Modeling the behaviour of light electric vehicles in urban environments

In recent years there has been a significant increase in the number of personal mobility vehicles (PMVs) of all kinds, used in urban scenarios. In this context, autonomous cars will encounter numerous problems due to their perception limitations and sometimes, unpredictable behaviours of PMV users. Therefore, it is particularly important to study the interaction between autonomous cars and those vehicles. Among PMVs, e-scooters have become increasingly popular. Despite this, very little attention has been given to studying their behaviour in shared spaces. This work aims to develop a behavioural model for e-scooters that can be integrated into the existing autonomous vehicles simulator SPACiSS. The behavioural model was created using the Social Force Model (SFM), adapting it to handle the different characteristics an e-scooter has with respect to classic SFM models. The updated model will be able to handle physical constraints such as the dynamic and kinematic due to the non-holonomicity of the vehicle as well as implementing a new defined force that models road elements such as lanes and sidewalks. The model was integrated into an ROS-based environment that handles the communication between the simulation and the visualisation process, as well as rendering the different scenarios and their agents. Due to the lack of real-world datasets, which are crucial to obtaining an accurate model, the calibration and the tests will be discussed only theoretically. Despite this, a crash analysis will be conducted. It has been possible because it was all on simulation, but it has proved fundamental to highlight the scenarios where the model has more difficulties and where it performs nicely. Despite the challenges encountered due to the lack of datasets, the model still proved to be effective and was able to simulate a large range of real-world situations. This work contributes to developing a behavioural model for an e-scooter that enriches the field of autonomous vehicle research. Future work should be focused on validating the model presented here with real-world data and expanding it to other personal mobility vehicles.