Daniel Wiekieta
supervisor: Piotr Samczyński
Presented, the off-road environment perception system is implemented in the autonomous vehicle to examine the load capacity of runway safety areas (RSA). It is an innovational use of cutting-edge camera, lidar and radar (CLR) sensor fusion technology to detect ICAO standard obstacles at airports. This type of multi-modal sensor fusion is caused by the risk of the system running in adverse weather conditions and the high cost of eventual collision with the airport infrastructure. The high-reliability requirements are affected also using a two-stage detection system currently considered the best. This system uses a Bayesian Neural Network (BNN) in the first stage and Support Vector Machines (SVMs) in the second one. The BNN does a sensor data fusion and extracts regions of interest (RoI) for the classification’s second stage.