A practical benchmark for testing analysis and verification methods for continuous as well as hybrid systems is suggested. It consists of a platoon of controlled vehicles. The vehicles exchange information via a communication network that is subject to failure of single nodes or a complete loss of communication. The worst case scenario corresponding to the transition from a fully functioning communication between the vehicles to a total loss of communication is considered in particular. The safety properties of the considered networked platoon are investigated. The system is modeled as a hybrid automaton. The continuous part represents the dynamical behavior of the platoon and the discrete part are spontaneous events related to the switching communication topology. The proposed example is a linear time invariant system. Category: academic Difficulty: challenge
Context and OriginsThe benchmark suggested here has been already proposed as a practical example for testing methods for computing reachable sets for continuous as well as for hybrid systems. In [2], a performance comparison between approximation methods based on support functions and zonotopes with invariant computation methods combining linear matrix inequality (LMI) technique and ellipsoids was presented. The tightness of the approximation and the computation effort were particularly taken into consideration. With the platoon example, the goal was to test which of these methods provides the shortest safe gaps between the vehicles. This example is therefore challenging because of its state dimension.
Benchmark Description and Testing GoalsWe consider now the platoon of n vehicles illustrated in Figure 1. The spacing error e i is in this context defined as the difference between the distance d i of the truck i to its predecessor and a reference distance d ref,i . The principal goal of the reachability analysis for this system is to determine a lower bound for d ref,i assuring collision-free driving. We assume that each G