Autonomous vehicle technology and advanced driver-assistance systems (ADAS) are advancing every year, and its reliability is dependent on both sophisticated engineering and testing. Much has been written about real word testing of autonomous vehicles, but little has been written regarding the simulation portion. The goal of this post is to briefly discuss virtual simulation technology available for autonomous vehicles and ADAS.
Many software applications exist, but can only perform a very specific type of simulation. As an example, a software application may be exclusively dedicated to simulating traffic conditions and a separate application may be dedicated to simulation vehicle handling and other dynamics.
Autonomous vehicle software has progressed far enough that these separate capabilities are now available together. One such simulation tool is the Vehicle Test Drive (VTD) platform and enables engineers to simulate various aspects of autonomous vehicles including:
- Vehicle Dynamics
- AI Driver
- 3D Environment
Below is a quick summary of real world details that can be simulated on VTD.
Vehicles can quickly change velocity and direction, and these sudden changes influence the dynamics of the vehicle significantly. Think of a vehicle suddenly maneuvering an obstacle. Other influences include how vehicle mechanical systems perform during operation, for example a steering wheel changes the direction of the wheels or the vehicle suspension reacts to changes in motion. The list of dynamic behaviors of a vehicle is quite extensive, and with VTD, accurate vehicle dynamic models are included in autonomous vehicle simulations.
Artificial Intelligence (AI) Driver
The AI is one of the most critical aspects of autonomous technology, and simulations must take into account the behavior of AI. Each company will develop its own artificial intelligence driver, and VTD allows connecting proprietary AI, allowing for simulations the include the AI driver.
An autonomous vehicle will never be by itself, and instead will be surrounded by other autonomous, human-driven or parked vehicles, bicycles, pedestrians and the list goes on. Such traffic conditions can be replicated in a simulation environment when using VTD.
The AI of an autonomous vehicle will not see the road in the traditional sense. Instead, AI will depend on various active safety sensors, sometimes combined together, such as RADAR, LiDar and cameras to identify road markers, other vehicles, pedestrians, sign posts and obstacles. These sensors are responsible for sending information to the AI, and the success of information exchange is dependent on timing, organization, and many other factors. The transmissions and exchange of signals from sensor to AI can be simulated in VTD.
Every highway, road and street is different. Some streets stretch mountain sides, some are paved while others are not, some roads are situated in quiet towns and others are located in dense cities. In order to perform an accurate autonomous vehicle simulation, the virtual road must be accurate. With VTD, the virtual environment can be created by digitizing roads using advanced mapping technology.
2019 MSC Software Global User’s Conference
This year’s global user conference will showcase the latest and upcoming simulation methods and their use for additive manufacturing, autonomous vehicles and more. You are invited to this year’s conference.