Learning Robust Control Policies for End-to-End Autonomous Driving From Data-Driven Simulation

In this work, we present a data-driven simulation and training engine capable of learning end-to-end autonomous vehicle control policies using only sparse rewards. By leveraging real, human-collected trajectories through an environment, we render novel training data that allows virtual agents to dri...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE robotics and automation letters Jg. 5; H. 2; S. 1142 - 1149
Hauptverfasser: Amini, Alexander, Gilitschenski, Igor, Phillips, Jacob, Moseyko, Julia, Banerjee, Rohan, Karaman, Sertac, Rus, Daniela
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Piscataway IEEE 01.04.2020
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
Access control
autonomous agents
Autonomous vehicles
data-driven simulation
Deep learning in robotics and automation
Machine learning
Policies
real world reinforcement learning
Roads & highways
Robust control
Semantics
Simulation
Training
ISSN:2377-3766, 2377-3766
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In this work, we present a data-driven simulation and training engine capable of learning end-to-end autonomous vehicle control policies using only sparse rewards. By leveraging real, human-collected trajectories through an environment, we render novel training data that allows virtual agents to drive along a continuum of new local trajectories consistent with the road appearance and semantics, each with a different view of the scene. We demonstrate the ability of policies learned within our simulator to generalize to and navigate in previously unseen real-world roads, without access to any human control labels during training. Our results validate the learned policy onboard a full-scale autonomous vehicle, including in previously un-encountered scenarios, such as new roads and novel, complex, near-crash situations. Our methods are scalable, leverage reinforcement learning, and apply broadly to situations requiring effective perception and robust operation in the physical world.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2377-3766
2377-3766
DOI:10.1109/LRA.2020.2966414