Drowsy driving is the primary cause of motor vehicle accidents and is a risk factor that leads to the loss of human life, remaining as a challenge for the global automotive industry. Recently, drowsy monitoring system has been actively studied for prediction system based machine learning. However, the challenges of automotive real-time constraints and flexibility should be taken into consideration against a large amount of heterogeneous data from vehicle network and other device. To solve this problem, we propose drowsy monitoring system based machine learning using GS1 standard. First, vehicle motion data is defined and modeled using the GS1 standard language for drowsy predict. Second, we propose an optimal algorithm selection and detail architecture for automotive real-time environments through machine learning algorithms (KNN, Naïve Bayes, Logistic Regression) and deep learning algorithms (RNN-LSTM). Finally, we describe system-wide integration and implementation through the open source hardware Raspberry Pi and the machine learning SW framework. We provide optimal LSTM architecture and implementation that takes into account the real-time environmental conditions and how to improve the readability and usability of the vehicle motion data. We also share the rapid prototyping methodology case of connected car systems without other sensor devices.