Research with TORCS
The publications below are a non-exhaustive selection of scholarly works whose authors report using TORCS in their experiments or research infrastructure. Inclusion indicates documented use of TORCS only; it does not imply affiliation, endorsement, reproducibility, or validation of the authors' methods or conclusions by the TORCS project.
Competition platform
-
Peer-reviewed Journal article, 2010
The 2009 Simulated Car Racing Championship
D. Loiacono et al. IEEE Transactions on Computational Intelligence and AI in Games, 2(2), 131-147
The authors report: Documents a TORCS-based competition framework in which submitted controllers drove simulated races.
Evidence checked in: title and abstract (championship built on the TORCS simulator)
Neuroevolution and evolutionary methods
-
Peer-reviewed Journal article, 2022
Overtaking Uncertainty With Evolutionary TORCS Controllers: Combining BLX With Decreasing Alpha Operator and Grand Prix Selection
M. Salem, A. M. Mora, and J. J. Merelo Guervós. IEEE Transactions on Games, 14(2), 318-327
The authors report: Evolves and evaluates TORCS controllers in racing and overtaking scenarios against other controllers.
Evidence checked in: title and abstract (evolutionary TORCS controllers)
-
Peer-reviewed Journal article, 2010
Learning to Drive in the Open Racing Car Simulator Using Online Neuroevolution
L. Cardamone, D. Loiacono, and P. L. Lanzi. IEEE Transactions on Computational Intelligence and AI in Games, 2(3), 176-190
The authors report: Uses TORCS tracks and driving tasks to evolve controllers online and examine transfer across tracks.
Evidence checked in: abstract (evolution of driving controllers in the Open Racing Car Simulator)
Imitation learning
-
Peer-reviewed Conference paper, 2019
Gaze Training by Modulated Dropout Improves Imitation Learning
Y. Chen, C. Liu, L. Tai, M. Liu, and B. E. Shi. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 7756-7761
The authors report: Collects driving controls and eye-tracking data in TORCS for gaze-modulated imitation learning.
Evidence checked in: abstract (driving environment with eye tracking; TORCS named in methods)
-
Peer-reviewed Conference paper, 2009
Controller for TORCS created by imitation
J. Muñoz, G. Gutiérrez, and A. Sanchis. IEEE Symposium on Computational Intelligence and Games (CIG), 271-278
The authors report: Trains a neural-network TORCS controller from software-controller and human driving examples.
Evidence checked in: title (controller for TORCS created by imitation)
Vision-based reinforcement learning
-
Peer-reviewed Conference paper, 2013
Evolving large-scale neural networks for vision-based reinforcement learning
J. Koutnik, G. Cuccu, J. Schmidhuber, and F. Gomez. GECCO '13 (Genetic and Evolutionary Computation Conference), 1061-1068
The authors report: Uses TORCS as a vision-based reinforcement-learning task for evolved neural networks.
Evidence checked in: abstract (vision-based reinforcement learning in the racing game)
Autonomous driving and deep reinforcement learning
-
Peer-reviewed Journal article, 2025
Adaptive enhancements of autonomous lane keeping via advanced PER-TD3 framework
X. Peng, J. Liang, X. Zhang, H. Yang, and W. Lei. Frontiers in Artificial Intelligence, 8, article 1688764
The authors report: Reports validation of a reinforcement-learning lane-keeping framework on the TORCS platform.
Evidence checked in: abstract (PER-TD3 lane keeping tested on TORCS)
-
Peer-reviewed Journal article, 2025
Physics encoded blocks in residual neural network architectures for digital twin models
M. S. Zia et al. Machine Learning, 114(8), article 180
The authors report: Uses TORCS as one experimental domain for autonomous steering models that combine learned components with Pure Pursuit physics.
Evidence checked in: abstract (TORCS listed among the experimental domains)
-
Peer-reviewed Journal article, 2021
End-to-End Autonomous Driving Through Dueling Double Deep Q-Network
B. Peng et al. Automotive Innovation, 4(3), 328-337
The authors report: Applies a Dueling Double Deep Q-Network to a TORCS lane-keeping experiment using images and vehicle-motion information.
Evidence checked in: abstract (lane-keeping experiments on the TORCS simulator)
-
Peer-reviewed Journal article, 2021
MADRaS: Multi Agent Driving Simulator
A. Santara et al. Journal of Artificial Intelligence Research (JAIR), 70, 1517-1555
The authors report: Builds a multi-agent research environment on TORCS with programmable traffic, communication, stochastic actions, and a Gym-style interface.
Evidence checked in: abstract (multi-agent driving simulator built over TORCS)
-
Conference paper Conference paper, 2020
Short-Term Trajectory Planning in TORCS using Deep Reinforcement Learning
E. Capo and D. Loiacono. IEEE Symposium Series on Computational Intelligence (SSCI), 2327-2334
The authors report: Uses TORCS to study higher-level trajectory-planning actions with deep reinforcement-learning agents.
Evidence checked in: title and abstract (trajectory planning in TORCS)
-
Peer-reviewed Journal article, 2017
Deep Reinforcement Learning framework for Autonomous Driving
A. El Sallab, M. Abdou, E. Perot, and S. Yogamani. Electronic Imaging, 29(19), 70-76
The authors report: Reports testing deep reinforcement-learning driving methods in TORCS.
Evidence checked in: abstract (experiments on the TORCS racing simulator)
-
Peer-reviewed Conference paper, 2017
Virtual to Real Reinforcement Learning for Autonomous Driving
X. Pan, Y. You, Z. Wang, and C. Lu. British Machine Vision Conference (BMVC), paper 11
The authors report: Trains driving policies in TORCS and transforms simulator imagery for virtual-to-real experiments.
Evidence checked in: abstract (policy trained in the TORCS game environment)
-
Peer-reviewed Conference paper, 2015
DeepDriving: Learning Affordance for Direct Perception in Autonomous Driving
C. Chen, A. Seff, A. Kornhauser, and J. Xiao. IEEE International Conference on Computer Vision (ICCV), 2722-2730
The authors report: Uses TORCS imagery and driving data to train and evaluate an affordance-based perception and control approach.
Evidence checked in: abstract (training data from the TORCS game engine)
Brain-computer and human-signal research
-
Peer-reviewed Conference paper, 2014
A brain-computer interface for shared vehicle control on TORCS car racing game
D. Kim and S.-B. Cho. International Conference on Natural Computation (ICNC), 550-555
The authors report: Uses TORCS to evaluate shared control combining EEG-derived commands with automated driving control.
Evidence checked in: title and abstract (shared vehicle control on the TORCS game)
-
Peer-reviewed Journal article, 2011
EEG potentials predict upcoming emergency brakings during simulated driving
S. Haufe et al. Journal of Neural Engineering, 8(5), article 056001
The authors report: Uses a TORCS-based driving environment while recording EEG, EMG, and braking responses.
Evidence checked in: methods section (driving simulation based on the open-source game TORCS)
Corrections and missing works
This list is maintained by review: a work is included only when its TORCS use is explicit in the paper itself, and each entry records where that evidence was found. If you know of a missing publication or spot wrong metadata, please report it with the paper's DOI - corrections are applied after verification.
Reported uses in events and community work (not peer-reviewed scholarship) are listed on a separate page: Uses and events.