Classification and detection of road users using neural networks and Active Shape models

Autonomous vehicles interpret their environment based on their sensor data. 360° laser scanners provide comprehensive and highly accurate information about the distance of objects. Predicting the behavior of road users differs between cars, trucks/buses, cyclists and pedestrians. The exact position of the different road users depends on their orientation and geometric dimensions. Active Shape models offer the possibility to estimate the center of objects by estimating deformable models, based on CAD plans and taking into account their orientation. This work is part of the automation process to create lane-specific maps from data of daily traffic. For this purpose, road users are classified and their exact position in complex intersection scenarios is determined. The road users are detected and classified using a neural network and then precisely positioned using Active Shape models. The aim of the work is to implement a robust detection of road users on 360° laser scans.