Automated Driving systems based on Model Predictive Control and Artificial Potential Fields applied to Shuttle Buses in Restricted Areas

In the last few years, shared mobility is increasing vs personal owned vehicles; one application that has become an important element in the field of Automated driving and urban mobility is the autonomous shuttle bus. The introduction into the automotive market is thought to be a gradual process that involves progressive work starting from restricted areas to reach robotaxi from 2030 by increasing the ODD. In the present work, the AD system of a shuttle bus driving in a restricted area is considered. In this functionality, the vehicle should be capable of reacting to the surrounding environment, making decisions based on sensors data, and consequently performing driving manoeuvres in autonomous way, since the considered level of automation is SAE L4. The accounted scenario in this application is a restricted area with urban roads, including stations, intersections, roundabout and U turns as well. The used approach is a combination of Artificial Potential Fields~(APF) and Non Linear Model Predictive Controller (NMPC), where the vehicle dynamics are simulated using a non linear model with discrete-time equations. In particular, the optimization problem solved by the controller includes the APF terms in the cost function, and it is subject to kinematic constraints mainly due to physical limitation and comfort performances. The potential functions are implemented in order to generate a reference path trajectory, according to the fact that the vehicle must maintain the center of the road's lane and accomplish its tasks during the journey. Since the beginning, the functional control architecture was configured within a automotive eco-system contexts, using a multimodal sensing approach (on-board sensors, environmental data, off-board information, etc.) to provide relevant content and functionality. In the present work, the controller architecture is supported by a Simplified Behavioural Logic, described as a Finite State Machine, that has the role of choosing the appropriate response based on environment evaluations. These analysis are made according to the data provided by vehicle's on-board sensors and by the infrastructure, in which the only type of exchanged information regard other possible surrounding vehicles. The shuttle is equipped with all the needed intelligence and capability of handling V2X communication, facilitating perfect integration into the traffic ecosystem. Moreover, in this thesis work, the bus can perform some basic manoeuvres such as: maintain the reference trajectory path with related speed, stop at designated station points, overtake fixed obstacles on the shuttle's lane, give precedence in case of presence of other vehicles and follow heading car using Adaptive Cruise Control (ACC), while maintaining a certain security distance and velocity. To verify the efficacy of the implemented controller and architecture, and to show all the capability of the shuttle, extensive numerical simulations are made. Different realistic scenarios are considered, also in complex multi-vehicles conditions, affirming the shuttle's robustness and adaptability in real-world situations.