Learning to drive from a world on rails

TL;DR

This paper introduces a model-based, vision-driven driving policy learned from pre-recorded logs under a world-on-rails assumption, achieving high performance and sample efficiency in simulation benchmarks.

Contribution

It proposes a novel approach that simplifies learning by assuming a nonreactive environment, enabling efficient training of a vision-based driving policy from pre-recorded data.

Findings

Ranks first on the CARLA leaderboard with 25% higher score

Uses 40 times less data than previous methods

Outperforms state-of-the-art model-free RL in ProcGen benchmarks

Abstract

We learn an interactive vision-based driving policy from pre-recorded driving logs via a model-based approach. A forward model of the world supervises a driving policy that predicts the outcome of any potential driving trajectory. To support learning from pre-recorded logs, we assume that the world is on rails, meaning neither the agent nor its actions influence the environment. This assumption greatly simplifies the learning problem, factorizing the dynamics into a nonreactive world model and a low-dimensional and compact forward model of the ego-vehicle. Our approach computes action-values for each training trajectory using a tabular dynamic-programming evaluation of the Bellman equations; these action-values in turn supervise the final vision-based driving policy. Despite the world-on-rails assumption, the final driving policy acts well in a dynamic and reactive world. At the time of writing, our method ranks first on the CARLA leaderboard, attaining a 25% higher driving score while using 40 times less data. Our method is also an order of magnitude more sample-efficient than state-of-the-art model-free reinforcement learning techniques on navigational tasks in the ProcGen benchmark.