Risk-Aware Reasoning for Autonomous Vehicles

TL;DR

This paper presents a comprehensive risk-aware system architecture for autonomous vehicles that explicitly reasons about uncertainty in perception, intention, and planning to ensure safety and collision risk bounds.

Contribution

It introduces a novel white-box architecture integrating perception, intention recognition, and planning subsystems for risk-aware autonomous driving.

Findings

Quantifies uncertainty in perception and communication modalities.

Predicts driving style and intentions of other agents.

Propagates uncertainty through planning to bound collision risk.

Abstract

A significant barrier to deploying autonomous vehicles (AVs) on a massive scale is safety assurance. Several technical challenges arise due to the uncertain environment in which AVs operate such as road and weather conditions, errors in perception and sensory data, and also model inaccuracy. In this paper, we propose a system architecture for risk-aware AVs capable of reasoning about uncertainty and deliberately bounding the risk of collision below a given threshold. We discuss key challenges in the area, highlight recent research developments, and propose future research directions in three subsystems. First, a perception subsystem that detects objects within a scene while quantifying the uncertainty that arises from different sensing and communication modalities. Second, an intention recognition subsystem that predicts the driving-style and the intention of agent vehicles (and pedestrians). Third, a planning subsystem that takes into account the uncertainty, from perception and intention recognition subsystems, and propagates all the way to control policies that explicitly bound the risk of collision. We believe that such a white-box approach is crucial for future adoption of AVs on a large scale.