On the Choice of Data for Efficient Training and Validation of End-to-End Driving Models
On the map: End-to-end autonomous driving → End-to-end system architecture
Abstract (arXiv)
The emergence of data-driven machine learning (ML) has facilitated significant progress in many complicated tasks such as highly-automated driving. While much effort is put into improving the ML models and learning algorithms in such applications, little focus is put into how the training data and/or validation setting should be designed. In this paper we investigate the influence of several data design choices regarding training and validation of deep driving models trainable in an end-to-end fashion. Specifically, (i) we investigate how the amount of training data influences the final driving performance, and which performance limitations are induced through currently used mechanisms to generate training data. (ii) Further, we show by correlation analysis, which validation design enables the driving performance measured during validation to generalize well to unknown test environments. (iii) Finally, we investigate the effect of random seeding and non-determinism, giving insights which reported improvements can be deemed significant. Our evaluations using the popular CARLA simulator provide recommendations regarding data generation and driving route selection for an efficient future development of end-to-end driving models.
Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in End-to-end system architecture, one of 9 directions in End-to-end autonomous driving. Builds on 41 of its 52 references that are in the map. First posted on arXiv 1 Jun 2022.
Builds on
41 of 52 references.
Closest ancestors · 1
Same direction, newest first
Other references · 40
- 2017Attention Is All You Needcited by 59,932
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2019PyTorch: An Imperative Style, High-Performance Deep Learning Librarycited by 20,160
- 2014Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translationcited by 4,016
- 2016DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFscited by 3,239
- 2019Scalability in Perception for Autonomous Driving: Waymo Open Datasetcited by 2,436
- 2017CARLA: An Open Urban Driving Simulatorcited by 2,431
- 2016End to End Learning for Self-Driving Carscited by 1,000
- 2016Playing for Data: Ground Truth from Computer Gamescited by 816
- 2017End-to-end Driving via Conditional Imitation Learningcited by 479
- 2019Learning by Cheatingcited by 361
- 2019Exploring the Limitations of Behavior Cloning for Autonomous Drivingcited by 343
28 more
- 2018Learning to Drive in a Daycited by 266
- 2021NEAT: Neural Attention Fields for End-to-End Autonomous Drivingcited by 174
- 2020Perceive, Predict, and Plan: Safe Motion Planning Through Interpretable Semantic Representationscited by 150
- 2020Self-Supervised Monocular Depth Estimation: Solving the Dynamic Object Problem by Semantic Guidancecited by 148
- 2016End-to-end Learning of Driving Models from Large-scale Video Datasetscited by 147
- 2019Self-Driving Cars: A Surveycited by 147
- 2019End-to-End Model-Free Reinforcement Learning for Urban Driving using Implicit Affordancescited by 147
- 2020Can Autonomous Vehicles Identify, Recover From, and Adapt to Distribution Shifts?cited by 139
- 2021DriveGAN: Towards a Controllable High-Quality Neural Simulationcited by 125
- 2021Learning to drive from a world on railscited by 114
- 2018Driving Policy Transfer via Modularity and Abstractioncited by 91
- 2019Urban Driving with Conditional Imitation Learningcited by 84
- 2018Conditional Affordance Learning for Driving in Urban Environmentscited by 70
- 2018On Offline Evaluation of Vision-based Driving Modelscited by 62
- 2019Simulation-based reinforcement learning for real-world autonomous drivingcited by 51
- 2021Multi-task Learning with Attention for End-to-end Autonomous Drivingcited by 49
- 2018Learning to Drive from Simulation without Real World Labelscited by 39
- 2018Variational End-to-End Navigation and Localizationcited by 34
- 2019Discrete Residual Flow for Probabilistic Pedestrian Behavior Predictioncited by 34
- 2019Driving with Style: Inverse Reinforcement Learning in General-Purpose Planning for Automated Drivingcited by 30
- 2021Trans4Trans: Efficient Transformer for Transparent Object Segmentation to Help Visually Impaired People Navigate in the Real Worldcited by 30
- 2018End-to-end Multi-Modal Multi-Task Vehicle Control for Self-Driving Cars with Visual Perceptioncited by 27
- 2020Explaining Autonomous Driving by Learning End-to-End Visual Attentioncited by 25
- 2020Perceive, Attend, and Drive: Learning Spatial Attention for Safe Self-Drivingcited by 24
- 2020Unsupervised BatchNorm Adaptation (UBNA): A Domain Adaptation Method for Semantic Segmentation Without Using Source Domain Representationscited by 24
- 2016Learning to Drive using Inverse Reinforcement Learning and Deep Q-Networkscited by 15
- 2021Carl-Lead: Lidar-based End-to-End Autonomous Driving with Contrastive Deep Reinforcement Learningcited by 11
- 2018Real-to-Virtual Domain Unification for End-to-End Autonomous Drivingcited by 10
Built on by
Citing papers per year: 2023: 1; 2024: 1.
2 papers.
Similar, not cited
Close in wording, no citation either way
- 2026How Far Can 5,500 Hours of Driving Take You? A Scaling Law Analysis of Video Diffusion Models
- 2024Data Scaling Laws for Imitation Learning-Based End-to-End Autonomous Driving
- 2025Data Scaling Laws for End-to-End Autonomous Driving
- 2023Scaling Planning for Automated Driving using Simplistic Synthetic Data
- 2026A Survey on End-to-End Autonomous Driving Training from the Perspectives of Data, Strategy, and Platform
- 2023Recent Advancements in End-to-End Autonomous Driving using Deep Learning: A Survey
- 2024Hidden Biases of End-to-End Driving Datasets
- 2026Post-Training in End-to-End Autonomous Driving