An Open Source Co-simulation Platform for Self-driving Vehicles
Author | : Yishen Jin |
Publisher | : |
Total Pages | : 0 |
Release | : 2021 |
Genre | : Automated vehicles |
ISBN | : |
Download An Open Source Co-simulation Platform for Self-driving Vehicles Book in PDF, Epub and Kindle
With the increasing demands for testing Automated Vehicles (AVs) and Advanced Driver Assistance Systems (ADAS), a large-scale virtual verification and validation framework becomes valuable for three reasons. First, for AV and ADAS software testing, it is infeasible to cover on-road conditions exhaustively. Second, developing a virtual testing environment can reduce operating costs greatly. Third, software failure in AVs or ADAS is safety-critical and can result directly in fatal accidents. To address the aforementioned issues, this work focuses on developing an open-source platform for virtual testing with the capability of the generation of large-scale traffic simulations, synchronization between traffic scenes and 3D environment, and integration with existing sensor models. Specifically, a virtual validation and verification environment framework for AV software testing is developed in this work by integrating a microscopic traffic simulator, Simulation of Urban Mobility (SUMO), with a 3D-rendering software, Unreal Engine (UE). In order to incorporate the variability in testing scenarios such as surrounding dynamic objects, obstacles, road networks, and infrastructure features, the framework provides a modular software block-set for the virtual testing of AV/ADAS controllers. This work presents the architecture of the synchronization of information from vehicles, traffic signals, and pedestrians between SUMO and UE. With the platform developed, large-scale test cases can be generated efficiently in parallel between SUMO and UE. Specific test cases can be visualized and analyzed individually. As a result, edge cases with low probability but catastrophic outcomes can be tested safely in the virtual environment.