Can Autonomous Vehicles Identify, Recover From, and Adapt to Distribution Shifts?

TL;DR

This paper introduces RIP and AdaRIP, epistemic uncertainty-aware planning methods for autonomous vehicles that detect, recover from, and adapt to distribution shifts, improving robustness in novel driving scenes.

Contribution

The paper proposes RIP and AdaRIP, novel planning algorithms that handle distribution shifts in autonomous driving, and introduces CARNOVEL, a new benchmark for evaluating control robustness to OOD scenarios.

Findings

Outperform current state-of-the-art in nuScenes prediction challenge.

Effective detection and recovery from distribution shifts.

New benchmark CARNOVEL for autonomous driving OOD robustness.

Abstract

Out-of-training-distribution (OOD) scenarios are a common challenge of learning agents at deployment, typically leading to arbitrary deductions and poorly-informed decisions. In principle, detection of and adaptation to OOD scenes can mitigate their adverse effects. In this paper, we highlight the limitations of current approaches to novel driving scenes and propose an epistemic uncertainty-aware planning method, called \emph{robust imitative planning} (RIP). Our method can detect and recover from some distribution shifts, reducing the overconfident and catastrophic extrapolations in OOD scenes. If the model's uncertainty is too great to suggest a safe course of action, the model can instead query the expert driver for feedback, enabling sample-efficient online adaptation, a variant of our method we term \emph{adaptive robust imitative planning} (AdaRIP). Our methods outperform current state-of-the-art approaches in the nuScenes \emph{prediction} challenge, but since no benchmark evaluating OOD detection and adaption currently exists to assess \emph{control}, we introduce an autonomous car novel-scene benchmark, \texttt{CARNOVEL}, to evaluate the robustness of driving agents to a suite of tasks with distribution shifts.