In a joint Hungarian-Vietnamese research project launched three years ago, researchers from HUN-REN SZTAKI used a hybrid learning method to train self-driving vehicles for safe emergency manoeuvres.
The research group led by Professor Péter Gáspár also included researchers from the University of Transport and Communications (UTC) in Vietnam. The project, entitled Emergency Path Planning for Cooperative Autonomous Vehicles, supported by the National Research, Development and Innovation Office (NKFIH), developed a control system that would enable autonomous vehicles, communicating with each other and the environment, to perform safe evasive manoeuvres in an emergency situation.
Researchers performed tests on an autonomous Lexus RX 450h not only in simulation, but also in a real environment on the ZalaZONE proving ground.

Maintaining the driving stability of vehicles has long been a research topic in control theory. The systems currently in use provide support to the driver when the vehicle condition requires These systems stabilize the vehicle using classic control methods, using the wheel brake when skidding.

In the research, a control system was developed based on a combination of machine learning and traditional control technology solutions. This system is able to consider environmental information and ensure the planning and execution of the vehicle’s safe trajectory.
The project was an experimental development combining vehicle dynamics, sensor data integration and machine learning assisted vehicle control, respecting the development processes of the automotive industry.