Poster: A Case for Heterogenous Co-Simulation of Cooperative and Autonomous Driving

Many exciting future research topics in the field of Cooperative Autonomous Vehicles (CAVs) require the simulation of both connectivity and automation components. However, existing simulation tools focus on only one of these two aspects while making idealistic assumptions about the other. In this work, we motivate the use of established libraries such as gRPC to couple existing independent simulation tools tailored to either connectivity or automation, and demonstrate the feasibility of such an approach. We also describe an Open Source reference implementation coupling Carla and Veins.

I. Introduction

For many years, the Cooperative Autonomous Vehicle (CAV) research community has relied on simulations that really only focus on one of the two aspects - cooperation or autonomy - and model the other as an idealized process. Thus, in the context of autonomous driving research, networking and communication have often been considered a solved problem, leading to assumptions such as unit disk coverage and channels with infinite capacity, uninterrupted availability, and perfect robustness. Conversely, in the context of cooperative driving research, sensing has often been considered as an abstract process generating opaque chunks of data to be shared.