State estimation and autonomous control of heavy-duty vehicles: a Markovian approach

• Autor(a) principal: Lucas Barbosa Marcos
• Data de Publicação: 2021
• Tipo de documento: Tese
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://doi.org/10.11606/T.18.2021.tde-27052022-100628

The past few years have seen a massive improvement in self-driving vehicle technology. However, many challenges remain ahead. For example, the autonomous control of heavyduty vehicles is still an issue because it demands robustness enough to endure huge payload variations. Also, there are still challenges concerning state estimation. For instance, take driveshaft torsion: even though it is a fundamental variable in vehicle dynamics, it is difficult to be measured or estimated due to the need for high precision encoders or because of integration estimation errors. Furthermore, gear shifting in the driveline affects state estimation and autonomous control, as it abruptly changes powertrain dynamics. Another issue is the influence of the road slope, which disturbs the system, and may or may not be measured. This thesis proposes robust discrete-time Markov jump linear system techniques for estimating driveshaft torsion and achieving autonomous driveline control. The filtering techniques are applied in two situations: with available road slope information and with unknown road slope. The algorithms are tested for a truck bodywork. Experiments show that the estimation delivers online results as accurate as offline estimation methods, especially when the road slope is known. The proposed filter is capable of estimating the torsion even in scenarios of high plant uncertainty, where an LMI-based filter only finds a highly oscillatory solution. Also, the proposed recursive controller outperforms its LMI-based counterpart in terms of tracking error and can complete the test track in scenarios where the nominal LMI-based version cannot.