Embedded Reinforcement Learning: Paving the Roads Towards Fuel Efficiency
Resumo
This paper presents a proof of concept for an Advanced Driver Assistance System (ADAS) based on embedded Reinforcement Learning (RL) aimed at fuel consumption optimization. Using the Asynchronous Advantage ActorCritic (A3C) algorithm, the model was integrated into a mobile application that processes vehicle telemetry data (OBD-II), geolocation, and known road topography. Operating as a passive level 0 ADAS, the system provides real-time acceleration suggestions to the driver. To validate this pilot study, evaluations were conducted through simulations and preliminary real-world road tests on a fixed route. The results indicated a 9% reduction in consumption in a simulated environment and efficiency gains between 7.64% and 14.01% during the on-road tests. We conclude that the proposed architecture demonstrates technical feasibility for promoting fuel savings, establishing a baseline for future validations with larger driver samples and dynamic routing.Referências
(2023). Transport and environment statistics: 2023. United Kingdom. Departament of Transports, Londres.
(2024). CO2 emission performance standards for cars and vans. European Comission, Bélgica.
Abediasl, H., Ansari, A., Hosseini, V., Koch, C. R., and Shahbakhti, M. (2023). Real-time vehicular fuel consumption estimation using machine learning and on-board diagnostics data. Journal of Automobile Engineering, 2020:14 p.
Abukhalil, T., AlMahafzah, H., Alksasbeh, M., and Alqaralleh, B. (2020). Fuel consumption using obd-ii and support vector machine model. Journal of Robotics, 2020:9 p.
Adheesh, S. R., Vasisht, M. S., and Ramasesha, S. K. (2016). Air-pollution and economics: diesel bus versus electric bus. Current Science, 110(5):p. 858–862.
de Goede, M., Hogema, J., Vermeulen, R., and Bleuanus, S. (2010). Effects of a gear-shift indicator and a fuel economy meter on fuel consumption. Países baixos.
Evans, L. (1979). Driver behaviour effects on fuel consumption in urban driving. Human Factors, 21(4):p. 389–398.
Juwono, J. N., Julienne, N. D. B., Yogatama, A. S., and Widianto, M. H. (2023). Motorized vehicle diagnosis design using the internet of things concept with the help of tsukamoto’s fuzzy logic algorithm. Journal of Robotics and Control, 4:p. 202–216.
Kamal, M., Mukai, M., Murata, J., and Kawabe, T. (2011). Ecological vehicle control on roads with up-down slopes. Intelligent Transportation Systems, 12:p. 783–794.
Massoud, R., 1, R. B., Poslad, S., Gloria, A. D., and Bellotti, F. (2021). Iot sensing for reality-enhanced serious games, a fuel-efficient drive use case. Sensors, 21(10):17 p.
Mnih, V., Badia, A. P., Mirza, M., Graves, A., Lillicrap, T., Harley, T., Silver, D., and Kavukcuoglu, K. (2016). Asynchronous methods for deep reinforcement learning. In International conference on machine learning, pages 1928–1937, Nova Iorque. JMLR.
Mohammadnazar, A., Khattak, Z. H., and Khattak, A. J. (2023). Assessing driving behavior influence on fuel efficiency using machine-learning and drive-cycle simulations. Transportation Research Part D: Transport and Environment, 126:15 p.
Pagliarini, C. (2024). 1,4 milhão de veículos totalmente elétricos vendidos no brasil. Bright Consulting.
Rykała, M., Grzelak, M., Łukasz Rykała, Voicu, D., and Stoica, R.-M. (2023). Modeling vehicle fuel consumption using a low-cost obd-ii interface. Energies, 16:23 p.
Statista (2024). Average annual brent crude oil price from 1976 to 2024.
Sussman, R. (2018). Car buyers rank fuel economy as a top priority. ACEEE.
Yao, Y., Zhao, X., Liu, C., Rong, J., Zhang, Y., Dong, Z., and Su, Y. (2020). Vehicle fuel consumption prediction method based on driving behavior data collected from smartphones. Journal of Advanced Transportation, 2020:11 p.
Yen, M., Tian, S., Lin, Y., Yang, C.-W., and Chen, C.-C. (2021). Combining a universal obd-ii module with deep learning to develop an eco-driving analysis system. Applied Sciences, 11:17 p.
(2024). CO2 emission performance standards for cars and vans. European Comission, Bélgica.
Abediasl, H., Ansari, A., Hosseini, V., Koch, C. R., and Shahbakhti, M. (2023). Real-time vehicular fuel consumption estimation using machine learning and on-board diagnostics data. Journal of Automobile Engineering, 2020:14 p.
Abukhalil, T., AlMahafzah, H., Alksasbeh, M., and Alqaralleh, B. (2020). Fuel consumption using obd-ii and support vector machine model. Journal of Robotics, 2020:9 p.
Adheesh, S. R., Vasisht, M. S., and Ramasesha, S. K. (2016). Air-pollution and economics: diesel bus versus electric bus. Current Science, 110(5):p. 858–862.
de Goede, M., Hogema, J., Vermeulen, R., and Bleuanus, S. (2010). Effects of a gear-shift indicator and a fuel economy meter on fuel consumption. Países baixos.
Evans, L. (1979). Driver behaviour effects on fuel consumption in urban driving. Human Factors, 21(4):p. 389–398.
Juwono, J. N., Julienne, N. D. B., Yogatama, A. S., and Widianto, M. H. (2023). Motorized vehicle diagnosis design using the internet of things concept with the help of tsukamoto’s fuzzy logic algorithm. Journal of Robotics and Control, 4:p. 202–216.
Kamal, M., Mukai, M., Murata, J., and Kawabe, T. (2011). Ecological vehicle control on roads with up-down slopes. Intelligent Transportation Systems, 12:p. 783–794.
Massoud, R., 1, R. B., Poslad, S., Gloria, A. D., and Bellotti, F. (2021). Iot sensing for reality-enhanced serious games, a fuel-efficient drive use case. Sensors, 21(10):17 p.
Mnih, V., Badia, A. P., Mirza, M., Graves, A., Lillicrap, T., Harley, T., Silver, D., and Kavukcuoglu, K. (2016). Asynchronous methods for deep reinforcement learning. In International conference on machine learning, pages 1928–1937, Nova Iorque. JMLR.
Mohammadnazar, A., Khattak, Z. H., and Khattak, A. J. (2023). Assessing driving behavior influence on fuel efficiency using machine-learning and drive-cycle simulations. Transportation Research Part D: Transport and Environment, 126:15 p.
Pagliarini, C. (2024). 1,4 milhão de veículos totalmente elétricos vendidos no brasil. Bright Consulting.
Rykała, M., Grzelak, M., Łukasz Rykała, Voicu, D., and Stoica, R.-M. (2023). Modeling vehicle fuel consumption using a low-cost obd-ii interface. Energies, 16:23 p.
Statista (2024). Average annual brent crude oil price from 1976 to 2024.
Sussman, R. (2018). Car buyers rank fuel economy as a top priority. ACEEE.
Yao, Y., Zhao, X., Liu, C., Rong, J., Zhang, Y., Dong, Z., and Su, Y. (2020). Vehicle fuel consumption prediction method based on driving behavior data collected from smartphones. Journal of Advanced Transportation, 2020:11 p.
Yen, M., Tian, S., Lin, Y., Yang, C.-W., and Chen, C.-C. (2021). Combining a universal obd-ii module with deep learning to develop an eco-driving analysis system. Applied Sciences, 11:17 p.
Publicado
19/10/2026
Como Citar
ARAÚJO, Laura L. dos Santos; GORLA, Paulo H. K.; ECHEVERRIA, Allan C. F.; PFITSCHER, Ricardo J..
Embedded Reinforcement Learning: Paving the Roads Towards Fuel Efficiency. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 2643-2654.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.30072.
