How Can Sensor Simulation Help Make Autonomous Cars Safer?

Autonomous cars rely on sensors to detect their environments. The correct processing of the sensors’ signals is a crucial factor for the safety of autonomous vehicles – and therefore must be thoroughly tested and validated. By simulating not only the testing scenario but also sensor models and their physical characteristics, you can validate the complete processing chain for different sensor technologies, efficiently, reliably, and in controlled conditions.

Transcript

00:00:00 Spotlight How can sensor simulation help make autonomous cars safer? Hi there. Have you ever wondered how autonomous cars perceive the streets and environment that they're about to navigate? Autonomous cars rely on sensors such as camera, radar, lidar or ultrasonic sensors, these sensors help detect the car's environment. The correct processing of these perceptions is a crucial factor with regard to the safety of autonomous vehicles and therefore must be thoroughly tested and validated. To do this, you can rely on real world testing or traditional lab setups, yet wouldn't it be far more efficient to test the functionality and

00:00:55 integrity of the perception earlier in the development process? For example, can you validate an ECU without a real hardware sensor? Enter sensor simulation, the integrated dSPACE toolchain simulates all the components of the car, such as engine and drivetrain, brake hydraulics and vehicle dynamics, as well as its environment. Road network and traffic signs, buildings, other vehicles and pedestrians allow for the creation of complex traffic scenarios. And of course, you can also emulate different sensor types that perceive these environments, such as radar, lidar and camera, as well as the effects and impairments that these sensor systems tend to suffer.

00:01:49 For example, for cameras sensors, you can implement geometrical distortion, vigneting, chromatic aberration or an obstruction of the viewing field in form of dirt or damage to the lens. For lidar and radar sensors, material properties are as important as weather conditions such as rain, snow or fog. The virtual testing scenario gives you complete control over these properties and conditions. In short, by simulating not only the testing scenario, but also the sensor model with its physical characteristics. You can validate the whole processing chain for different sensor technologies

00:02:36 reliably, efficiently and under controlled conditions. If you have any questions about sensor simulation, just drop us an email. dSPACE, your partner in simulation and validation.