Proposal of Hybrid Deep Learning Algorithm for In Cabin Monitoring System

An autonomous vehicle is defined as a vehicle that is capable of navigating and operating independently, without the necessity for input from a driver or passengers. The Society of Automotive Engineers (SAE) has established internationally recognised standards for the classification of autonomous driving technology, delineating the various levels of autonomy. The development of Level 4 autonomous vehicles is currently being led by leading technology companies, including Google, Nvidia, and Tesla. For vehicles classified as Level 3 or above, the autonomous system is required to fulfil the role of the driver. This necessitates the development of advanced decision-making algorithms to support high-level autonomy. To create an effective deep learning-based autonomous driving system, it is essential to have a diverse array of scenarios and a large dataset. This study introduces a novel image captioning algorithm that utilises a hybrid CNN-LSTM model to assess passenger conditions within the vehicle and generate corresponding scenarios. Furthermore, the study evaluates the suitability of the produced data for training a monitoring system through simulated environments