Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Good Data Is All Imitation Learning Needs

Amir Samadi, Konstantinos Koufos, Kurt Debattista, Mehrdad Dianati

TL;DR

The use of Counterfactual Explanations (CFEs) are introduced as a novel data augmentation technique for end-to-end ADS, to enhance the robustness of traditional teacher-student models, imitation learning, and behaviour cloning in the context of Autonomous/Automated Driving Systems.

Abstract

In this paper, we address the limitations of traditional teacher-student models, imitation learning, and behaviour cloning in the context of Autonomous/Automated Driving Systems (ADS), where these methods often struggle with incomplete coverage of real-world scenarios. To enhance the robustness of such models, we introduce the use of Counterfactual Explanations (CFEs) as a novel data augmentation technique for end-to-end ADS. CFEs, by generating training samples near decision boundaries through minimal input modifications, lead to a more comprehensive representation of expert driver strategies, particularly in safety-critical scenarios. This approach can therefore help improve the model's ability to handle rare and challenging driving events, such as anticipating darting out pedestrians, ultimately leading to safer and more trustworthy decision-making for ADS. Our experiments in the CARLA simulator demonstrate that CF-Driver outperforms the current state-of-the-art method, achieving a higher driving score and lower infraction rates. Specifically, CF-Driver attains a driving score of 84.2, surpassing the previous best model by 15.02 percentage points. These results highlight the effectiveness of incorporating CFEs in training end-to-end ADS. To foster further research, the CF-Driver code is made publicly available.

Good Data Is All Imitation Learning Needs

TL;DR

The use of Counterfactual Explanations (CFEs) are introduced as a novel data augmentation technique for end-to-end ADS, to enhance the robustness of traditional teacher-student models, imitation learning, and behaviour cloning in the context of Autonomous/Automated Driving Systems.

Abstract

In this paper, we address the limitations of traditional teacher-student models, imitation learning, and behaviour cloning in the context of Autonomous/Automated Driving Systems (ADS), where these methods often struggle with incomplete coverage of real-world scenarios. To enhance the robustness of such models, we introduce the use of Counterfactual Explanations (CFEs) as a novel data augmentation technique for end-to-end ADS. CFEs, by generating training samples near decision boundaries through minimal input modifications, lead to a more comprehensive representation of expert driver strategies, particularly in safety-critical scenarios. This approach can therefore help improve the model's ability to handle rare and challenging driving events, such as anticipating darting out pedestrians, ultimately leading to safer and more trustworthy decision-making for ADS. Our experiments in the CARLA simulator demonstrate that CF-Driver outperforms the current state-of-the-art method, achieving a higher driving score and lower infraction rates. Specifically, CF-Driver attains a driving score of 84.2, surpassing the previous best model by 15.02 percentage points. These results highlight the effectiveness of incorporating CFEs in training end-to-end ADS. To foster further research, the CF-Driver code is made publicly available.
Paper Structure (18 sections, 2 equations, 3 figures, 2 tables)

This paper contains 18 sections, 2 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Background and Related work
  3. Counterfactual Explanations
  4. CARLA Drivers and Imitation Learning
  5. Methodology
  6. Problem Definition
  7. Counterfactual Driver
  8. Data Collection and Tree-based Classifier:
  9. CF Example Generation:
  10. Collect demonstration data:
  11. Training the Immitation Learner with Augmented Data:
  12. Evaluations
  13. Implementation Details
  14. Performance Metrics
  15. Performance Evaluation
  16. ...and 3 more sections

Figures (3)

  • Figure 1: Data Augmentation with CFEs for Improved Knowledge Distillation.
  • Figure 2: Overview of the three-step process used to train the proposed CF-Driver. So lid lines surrounding the model boxes indicate fixed models, while dashed lines represent models undergoing training.
  • Figure 3: Illustration of the dataset consisting of front camera and Bird's Eye View frames, with original and counterfactual examples. In Scenario 1, the top row shows an original frame from Town01, where the vehicle is approaching a green traffic light. The bottom row demonstrates a counterfactual example generated by the CF generator, where the traffic light has been changed to red. The expert driver labels this counterfactual scenario as "stop". In Scenario 2, the CF generator modified the motorcycle's location, bringing it closer to the ego-vehicle. This change shifts the expert's decision from "go" to "stop", demonstrating how counterfactual examples can refine the imitation learner's decision boundary.