RECORD
Learning a Decision Module by Imitating Driver's Control Behaviors
Artificial Intelligence · Submitted 30 Nov 2019
abstract compiled · 2026-09-23 · no language-model review
The authors' main claim We show in the simulation experiments that our modular driving agent can generalize its driving decision and control to various complex scenarios where the rule-based programs fail.
The compiled record
The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
Open the compiled record
The abstract, sentence by sentence
Autonomous driving systems have a pipeline of perception, decision, planning, and control. The decision module processes information from the perception module and directs the execution of downstream planning and control modules. On the other hand, the recent success of deep learning suggests that this pipeline could be replaced by end-to-end neural control policies, however, safety cannot be well guaranteed for the data-driven neural networks. In this work, we propose a hybrid framework to learn neural decisions in the classical modular pipeline through end-to-end imitation learning. This hybrid framework can preserve the merits of the classical pipeline such as the strict enforcement of physical and logical constraints while learning complex driving decisions from data. To circumvent the ambiguous annotation of human driving decisions, our method learns high-level driving decisions by imitating low-level control behaviors. We show in the simulation experiments that our modular driving agent can generalize its driving decision and control to various complex scenarios where the rule-based programs fail. It can also generate smoother and safer driving trajectories than end-to-end neural policies.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Autonomous driving systems have a pipeline of perception, decision, planning, and control. | background 1.00 | 0.05 | 0.02 | asserted 1.00 |
| s1 | The decision module processes information from the perception module and directs the execution of downstream planning and control modules. | background 1.00 | 0.05 | 0.03 | asserted 1.00 |
| s2 | On the other hand, the recent success of deep learning suggests that this pipeline could be replaced by end-to-end neural control policies, however, safety cannot be well guaranteed for the data-driven neural networks. | background 0.85 | 0.05 | 0.04 | hedged possibility 0.88 |
| s3 | In this work, we propose a hybrid framework to learn neural decisions in the classical modular pipeline through end-to-end imitation learning. | contribution 0.94 | 0.88 | 0.13 | asserted 0.99 |
| s4 | This hybrid framework can preserve the merits of the classical pipeline such as the strict enforcement of physical and logical constraints while learning complex driving decisions from data. | contribution 0.80 | 0.44 | 0.22 | asserted 0.70 |
| s5 | To circumvent the ambiguous annotation of human driving decisions, our method learns high-level driving decisions by imitating low-level control behaviors. | method 0.94 | 0.86 | 0.07 | asserted 1.00 |
| s6 | We show in the simulation experiments that our modular driving agent can generalize its driving decision and control to various complex scenarios where the rule-based programs fail. | result 1.00 | 0.96 | 0.35 | asserted 1.00 |
| s7 | It can also generate smoother and safer driving trajectories than end-to-end neural policies. | result 1.00 | 0.92 | 0.25 | asserted 0.98 |
What was examined, and what was not
- arXiv listing abstract v3
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
The paper
Original abstract
Autonomous driving systems have a pipeline of perception, decision, planning, and control. The decision module processes information from the perception module and directs the execution of downstream planning and control modules. On the other hand, the recent success of deep learning suggests that this pipeline could be replaced by end-to-end neural control policies, however, safety cannot be well guaranteed for the data-driven neural networks. In this work, we propose a hybrid framework to learn neural decisions in the classical modular pipeline through end-to-end imitation learning. This hybrid framework can preserve the merits of the classical pipeline such as the strict enforcement of physical and logical constraints while learning complex driving decisions from data. To circumvent the ambiguous annotation of human driving decisions, our method learns high-level driving decisions by imitating low-level control behaviors. We show in the simulation experiments that our modular driving agent can generalize its driving decision and control to various complex scenarios where the rule-based programs fail. It can also generate smoother and safer driving trajectories than end-to-end neural policies.
Human review
How the literature uses it
1 arXiv paper cite this in 1 sentence
Papers citing it
Cites it as background 1 paper
-
End-to-end control approaches such as imitation learning have also been investigated to obtain the driving policy based on image inputs [ref] .
Multi-task Safe Reinforcement Learning for Navigating Intersections in Dense Traffic · Introduction · 0.80
What it cites · 3 sentences without a confident label
Cites as support 1 paper
-
Nor do they have robustness or stability guarantees [ref] , which are of great importance to the real-world deployment of classical controllers.
Adversarial Attacks on Neural Network Policies · Introduction · 0.99
Cites as background 2 papers
-
More recently, Kuefler et al. [ref] and Behbahani et al. [ref] learn an end-to-end policy in a GAIL [ref] -like manner.
Learning from Demonstration in the Wild · Related Work · 0.99
-
Rhinehart et al. [ref] propose to learn a generative predictor for vehicles' coordinate sequence.
Deep Imitative Models for Flexible Inference, Planning, and Control · Related Work · 0.84
Discussion
Know something about this paper? Leave a comment, or highlight any passage above to comment on it.