Main feature of the research is the proposal of a novel Intrusion Detection System (IDS) for autonomous cars based on the newest machine learning techniques. The main task for the IDS will be the detection of cyberattacks and their classification which could include different types like DDoS, Fuzzy, Impersonation, and ordinary "Free" traffic. The whole process of building and testing models is based on CAN intrusion-dataset which describes the communication among vehicles in terms of Messaged, Byte-level signals, and Target labels. Besides, some of the machine learning methods such as Random Forest, Gradient Boosting, Adaboost, LSTM, and CatBoost will be combined to perform the task of detecting and preventing threats. Thus, the identification of unauthorized persons trying to access vehicle networks will not only be very secure but also very adaptive because of the complete usage of these algorithms. Therefore, the security and reliability of the smart vehicle systems will be enhanced. In a nutshell, the development of a detection system has the intention that it would not just have a capability of scaling but at the same time also be able to guard against the ever-increasing cyber threats targeting the smart vehicles.

Keywords: Random Forest, Gradient Boosting, Adaboost, LSTM, CatBoost classifiers.