Keyword | Manufacturing process, Bluetooth low energy, Duration and Interval Hidden Markov Model, Classification tree, Hidden Semi Markov Model |
| Abstract | This study focuses on the application of an Indoor Positioning System (IPS) using Bluetooth Low Energy (BLE) network for estimating key manufacturing metrics, specifically cycle time. The manufacturing flow is represented as a Hidden Semi Markov Model (HSMM) problem, and a model is trained using RSSI data from the BLE network. An algorithm is proposed to determine the duration probability distribution, a critical parameter of the model. The study also addresses scenarios where no RSSI data is available, such as when products are stored away from BLE scanners. This problem is formulated as a Duration and Interval Hidden Markov Model (DIHMM), and a combined architecture of a classification tree and HSMM is proposed. The algorithms are extended to handle the DI-HMM problem. RSSI observation sequences are generated using software, real-world experiments, and simulations. The estimated HSMM duration probability is evaluated using the Kullback-Leibler Divergence (KLD), resulting in a score of 0.0573. The estimated duration and interval of the DI-HMM are compared to the actual values using a vector distance score of 0.4717. |
