Driving Policy Transfer via Modularity and Abstraction

TL;DR

This paper introduces a modular and abstracted approach to transfer autonomous driving policies from simulation to real-world robots, demonstrating successful deployment without fine-tuning across diverse conditions.

Contribution

It proposes a novel modular framework that encapsulates driving policies, facilitating effective transfer from simulation to real-world autonomous vehicles without additional training.

Findings

Successful transfer of policies to a 1/5-scale robotic truck

No fine-tuning required for deployment in new environments

Effective in diverse urban conditions across continents

Abstract

End-to-end approaches to autonomous driving have high sample complexity and are difficult to scale to realistic urban driving. Simulation can help end-to-end driving systems by providing a cheap, safe, and diverse training environment. Yet training driving policies in simulation brings up the problem of transferring such policies to the real world. We present an approach to transferring driving policies from simulation to reality via modularity and abstraction. Our approach is inspired by classic driving systems and aims to combine the benefits of modular architectures and end-to-end deep learning approaches. The key idea is to encapsulate the driving policy such that it is not directly exposed to raw perceptual input or low-level vehicle dynamics. We evaluate the presented approach in simulated urban environments and in the real world. In particular, we transfer a driving policy trained in simulation to a 1/5-scale robotic truck that is deployed in a variety of conditions, with no finetuning, on two continents. The supplementary video can be viewed at https://youtu.be/BrMDJqI6H5U