Behaviour modelling of vehicles at a Roundabout

This thesis concentrates on the motion prediction of the agents utilizing the behaviour model. In an autonomous environment, the agents need to beaware of other agents’ position and actions to minimize the use of emergency braking, thereby reducing collisions and damages. The data isused to forecast agents’ position in the environment and classify agents’ exits utilizing a variation of Recurrent Neural Network(RNN), namely, Long Short-Term Memory (LSTM) to determine the specific behaviour model. Additionally, the network performance is compared with other RNN architecture such as the Bi-LSTM and Bi-LSTM + LSTM stacked architecture to evaluate which model has the best performance. The results achieved in this thesis are comparable to prior literature. They have shown improvement in some aspects of modelling behaviour at a roundabout using data acquired from an infrastructure-based LiDAR sensor