Vehicle trajectory prediction through manual setting and machine learning

Paper in proceeding, 2022

The prediction of vehicle trajectory is an important step in safe driving. The relative positions or speeds of vehicles affect the interactions between them, which in turn produce predictions of trajectories. Through the trajectory prediction of each vehicle to form a complete safe driving environment. This paper will be divided into two parts for vehicle track, namely manual setting and machine learning. The advantage of manual setting is that it is convenient to observe the characteristics and then make other vehicles follow the corresponding behaviour. However, the disadvantage is that the characteristics that can be observed are incomplete, which makes the experimental error relatively large. The advantage of machine learning is that through previous experience to make all vehicles change in the corresponding situation. Through previous experience to make all vehicles change in the corresponding situation. An effective extraction and addition of the original data is helpful to reduce the final error. Our code base can be found in https://github.com/RR30/CIS_multi_vehicle_prediction/tree/Xinyu.