Investigations on pedestrian long-term trajectory prediction based on AI and environmental maps

In recent years, a great research interest came up on the automatic protection of road users like pedestrians, bicyclists, or cars. One main problem of it is the position estimation of all road users in order to avoid upcoming collisions. Additionally to that, accurate positioning systems that are able to predict the intention or the future trajectory of the road user from previous paths received increasing attention. This allows for predicting collisions and for being able to send an alert or even braking the car, which is necessary for collision avoidance systems. Besides the prediction of a car’s path, the intention prediction of Vulnerable Road Users (VRU) is increasingly investigated in the literature, which is more difficult due to the fact that it can even be more random. The term VRU often refers to pedestrians, cyclists and motorcyclists. However, according to [1], it is important to differentiate between the multiple road users. Thus, in this work, only pedestrians will be considered but the models and scenarios could be adapted to other types of VRU. The objective of this paper is to predict pedestrians’ long-term trajectories using Artificial Intelligence (AI) and environmental map, which aims to provide timely alerts for VRU in dangerous situations