TY - CHAP A1 - Cristófoli Duarte Silva, Letícia A1 - Funk Drechsler, Maikol A1 - Poledna, Yuri A1 - Huber, Werner A1 - Antonio Fiorentin, Thiago ED - Ariwa, Ezendu ED - Fong, Simon T1 - Synthetic Extreme Weather for AI Training: Concept and Validation T2 - Proceedings: 2023 Third International Conference on Digital Data Processing, DDP 2023 UR - https://doi.org/10.1109/DDP60485.2023.00044 Y1 - 2023 UR - https://doi.org/10.1109/DDP60485.2023.00044 SN - 979-8-3503-2901-8 SP - 188 EP - 194 PB - IEEE CY - Piscataway ER - TY - JOUR A1 - Amaral, Janaína R. A1 - Fiorentin, Thiago Antonio A1 - Göllinger, Harald T1 - Improvement of vehicle stability using a controller based on reinforcement learning JF - International Journal of Vehicle Systems Modelling and Testing (IJVSMT) UR - https://doi.org/10.1504/IJVSMT.2020.111677 KW - torque vectoring KW - neural fitted Q iteration KW - vehicle stability KW - vehicle handling KW - CarMaker KW - sideslip angle Y1 - 2020 UR - https://doi.org/10.1504/IJVSMT.2020.111677 SN - 1745-6436 SN - 1745-6444 VL - 14 IS - 2-3 SP - 97 EP - 112 PB - Inderscience CY - Genf ER - TY - JOUR A1 - Funk Drechsler, Maikol A1 - Fiorentin, Thiago Antonio A1 - Göllinger, Harald T1 - Actor-Critic Traction Control Based on Reinforcement Learning with Open-Loop Training JF - Modelling and Simulation in Engineering N2 - The use of actor-critic algorithms can improve the controllers currently implemented in automotive applications. This method combines reinforcement learning (RL) and neural networks to achieve the possibility of controlling nonlinear systems with real-time capabilities. Actor-critic algorithms were already applied with success in different controllers including autonomous driving, antilock braking system (ABS), and electronic stability control (ESC). However, in the current researches, virtual environments are implemented for the training process instead of using real plants to obtain the datasets. This limitation is given by trial and error methods implemented for the training process, which generates considerable risks in case the controller directly acts on the real plant. In this way, the present research proposes and evaluates an open-loop training process, which permits the data acquisition without the control interaction and an open-loop training of the neural networks. The performance of the trained controllers is evaluated by a design of experiments (DOE) to understand how it is affected by the generated dataset. The results present a successful application of open-loop training architecture. The controller can maintain the slip ratio under adequate levels during maneuvers on different floors, including grounds that are not applied during the training process. The actor neural network is also able to identify the different floors and change the acceleration profile according to the characteristics of each ground. UR - https://doi.org/10.1155/2021/4641450 Y1 - 2021 UR - https://doi.org/10.1155/2021/4641450 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-15862 SN - 1687-5605 VL - 2021 PB - Hindawi CY - New York ER -