Hidden Biases of End-to-End Driving Models (2023-06-13T00:00:00.000000Z)

TL;DR

TF++ is developed, a simple end-to-end method that ranks first on the Longest 6 and LAV benchmarks, gaining 11 driving score over the best prior work on Longest6.

Abstract

End-to-end driving systems have recently made rapid progress, in particular on CARLA. Independent of their major contribution, they introduce changes to minor system components. Consequently, the source of improvements is unclear. We identify two biases that recur in nearly all state-of-the-art methods and are critical for the observed progress on CARLA: (1) lateral recovery via a strong inductive bias towards target point following, and (2) longitudinal averaging of multimodal waypoint predictions for slowing down. We investigate the drawbacks of these biases and identify principled alternatives. By incorporating our insights, we develop TF++, a simple end-to-end method that ranks first on the Longest6 and LAV benchmarks, gaining 11 driving score over the best prior work on Longest6.

Authors

Andreas Geiger

18 papers

Kashyap Chitta

4 papers

Bernhard Jaeger

2 papers

TL;DR

Abstract

Authors

References44 items

Coaching a Teachable Student

ReasonNet: End-to-End Driving with Temporal and Global Reasoning

Think Twice before Driving: Towards Scalable Decoders for End-to-End Autonomous Driving

PlanT: Explainable Planning Transformers via Object-Level Representations

Model-Based Imitation Learning for Urban Driving

Safety-Enhanced Autonomous Driving Using Interpretable Sensor Fusion Transformer

Trajectory-guided Control Prediction for End-to-end Autonomous Driving: A Simple yet Strong Baseline

TransFuser: Imitation With Transformer-Based Sensor Fusion for Autonomous Driving

Learning from All Vehicles

VISTA 2.0: An Open, Data-driven Simulator for Multimodal Sensing and Policy Learning for Autonomous Vehicles

NEAT: Neural Attention Fields for End-to-End Autonomous Driving

End-to-End Urban Driving by Imitating a Reinforcement Learning Coach

Learning to drive from a world on rails

Multi-Modal Fusion Transformer for End-to-End Autonomous Driving

Representations

MP3: A Unified Model to Map, Perceive, Predict and Plan

Fighting Copycat Agents in Behavioral Cloning from Observation Histories

Learning Situational Driving

Label Efficient Visual Abstractions for Autonomous Driving

Shortcut learning in deep neural networks

Designing Network Design Spaces

Learning by Cheating

End-to-End Model-Free Reinforcement Learning for Urban Driving Using Implicit Affordances

Exploring the Limitations of Behavior Cloning for Autonomous Driving

ChauffeurNet: Learning to Drive by Imitating the Best and Synthesizing the Worst

Conditional Affordance Learning for Driving in Urban Environments

On the Convergence of Adam and Beyond

Decoupled Weight Decay Regularization

CARLA: An Open Urban Driving Simulator

End-to-End Driving Via Conditional Imitation Learning

Attention is All you Need

Layer Normalization

Gaussian Error Linear Units (GELUs)

End to End Learning for Self-Driving Cars

Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation

LASzip: Lossless Compression of Lidar Data

A Reduction of Imitation Learning and Structured Prediction to No-Regret Online Learning

The unscented Kalman filter for nonlinear estimation

Supplementary Material for Exploring Data Aggregation in Policy Learning for Vision-based Urban Autonomous Driving

Fighting Copycat Agents in Behavioral Cloning from Multiple Observations

Carla autonomous driving leaderboard

Kalman and bayesian filters in python

ALVINN, an autonomous land vehicle in a neural network

Sigma-point kalman filters for probabilistic inference in dynamic state-space models

Field of Study

Journal Information

Name

Page

Venue Information

Name

Type

URL

Alternate Names