The goal of the project is to realize an adversarial driving emulator in a driving simulator context. That is to build a tool that based on a Nonlinear Model Predictive Controller (NMPC) makes possible the interaction between an adversarial vehicle and a human-driven vehicle. For this purpose, initially a single virtual vehicle on the track will be considered to test the best controller configuration. Subsequently, two vehicles will be driven by the same controller in a simulated context. In this case, the internal model of the controller includes both vehicles in order to also predict the behaviour of the human vehicle in case of overtaking. In fact, the overtaking management will prove to be an important factor in describing the relationship between the two vehicles. Thereafter, one vehicle will be controlled by the virtual driver and the other will have a fixed predetermined trajectory. This simulation will aim to verify the controller's response to the unexpected behaviour of the second vehicle. To conclude, the controller that realizes this goal will have to take into account a variable configuration. In fact, the shape of the track and the initial conditions for the two vehicles will turn out to be determining factors for their interaction and the overtaking management on the track.
L'obiettivo del progetto è realizzare un emulatore di guida avversario da inserire in un contesto di simulatori di guida, ovvero costruire uno strumento che basato su un Controllore Predittivo per Modelli Non lineari (NMPC) renda possibile l'interazione tra un veicolo avversario e un veicolo pilotato da un umano. A tal proposito inizialmente sarà considerato un singolo veicolo virtuale sul tracciato per verificare la miglior configurazione del controllore. Successivamente due veicoli saranno pilotati dallo stesso controllore in un contesto simulato. In questo caso il modello interno del controllore comprenderà entrambi i veicoli, allo scopo di predire anche il comportamento del veicolo umano in caso di sorpasso. Infatti la gestione del sorpasso si rivelerà un fattore importante per descrivere la relazione tra i due veicoli. In seguito un veicolo sarà controllato dal driver virtuale e l'altro avrà una traiettoria fissa predeterminata. Questa simulazione avrà come scopo quello di verificare la risposta del controllore al comportamento inatteso del secondo veicolo. Per concludere il controllore che realizza questo obbiettivo dovrà tenere conto di una configurazione variabile. Infatti la forma del tracciato e le condizioni iniziali dei due veicoli risulteranno dei fattori determinanti per la loro interazione e la gestione del sorpasso su pista.
A NMPC based adversarial drivers for driving simulators
BRONCA, NICCOLÒ
2021/2022
Abstract
The goal of the project is to realize an adversarial driving emulator in a driving simulator context. That is to build a tool that based on a Nonlinear Model Predictive Controller (NMPC) makes possible the interaction between an adversarial vehicle and a human-driven vehicle. For this purpose, initially a single virtual vehicle on the track will be considered to test the best controller configuration. Subsequently, two vehicles will be driven by the same controller in a simulated context. In this case, the internal model of the controller includes both vehicles in order to also predict the behaviour of the human vehicle in case of overtaking. In fact, the overtaking management will prove to be an important factor in describing the relationship between the two vehicles. Thereafter, one vehicle will be controlled by the virtual driver and the other will have a fixed predetermined trajectory. This simulation will aim to verify the controller's response to the unexpected behaviour of the second vehicle. To conclude, the controller that realizes this goal will have to take into account a variable configuration. In fact, the shape of the track and the initial conditions for the two vehicles will turn out to be determining factors for their interaction and the overtaking management on the track.File | Dimensione | Formato | |
---|---|---|---|
Bronca_Niccolò.pdf
Open Access dal 19/07/2023
Dimensione
3.27 MB
Formato
Adobe PDF
|
3.27 MB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/31583