EMPERROR: A Flexible Generative Perception Error Model for Probing Self-Driving Planners

TL;DR

This paper introduces EMPERROR, a transformer-based generative perception error model that realistically simulates perception noise, enabling more thorough testing and evaluation of self-driving planners under realistic error conditions.

Contribution

The paper presents a novel transformer-based generative perception error model that better captures failure modes of perception systems for robust self-driving planner evaluation.

Findings

EMPERROR produces more realistic perception errors than previous models.

Using EMPERROR increases planner collision rates by up to 85%.

It improves the robustness testing of self-driving systems.

Abstract

To handle the complexities of real-world traffic, learning planners for self-driving from data is a promising direction. While recent approaches have shown great progress, they typically assume a setting in which the ground-truth world state is available as input. However, when deployed, planning needs to be robust to the long-tail of errors incurred by a noisy perception system, which is often neglected in evaluation. To address this, previous work has proposed drawing adversarial samples from a perception error model (PEM) mimicking the noise characteristics of a target object detector. However, these methods use simple PEMs that fail to accurately capture all failure modes of detection. In this paper, we present EMPERROR, a novel transformer-based generative PEM, apply it to stress-test an imitation learning (IL)-based planner and show that it imitates modern detectors more faithfully than previous work. Furthermore, it is able to produce realistic noisy inputs that increase the planner's collision rate by up to 85%, demonstrating its utility as a valuable tool for a more complete evaluation of self-driving planners.