Today, the autonomous driving industry understands the importance of road testing for self-driving vehicles, and the industry is spending a fortune on it. However, is road testing really enough to help us reach level 5 autonomy in the foreseeable future?
As long as we only have to check a few use cases (in the range of tens), we can easily test them on real roads. However, in order to assure safety for Autonomous Vehicles, the number of conditions to be evaluated scales quickly to millions and there is no way to tackle it without simulations.
Autonomous Vehicle (AV) simulation is different from the traditional vehicle simulations in a sense that apart from the vehicle itself, also the “environment” in which the vehicle operates is fundamental to assessing how it copes with all driving situations.
Let’s take a look at some of the key elements/models that are needed in order to create a highly realistic virtual environment for ADAS/Autonomous vehicle simulation.
- Traffic & Pedestrians/Animals
- 3D Environment & Road Network
- Vehicle Dynamics
- Weather & Lighting
- Scenarios & Massive Scaling