Beyond the Racetrack: Evaluating the Transferability of Formula 1 AI and Telemetry Technologies to Passenger Vehicles

This research investigates the level of transferability of selected Artificial Intelligence (AI) and Telemetry (TM) techniques from Formula 1 to passenger cars which are equipped with lower number of sensors and computing power. The research is not to copy a full fledged Formula 1 telemetry system, but to take over its analytical method and to use consumer grade OBD-II telemetry to deliver useful predictions. After data quality checks and duplicate identification, three OBD-II driving datasets, with a total number of 555,000 raw observations were analysed. Binary driver-state classification was conducted with Logistic Regression (as a baseline model) and with the nonlinear ensemble models Random Forest, Extra Trees, LightGBM and XGBoost. Due to the imbalance between the classes, the performance of the model was assessed using accuracy, macroprecision, macro-recall, Macro-F1 score and confusion-matrix analysis. The best accuracy on any one of the individual driver datasets was 99.9873%. The highest accuracy (99.9949%) and Macro-F1 (99.9924%) were obtained by the best models for the combined dataset. The driver datasets reported the greatest relationships with the classification target, with throttle position, accelerator-pedal position, engine load, and intake-related variables showing the strongest relationships. The results indicate that a rather modest sensor suite can be used to produce useful vehicle intelligence. The results should not be viewed as an immediate demonstration of how ready the technology is to be used in the field, however. Telemetry for the vehicles is sequential, meaning that the observations in the randomly split training and test sets can have a high degree of correlation. Other tests such as driver-wise, trip-wise, time-based and cross-vehicle tests are thus required for additional validation. Both tire and brake wear prediction are mentioned as future research directions, rather than results of experiments, since there do not exist any validated physical measurements of degradation from the available data sets. This paper focuses on the application of machine learning techniques in driver behavior classification, predictive maintenance, vehicle telemetry, OBD-II and CAN bus, and tire/brake degradation for application in Formula 1, and introduces the concepts of Random Forest and XGBoost models.

Authors

Institutions

Publication Details

Journal
International Journal for Research in Applied Science and Engineering Technology
Published
2026-11-17
DOI
https://doi.org/10.22214/ijraset.2026.85085
Primary Topic
Human-Automation Interaction and Safety
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Beyond the Racetrack: Evaluating the Transferability of Formula 1 AI and Telemetry Technologies to Passenger Vehicles

Akshata powar, Sheetal S. Shevkari, Prajwal Mane, Mahesh Rawate et al.
3 links
International Journal for Research in Applied Science and Engineering Technology
Human-Automation Interaction and Safety
article

Beyond the Racetrack: Evaluating the Transferability of Formula 1 AI and Telemetry Technologies to Passenger Vehicles

Akshata powar, Sheetal S. Shevkari, Prajwal Mane, Mahesh Rawate, Aarti Tonpe, Prathamesh Bhamare, Aishwarya Navagune
article en

Abstract

This research investigates the level of transferability of selected Artificial Intelligence (AI) and Telemetry (TM) techniques from Formula 1 to passenger cars which are equipped with lower number of sensors and computing power. The research is not to copy a full fledged Formula 1 telemetry system, but to take over its analytical method and to use consumer grade OBD-II telemetry to deliver useful predictions. After data quality checks and duplicate identification, three OBD-II driving datasets, with a total number of 555,000 raw observations were analysed. Binary driver-state classification was conducted with Logistic Regression (as a baseline model) and with the nonlinear ensemble models Random Forest, Extra Trees, LightGBM and XGBoost. Due to the imbalance between the classes, the performance of the model was assessed using accuracy, macroprecision, macro-recall, Macro-F1 score and confusion-matrix analysis. The best accuracy on any one of the individual driver datasets was 99.9873%. The highest accuracy (99.9949%) and Macro-F1 (99.9924%) were obtained by the best models for the combined dataset. The driver datasets reported the greatest relationships with the classification target, with throttle position, accelerator-pedal position, engine load, and intake-related variables showing the strongest relationships. The results indicate that a rather modest sensor suite can be used to produce useful vehicle intelligence. The results should not be viewed as an immediate demonstration of how ready the technology is to be used in the field, however. Telemetry for the vehicles is sequential, meaning that the observations in the randomly split training and test sets can have a high degree of correlation. Other tests such as driver-wise, trip-wise, time-based and cross-vehicle tests are thus required for additional validation. Both tire and brake wear prediction are mentioned as future research directions, rather than results of experiments, since there do not exist any validated physical measurements of degradation from the available data sets. This paper focuses on the application of machine learning techniques in driver behavior classification, predictive maintenance, vehicle telemetry, OBD-II and CAN bus, and tire/brake degradation for application in Formula 1, and introduces the concepts of Random Forest and XGBoost models.

International Journal for Research in Applied Science and Engineering TechnologyVol. 14(11)
G.S. Science, Arts And Commerce College (IN)
Human-Automation Interaction and Safety
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon