Search tips
 This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Autonomous Vehicles in the Cyberspace: Accelerating Testing via Computer Simulation

Technical Paper
2018-01-1078
ISSN: 0148-7191, e-ISSN: 2688-3627
DOI: https://doi.org/10.4271/2018-01-1078
Published April 3, 2018 by SAE International in United States
Annotation ability available
 
Sector:
Automotive  
Language: English

Abstract

We present an approach in which an open-source software infrastructure is used for testing the behavior of autonomous vehicles through computer simulation. This software infrastructure is called CAVE, from Connected Autonomous Vehicle Emulator. As a software platform that allows rapid, low-cost and risk-free testing of novel designs, methods and software components, CAVE accelerates and democratizes research and development activities in the field of autonomous navigation. CAVE is (a) heterogeneous and multi-agent, in that it supports the simulation of heterogeneous traffic scenarios involving conventional, assisted, and autonomous vehicles as well as pedestrians and cyclists; (b) open platform, as it allows any client that subscribes to a standard application programming interface (API) to remotely plug into the emulator and engage in multi-participant traffic scenarios that bring together autonomous agents from different solution providers; (c) vehicle-to-vehicle (V2V) communication emulation ready, owing to its ability to simulate the V2V data exchange enabled in real-world scenarios by ad-hoc dedicated short range communication (DSRC) protocols; and (d) open-source, as the software infrastructure will be available under a BSD3 license in a public repository for unrestricted use and redistribution. CAVE provides three immediate benefits. First, it serves as a development platform for algorithms that seek to establish path planning policies for autonomous vehicles operating in heterogeneous traffic scenarios; i.e., it enables the rapid and safe testing of “work in progress” piloting computer programs (PCPs). Second, it enables auditing of existing path planning policies by exposing connected and/or autonomous vehicles to scenarios that would be costly, time consuming and/or dangerous to consider in real-world testing. Third, the CAVE will provide a scalable, high-throughput, virtual proving ground that exposes heterogeneous traffic complexity which would not otherwise emerge in actual single-vehicle testing conducted in controlled environments. We present early results of a test case in which 30 autonomous vehicles negotiate a busy intersection in Madison, WI, without the need of traffic lights, simply by using sensors and communicating via DSRC.