Deterministic environment and weather simulation
aiSim's proprietary, real-time rendering engine addresses all the requirements for a physically correct environment and weather simulation with complete determinism and repeatability. In aiSim 4, a broad range of weather conditions, including snowstorms, heavy rain, fog, and sunshine, can be simulated by leveraging state-of-the-art ray-tracing and rasterization technologies. In addition, multiple road parameters, including road painting deterioration and road degradation, can be easily configured to create even more diverse and challenging environments for perception stacks.
Physics-based sensor simulation in real time
Sensor simulation is a pivotal piece of AD software stacks' testing chains as the information of the surrounding environment, known by the autonomous vehicle, is provided by the sensors. In aiSim 4, sensor models utilize a physics-based approach to accurately generate a comprehensive range of simulated conditions and environmental interactions. The extensive physics-based and validated sensor model library include cameras, lidars, radars, and ultrasonic sensors. A few of the many use cases for sensor simulation:
- Simulate highly complex sensor setups in real-time due to aiSim's highly scalable architecture, which allows the distribution of raytracing-based rendering over multiple GPUs.
- Test how sensor degradations, including condensation or mud, and lens distortionslike chromatic aberration, lens flare, or bloom, impact the perception capabilities of camera sensors.
- Prototype with new sensor configurations quickly to test their impact on perception performance. Camera, lidar, and radar sensor frustum visualization help speed up the optimal sensor placement on the ego-vehicle with regards to the system requirements.
Synthetic data generation at scale
Data is critical to neural network (NN) development. Automatic annotation is the most effective and cheapest way to generate training and validation data from real-world recordings. But what about hard-to-capture scenarios or corner cases that hardly ever occur in real life? Automated driving systems still need to be aware of such events.
aiSim, combined with aiFab, supports the simple generation of scenarios at scale, with domain randomization, to replicate the great variability in real-world data. It covers locations and assets for various AD use cases, including highway, urban, and parking scenarios. Once the data is generated, it provides statistics and visual feedback so that users can understand their synthetic data.
Cloud-native UI & Open SDK
In aiSim 4, all the GUI functionalities will be available in a web-native user interface, ensuring a seamless customer experience and deployment as well as eliminating installation requirements in the customer environment.
aiSim is provided with a comprehensive software development kit (SDK) that enables developers to leverage aiSim's capabilities. In the SDK of aiSim 4, a handful of APIs and interfaces enable tailoring aiSim to the users' needs and support a seamless and performance integration in customer toolchains.
- Sensor API allows the development of custom sensor models or the integration of 3rd party sensor models. The SDK comes with built-in sensor plugin implementations and source code examples that can be used instantly in the simulation.
- Scenario API enables to develop and use an application that can control the actors of a scenario, i.e., the static and dynamic objects in the scene such as exo-vehicles, pedestrians, obstacles, etc.
Mature 3D asset pipeline
aiSim 4 provides an extensive set of 3D assets and tools required for efficient content creation & import to easily set up high-fidelity 3D environments.
- Vehicle models: pre-built, high-quality models, including passenger vehicles, trucks, emergency vehicles, trailers, truck trailers, motorcycles, vans, and forklifts.
- VRUs: various pedestrians, bicyclists, and people with wheelchairs, segway, and scooters with recoloring option. Integration of 3rd-party 3D human asset library with posed, rigged, and animated models.
- Maps: covering all the relevant ODDs, including highway, urban, country roads, and parking garages. The modular map pipeline can handle the most common HD map formats & vendors. Integration with RoadRunner(TM) to create and import custom maps into aiSim.
- 3D assets: thousands of 3D models, including barriers, posts, road signs, signs, traffic lights, and other road furniture. 3D asset creation, import and customization are enabled by the aiSim Unreal Editor Plugin supporting Unreal 5(TM).
- Scenarios: a library of 1,500+ pre-defined scenario templates complemented by 100+ event triggers and 50+ evaluation condition parameters. Scenario creation and customization is enabled by aiSim's Scenario Editor.