Question about handling different missing beam scenarios with separate models #2
Description
This is a great piece of work! The approach to regressing missing DVL beams using deep learning is impressive. However, while running the code (LSTM model), I encountered a question. Based on the current implementation, it seems like the model is trained to handle specific missing beam scenarios, but I’m wondering if it’s necessary to train separate models for each missing scenario. For example:
A model for the scenario where one beam is missing. A model for the scenario where two beams are missing. A model for the scenario where three beams are missing. Thanks again for the great work!