Quantification of Distributionally Robust Risk of Cascade of Failures in Platoon of Vehicles

TL;DR

This paper introduces a distributionally robust risk framework to analyze and quantify cascading failures in autonomous vehicle platoons, considering network delays, noise, and communication structures, validated through simulations.

Contribution

It develops a novel risk quantification method for cascading failures in vehicle platoons using a distributionally robust approach with closed-form formulas.

Findings

Risk formulas show the impact of time-delay and noise.

Network connectivity influences cascade likelihood.

Framework aids in designing safer vehicle platoons.

Abstract

Achieving safety is a critical aspect of attaining autonomy in a platoon of autonomous vehicles. In this paper, we propose a distributionally robust risk framework to investigate cascading failures in platoons. To examine the impact of network connectivity and system dynamics on the emergence of cascading failures, we consider a time-delayed network model of the platoon of vehicles as a benchmark. To study the cascading effects among pairs of vehicles in the platoon, we use the measure of conditional distributionally robust functional. We extend the risk framework to quantify cascading failures by utilizing a bi-variate normal distribution. Our work establishes closed-form risk formulas that illustrate the effects of time-delay, noise statistics, underlying communication graph, and sets of soft failures. The insights gained from our research can be applied to design safe platoons that are robust to the risk of cascading failures. We validate our results through extensive simulations.