Enhancing the Realism of Autonomous Driving Simulation with Real-Time Co-Simulation.

Qiwei Chen,Tiexin Wang,Chengjie Lu,Tao Yue,Shaukat Ali
DOI: https://doi.org/10.1145/3550356.3561558
2022-01-01
Abstract:Autonomous driving simulators are commonly used to develop autonomous driving systems (ADS) since they provide the flexibility to experiment with scenarios that could even be dangerous in a real setting. This flexibility, however, comes with the possibility of experimenting with unrealistic scenarios. To this end, we present an initial co-simulation framework integrating OpenModelica and CARLA to enable real-time communication between them. As a proof of concept, we experimented with two Modelica models (air resistance and energy consumption). We connected the two models with CARLA to enable real-time communication between them to ensure the realism of scenarios in addition to connecting CARLA with OpenWeather through its API to access real weather conditions. We conducted experiments with a specific virtual electric vehicle (Tesla Model 3) running on the Town 06 map in CARLA. Results provide preliminary evidence that co-simulation with Modelica models improved the realism of the virtual vehicle in CARLA.
What problem does this paper attempt to address?