Ab­stract

Au­tonomous ve­hi­cles need a com­plete and ro­bust per­cep­tion of their en­vi­ron­ment to cor­rectly un­der­stand the sur­round­ing traf­fic scene and come to the right de­ci­sions. Mak­ing use of ve­hi­cle-to-ve­hi­cle (V2V) com­mu­ni­ca­tion can im­prove the per­cep­tion ca­pa­bil­i­ties of au­tonomous ve­hi­cles by ex­tend­ing the range of their own local sen­sors. For the de­vel­op­ment of ro­bust co­op­er­a­tive per­cep­tion sys­tems it is nec­es­sary to in­clude vary­ing en­vi­ron­men­tal con­di­tions to the sce­nar­ios used for val­i­da­tion. In this paper we pre­sent a new ap­proach to in­ves­ti­gate a co­op­er­a­tive per­cep­tion pipeline within sim­u­la­tion under vary­ing rain con­di­tions. We demon­strate our ap­proach on the ex­am­ple of a com­plete vi­sion-based co­op­er­a­tive per­cep­tion pipeline. Sce­nar­ios with a vary­ing num­ber of co­op­er­a­tive ve­hi­cles under dif­fer­ent syn­thet­i­cally gen­er­ated rain vari­a­tions are used to show the in­flu­ence of rain on local and co­op­er­a­tive per­cep­tion.